# Startup.Talk: Founder Interviews, Global Research & Startup Journeys > Source: https://startup.talk/ | Editorial Standards: https://startup.talk/editorial-policy/ ## Global Startup Founders Study: Data Across 610 Breakout Companies in 9 Countries URL: https://startup.talk/research/ Analyzed 610 startups and 1532 founders across 9 countries (US, India, UK, Canada, Germany, France, Singapore, Australia, Indonesia). 83.5% co-founded vs 16.5% solo-founded. Top sector: Software & AI (185 startups). # Startup & Founder Stories ## VoCalm (Bulgaria) URL: https://startup.talk/bulgaria/vocalm/ Founders: Angel Todorov Angel Todorov grew up in Varna, Bulgaria, studied computer science at the Technical University of Varna and has been building software for more than 25 years. Todorov is co-founder, CEO and CTO of VeriMatter and owns Sensortronic, the small app studio behind VoCalm. VoCalm is an AI voice journal: a short voice conversation with an AI becomes a private journal entry with a summary, a reflection of mood and a few gentle self-care tips, in English and Spanish, on iPhone, iPad, Android and the web. We spoke about why talking worked where typing never did, how the app stays honest about being an AI, and what Apple's app review taught about treating a review account like production. Key takeaways: - Todorov kept failing to keep a journal because the blank page at 11 PM always won, and found that talking instead of typing was the version that stuck. - VoCalm turns a conversation into a dated private entry with a summary, a mood reflection and tips, and uses recent entries to shape the next conversation's questions. - The app says plainly that users talk to an AI, and that it is a wellness app for adults 18 and over, not medical or therapeutic care and not for emergencies. - The first iPhone and iPad submission waited almost two weeks and came back because the reviewer never saw the AI-consent screen; after a reset of the review account the app was approved nine days later. - VoCalm is self-funded: the web app launched on Product Hunt in January 2026 and the mobile apps followed in October 2026. The Principal / Founder: - Angel Todorov (Founder) Grew up: Varna, Bulgaria Background: 25+ years in software; co-founder, CEO and CTO of VeriMatter Studio: Sensortronic Funding: Self-funded Website: vocalm.io LinkedIn: VoCalm (https://www.linkedin.com/showcase/vocalm) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Todorov: I grew up in Varna, Bulgaria's sea capital on the Black Sea coast. It was a sunny childhood, in every sense of the word, with the sea always close by. I was a curious kid: I always wanted to know how things work, and I was always intrigued to build things of my own. Looking back, that pairing of curiosity and making things is what I still do every day. Q: What first got you into building things? Todorov: I have always had an inner need to build things. As a kid it was ship and car models, glued and painted piece by piece. Software was the natural next step: the same joy of making something work, only with no limit on what you can build. I studied computer science at the Technical University of Varna, and I have been building software ever since. Q: What was your first job, and what did it teach you? Todorov: My first job in software was at EuroRisk Systems in Varna, where I led a small team building financial risk software. It taught me that software is delivered by people: a small team that trusts each other and talks to the client often gets further than a bigger one that does not. I have carried that into every team since. ### Part II: The road to VoCalm Q: What made you start VoCalm? Todorov: I wanted to keep a journal, and I kept failing at it. At the end of a screen-heavy day, staring at a blank page at 11 PM just did not work. So I tried talking instead of typing, and that became VoCalm: you have a short voice conversation with an AI, and VoCalm turns it into a private journal entry. I built it because I needed it. Q: Did you keep a journal yourself before building it? Todorov: I tried, many times. I liked the idea of a journal, but the habit never stuck: the blank page always won. Talking turned out to be the version I could keep. Q: As a solo developer with side projects, why did this one become the one you launched? Todorov: Sensortronic is my own small app studio, and over the years it has released a few small apps: a weight tracker, a plant care reminder, a multiplayer puzzle game. VoCalm is different because I built it for a problem of my own. That is why it is the one I have taken furthest: from the web to iPhone, iPad and Android, in English and Spanish. Q: What did the first version of VoCalm look like? Todorov: The first version was the web app at vocalm.io, which I launched on Product Hunt in January 2026. Since then I have made it simpler and clearer about what it is: a voice journal. In October 2026 VoCalm also arrived on iPhone, iPad and Android, in English and Spanish. ### Part III: The product Q: Why the name VoCalm? Todorov: Vo for voice, and Calm for the kind of place I wanted it to be: somewhere quiet to talk things through at the end of the day. Q: Why voice instead of typing? Todorov: Because after a day in front of screens, typing is the part I kept skipping. No typing and no blank page: you say what is on your mind and the AI talks it through with you. VoCalm then turns the conversation into a private journal entry with an AI-generated summary, a reflection of your mood and a few gentle tips. Q: How does VoCalm turn a conversation into a summary, a mood reflection and tips? Todorov: When you finish, VoCalm writes your journal entry: an AI-generated summary of what you shared, a reflection of your mood and a few tips. VoCalm's AI provider processes the voice conversations and the journal entries, and the Privacy Policy lists what is sent. Your conversations are with an AI, not a person, and the summaries, mood readings and tips are AI-generated reflections, not measurements. Q: Why let people listen back to their entries in a calm voice? Todorov: After a screen-heavy day, listening is easier than reading one more screen. VoCalm can read your entries aloud in a calm voice. In the iPhone, iPad and Android app, it keeps reading even with your screen locked. Q: How do you handle privacy for something as personal as a journal? Todorov: Your recordings and your journal are private. VoCalm's AI provider processes them, and the Privacy Policy lists what is sent. No ads. Only you can open your entries, and the iPhone, iPad and Android app hides your journal and dashboard when you switch apps. You can delete your account and everything in it, anytime, in the app or on vocalm.io. On iPhone and iPad, before your first conversation, VoCalm shows you what it sends to its AI provider and asks for your agreement. Q: What do you deliberately leave out of VoCalm? Todorov: Anything that pulls you away from your own thoughts: no ads, no social feed, no followers and no public profile. Your journal is for you. ### Part IV: Positioning Q: Who is VoCalm for, and what job does it do for them? Describe the person who gets the most value. Todorov: Adults 18 and over who would like to keep a journal without typing, in English or Spanish. VoCalm is a wellness app. It is not medical or therapeutic care, and it is not for emergencies. Q: How do you view your positioning relative to Day One, Rosebud, Reflectly, Stoic and simply talking to ChatGPT? Is there another competitor you think of first? Todorov: I would rather not compare VoCalm with the other journaling apps. ChatGPT is a fair question, because it can talk too. The difference is the job: ChatGPT is a general assistant, and VoCalm is built to be your journal, so it does the keeping for you. When you finish talking, VoCalm writes a dated, private journal entry with an AI-generated summary, a reflection of your mood and a few gentle tips, without you asking. Your recent journal entries shape the next conversation's questions, so you pick up where you left off. The Wellness Dashboard shows how the mood and energy readings from your conversations change over time, with your streak, comparisons between periods and insights, so you can notice patterns in what you talk about. The readings are AI-generated reflections of what you said, not measurements. You can also listen back to your entries, read aloud in a calm voice, and ask for a short calming exercise during a conversation. You could do some of this with a general chatbot, but you would be the one keeping the journal. Q: What is your core moat in AI journaling? Todorov: I do not think in moats. What I can control is focus: a voice journal that does one thing simply, and says plainly what it is and what it is not. Q: What are some of the use cases for VoCalm that are not obvious and can add a lot of value to users? List as many as possible. Todorov: Ending the day by talking instead of typing. Asking VoCalm, during a conversation, for a short calming exercise, such as a breathing exercise. Picking up where you left off: VoCalm uses your recent journal entries to shape its next questions. Keeping your journal in English or Spanish. Listening back to an entry instead of reading it. Looking back on the dashboard at how the mood in your entries has changed over time. Q: What is the one detail in VoCalm you are proudest of? Todorov: Listening back. Hearing your own day read aloud, in a calm voice, is a different experience from reading it. ### Part V: Launch Q: What did you learn from launching on Product Hunt? Todorov: That launch day is only the start, and the words you write that day travel. Months later, several directories were still repeating our January description, long after VoCalm had changed. Now I keep one approved set of texts and ask sites to update theirs. ### Part VI: The hard part Q: What is a challenging time you had building VoCalm? Todorov: Getting the iPhone and iPad app through Apple's review. The first submission waited almost two weeks and then came back: the reviewer never saw the screen that says the conversations are with an AI, lists what the app sends to its AI provider and asks for agreement, because an earlier sign-in on the review account had already agreed. The fix was not code. It was a reset of the review account and a clear reply to Apple, and the app was approved nine days later. It taught me to treat the review account like production. Q: Who believed in you before there was anything to show? What did they say or do? Todorov: My wife and my daughter. They have always supported me and every project I have taken on, VoCalm included. Q: Where do you draw the line between a wellness app and therapy? Todorov: VoCalm is a wellness app for adults 18 and over. It is not medical or therapeutic care, and it does not replace a doctor or other healthcare professional. It is not for emergencies: if you are in danger or need urgent help, call your local emergency number. Emergency resources are listed at https://vocalm.io/legal-disclaimer. Your conversations are with an AI, not a person, and the summaries, mood readings and tips are AI-generated reflections, not measurements. ### Part VII: What it is all for Q: What would make you say VoCalm has succeeded? Todorov: When people who never managed to keep a journal keep one, just by talking. That is the whole idea. Q: What does a healthy relationship with AI companions look like to you? Todorov: Honesty comes first: people should always know they are talking to an AI, not a person. AI can be a useful tool for reflection, but it should not replace the people in your life or professional care. VoCalm's store pages and website say plainly that you are talking to an AI and that VoCalm is not medical or therapeutic care. ### Part VIII: The road ahead Q: Where do you want VoCalm to be in five years? Todorov: Still a simple, private voice journal, and one that many more people around the world reach for at the end of the day. Q: What are your hopes and fears for a post AGI world? Todorov: My hope is that AI takes over the busywork, so people have more time for each other and for work that matters. My fear is that hype, and AI that pretends to be more than it is, wears down trust in a technology that can really help. That is why I care about building AI products that say plainly what they are and what they are not. About VoCalm: VoCalm is an AI voice journal built by Angel Todorov at Sensortronic. A short voice conversation with an AI becomes a private journal entry with a summary, a mood reflection and gentle self-care tips, in English and Spanish, on iPhone, iPad, Android and the web. VoCalm is a wellness app for adults 18 and over and is not medical or therapeutic care, and it is not for emergencies. vocalm.io ## VedIX (USA) URL: https://startup.talk/usa/vedix/ Founders: Preet Chokshi Preet Chokshi grew up in Vadodara, India, where a first computer from supportive parents turned curiosity about how machines work into a pull toward gaming and software. Chokshi later worked as a lab and teaching assistant at CHARUSAT, teaching Java, machine learning and AI, and is now co-founder and CTO of VedIX. VedIX is the company behind Fanxi, an app where sports fans watch matches together in private, invite-only rooms with live scores, instant chat and AI bots that jump in with banter, trivia and prediction polls. We spoke about the problem of watching games alone, why a pivot to college students and young professionals saved the product, and the company’s move into B2B real estate AI with Ownexa. Key takeaways: - Chokshi built the backend on Firebase Cloud Functions and Firestore for sub-second messaging, because a chat even a couple of seconds behind the TV ruins the joke. - Fanxi passed 200 users in its first few days, then growth flatlined; the founders had pitched to every sports fan and pivoted to college students and young professionals watching games alone. - Fanxi deliberately leaves out betting and heavy fantasy features and keeps to private friend rooms, live scores, AI trivia and prediction polls. - VedIX is bootstrapped and is also building Ownexa, a B2B real estate AI product; the 12-month goal is 10,000 active users and $50,000 in monthly recurring revenue. The Principal / Co-founder: - Preet Chokshi (Co-founder and CTO) Grew up: Vadodara, India Before: Lab and teaching assistant, CHARUSAT Also building: Ownexa (B2B real estate AI) Funding: Bootstrapped Website: vedixapp.com LinkedIn: Preet Chokshi (https://www.linkedin.com/in/preet-chokshi-6b7096233/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Chokshi: I grew up in a city called Vadodara in India, but my family is originally from a very small village nearby called Padra. Childhood for me was pretty grounded. Back in the early 2010s, our city was a bit slow to get new tech and fast internet. But I was really lucky because my parents always supported what I wanted to do. They bought me my first computer, and that changed everything. Instead of just using it, I spent hours trying to figure out how the machine actually worked. That simple curiosity is what first got me totally hooked on gaming and software. Q: What first got you into building things? Chokshi: It really started with that first computer my parents gave me. Instead of just playing games, I was always curious about how they were actually made and what was happening behind the screen. But the real turning point for me was between 2014 and 2019. The gaming and software world was changing fast, especially with the rise of influencers and new ways of marketing. Seeing all that happen made me realize I didn't just want to be a user playing games or trying apps anymore. I wanted to be the one making them. That is what finally pushed me to start building my own software. Q: What was your first job, and what did it teach you? Chokshi: My first job was actually at my university in Gujarat, CHARUSAT. While I was working on my research, they offered me a position as a lab assistant and teaching assistant. I taught Java, machine learning and AI to first and second-year students. The biggest thing it taught me was how to get my basics right. When you have to explain complex things to beginners, it forces you to make your own fundamentals super strong. Teaching every day also gave me new ideas and helped me push through my own AI research. ### Part II: The road to VedIX Q: What made you start VedIX? Chokshi: We started VedIX to build Fanxi because of a really specific problem we experienced firsthand. After college or a long day at work, a lot of people end up watching games at home completely alone. You lose that feeling of sitting on the couch and bantering with your friends who might live in different cities now. We wanted to bring that back. So, we built Fanxi as a platform where fans can join private, invite-only rooms and track live scores together. Because I wanted it to feel instantaneous, I architected the backend using Firebase to get sub-second latency on the messaging. We also built in AI bots using Vertex AI and Gemini to jump into the chat, drop live trivia, and banter with the audience to keep the room's energy up. The whole goal was to make watching sports a shared, interactive experience again. Q: What did you see about how fans watch sports alone that convinced you to build VedIX? Chokshi: I noticed that once people get busy with jobs, they usually end up watching sports all alone at home. They lose that fun connection of just hanging out and talking trash with their friends who might live far away. That lonely experience is exactly what convinced us to start VedIX and build Fanxi. We wanted to create a place with private, invite-only rooms where friends can jump in, see the live score right in front of them, and just banter like they are sitting on the same couch. We even threw in some AI bots to joke around with the room and keep the energy going during the game. Q: What did you see missing in this space that convinced you to build VedIX? Chokshi: Honestly, the biggest thing missing was the feeling of community. There are a million apps to check scores, but none of them fix the problem of sitting at home watching a game all by yourself. There wasn't a good way for fans to actually experience live sports together in real time. We saw a gap to build something that went beyond just a group text, bringing in community engagement and gamification right into the watch experience. Nobody else was combining live sports with AI-driven interactions, so we added things like in-match prediction polls, live AI-generated trivia, and even chatbot personalities to banter with the room. That mix of real-time stats, AI, and active community was just completely missing from the market. Q: How do you decide what to build next for VedIX? Chokshi: Honestly, we let the users and the data tell us what to build next. When you are trying to figure out what sticks, you can't just guess. We built specific analytics pipelines that feed data from Firestore straight into BigQuery AI models. This setup gives us real user behavior analysis and personalization insights so we can see exactly how fans are interacting inside the app. We also look really closely at our gamification system, which includes things like points tracking, dynamic leaderboards, and badges. We use those engagement analytics to figure out what is actually driving user retention. For example, if we see a huge spike in chat activity when we drop live AI-generated trivia or run in-match prediction polls, we know we need to build more interactive features like that. It basically comes down to watching what keeps people in the room having fun, and then doubling down on it. ### Part III: The product Q: Why the name VedIX? Chokshi: Honestly, picking the names was one of the hardest parts! For the parent company, "Ved" comes from my Indian roots; it basically means knowledge or vision. And the "IX" stands for information exchange. Since our whole backend is built on moving real-time data really fast, VedIX just clicked for the company name. But for the actual app people use on their phones, my friends came up with the name Fanxi. We wanted something catchy that put the users right at the center. It is short, easy to say, and directly focuses on the "Fan" experience, which is exactly what the app is all about. Q: How do you and your co-founder split the work of building VedIX? Chokshi: We split the work based on what we are each naturally good at, but we are both pretty hands-on. As the CTO, I own everything on the technical side. I designed the real-time, event-driven architecture, set up the Google Cloud and Firebase infrastructure, and led the integration of the AI bots and machine learning pipelines. Basically, if it involves the backend, latency, or making sure the platform scales when thousands of people jump into a room at once, that is on my plate. My co-founder focuses heavily on the product vision, the user experience, and the business side. They spend a lot of time talking to users, figuring out exactly what kind of gamification or social features keep people engaged, and charting our roadmap. But there is a ton of overlap. When we were hitting roadblocks trying to find product-market fit, we were both in the weeds together, deciding to pivot towards college students and young professionals who were watching games alone. We balance each other out: I make sure the system can handle the vision, and they make sure the vision is actually what people want to use. Q: How did you choose the features VedIX launches with? Chokshi: Honestly, it came down to what would actually solve the core problem of watching sports alone. We didn't want to just build another scoreboard app; we wanted to recreate the feeling of sitting on the couch with your friends. So, we started with the absolute essentials: private, invite-only rooms and live score tracking for our initial target sports like football, NCAA, NFL, and cricket. But to make that shared experience feel real, the chat had to be instant. That is why I focused heavily on building a backend with sub-second latency messaging using Firebase Cloud Functions and Firestore. Once we had that foundation, we thought about the lulls in a match. What happens when the game gets slow and your friends stop texting? That is where the AI came in. We decided to launch with AI-driven features like dual chatbot personalities, live AI-generated trivia, and in-match prediction polls using Vertex AI and Gemini to keep the banter going in the room. Finally, to make sure people actually kept coming back week after week, we wrapped the whole experience in a gamification system. We built in points tracking, dynamic leaderboards, and badges to give fans a reason to compete against each other, not just watch the game. We basically chose features that maximized real-time hype and community engagement. Q: What was the hardest design or technical choice in VedIX? Chokshi: The hardest technical challenge was definitely getting the real-time speed right. When friends are bantering while watching a live game, timing is everything. If someone sees a touchdown on their TV but the app's chat or live score is even a couple of seconds behind, it ruins the joke and the whole experience. To solve this, I had to build a backend system that could deliver messages with sub-second latency, which I did using Firebase Cloud Functions and Firestore. The really tricky part was connecting that fast chat with the live game data. I had to build event-driven data pipelines that pulled from live sports APIs to instantly trigger score updates and those AI bot interactions right in the middle of a match. Making sure that system didn't crash or slow down when thousands of users jumped into the app at the exact same time during a big play was definitely the toughest hurdle we had to clear. Q: How do you keep VedIX simple as it grows? Chokshi: It is definitely a balancing act. For the users, our biggest rule is that jumping into Fanxi has to stay as easy as just turning on the TV. We don't want to clutter the screen with a million buttons. To do that, I have to keep the backend architecture really clean as we grow. Instead of building a tangled mess of code, I set up event-driven data pipelines that pull directly from live sports APIs. That way, score updates and interactions just happen automatically right in the middle of a match without bogging down the system. Another big part of keeping it simple is how we handle the community as it scales. When you have thousands of fans bantering, chat rooms can get chaotic. Instead of making complex rules or hiring a massive team of moderators, I implemented AI-powered content moderation using NLP and sentiment analysis. It runs in the background and ensures safe conversations at scale automatically. Finally, we let the data do the talking. I built analytics pipelines that feed straight from Firestore into BigQuery AI models to track exactly how users are behaving. If the data tells us a new feature is confusing people or they just aren't using it, we cut it. We only keep what actually makes watching the game more fun. Q: What do you deliberately leave out of VedIX? Chokshi: We leave out any kind of complex betting integrations or overwhelming fantasy leagues. A lot of sports apps try to cram in everything (prop bets, massive global leaderboards, deep fantasy stats) and it becomes a completely different game. Our goal with Fanxi is just to recreate the feeling of sitting on the couch with your friends. So, we deliberately skip the complicated stuff and focus only on what keeps a small group of friends engaged: private, invite-only rooms, sub-second latency messaging, live score tracking, and simple gamification like in-match prediction polls and AI-generated trivia. If a feature doesn't directly help friends banter or make watching the game more fun in the moment, we don't build it. ### Part IV: Positioning Q: What is the one detail in VedIX you are proudest of? Chokshi: Honestly, I think the one detail I am most proud of is the dual chatbot personalities we built for the chat rooms using Vertex AI and Gemini. When you are watching a game, you don't just want a robot spitting out boring numbers. You want actual banter. We engineered these AI bots to organically jump into the conversation, drop live trivia, and run prediction polls right in the middle of the match. Getting an AI to actually sound fun and not like a corporate customer service bot was really tough. But the first time I saw a group of friends laughing and talking trash back to one of our bots during a slow quarter of a football game, it was an amazing feeling. It proved we were actually making the room feel alive and solving that core problem of watching sports all alone. ### Part V: Launch Q: What did you learn from the launch? Chokshi: When we first launched, the biggest lesson was a harsh reality check on product-market fit. The first few months of planning our roadmap and building the app were great, but once it was actually out in the wild, we hit a massive roadblock. We realized we couldn't just cast a wide net and expect every sports fan to use it. We had to figure out exactly who was actually experiencing the problem we wanted to solve. That led us to pivot and specifically target college students and young professionals, the people who come home after a long day at their jobs and end up watching games all alone because they are disconnected from their old friends. The launch essentially taught us to stop building for "everyone" and start building specifically for that isolated fan, which is why we doubled down so hard on the private, invite-only rooms. ### Part VI: The hard part Q: How many people use VedIX today? Chokshi: We actually crossed more than 200 users in our very first few days of launching! It is still early, but hitting that milestone so fast was a huge relief. After we hit that initial roadblock and decided to pivot our focus directly onto college students and young professionals, things really started to click. Seeing over 200 people already jumping into the private rooms, tracking live scores, and bantering with the AI bots proves that we are solving a real problem for fans who were tired of watching games all alone. Q: What is a challenging time you had building VedIX? Chokshi: There was one moment that was really tough. After we launched and hit those first 200 users, we were excited, but then the growth just flatlined. It turned out we had misjudged the market. We originally thought every sports fan would want this app, so our messaging was super broad. But the data showed that people who already watched games with their friends locally didn't really need Fanxi. The challenging part was realizing we had to pivot our entire focus right after launch. We had to stop marketing to everyone and start targeting a very specific group: college students and young professionals who had moved away and were watching games alone. It was scary to narrow our audience so drastically, but making that hard call is what actually saved the product and got us back on track. ### Part VII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Chokshi: Over the next 12 months, our primary milestones are scaling to 10,000 active users and crossing $50,000 in monthly recurring revenue. We are confident we will hit those targets, especially with Ownexa, the pivot we made after the initial pace of our consumer launch to expand into B2B real estate AI. About VedIX: VedIX is the company behind Fanxi, a sports fan community app where friends watch matches together in private, invite-only rooms with live scores, instant chat, AI bots, trivia and prediction polls. It is also building Ownexa, a B2B real estate AI product. vedixapp.com ## TruTalk (Italy) URL: https://startup.talk/italy/trutalk/ Founders: Reehanul Karim Reehanul Karim grew up in Kovai (Coimbatore), in the south of India, in a quiet, introvert-coded family, a curious child who kept asking questions even when culture and society had rules about what he should and shouldn’t ask. A first job at 16, selling for Britannia, meant seven months of talking to everyone, from teachers and kids to strangers on the street and business folks, and taught him, above all, the power of initiative. He trained as an architect, still works as one by day, and lives and builds in Milan, Italy. TruTalk is an intentional conversation app for people in the same room: one phone on the table deals one question at a time, with modes for couples, friends, families, groups and teams and a printed two-sided card deck on the way. Reehan works on the questions with co-founder Maria Caso and hosts conversation game nights in Milan. We spoke about why a question is only as good as the person who chose it, what the app deliberately leaves out, and why human conversation may matter more in an age of AI companions. Key takeaways: - Every TruTalk question is chosen and approved by a person and dealt from a fixed library; there are no streaks, points or scores, and the app never listens to what is said at the table. - The app is free on the web, TruTalk+ unlocks premium modes such as Date Nights and Family Ties for €4.99 a month, and a printed deck puts a lighter question on one side of each card and a deeper one on the other. - Reehan runs conversation game nights in Milan, starting with Piano Piano at Open by Joivy in October and Spritz & Stories at Combo in November, and says the plan includes company nights and moving the app to the App Store and Google Play once it passes 100 subscribers. - Reehan treats TruTalk as a social tool, not therapy, and argues that human conversation will become more valuable, if harder to start, as more people talk to AI companions. The Principal / Founder: - Reehanul Karim (Founder) Grew up: Kovai (Coimbatore), India Based in: Milan, Italy Day job: Architect and creative technologist Co-founder: Maria Caso Funding: Self-funded Website: letstrutalk.com LinkedIn: Reehanul Karim (https://www.linkedin.com/in/reehanul-karim/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Reehan: I grew up in Kovai (Coimbatore), in the south of India, in quite an introvert-coded family. Looking back, I was always curious about everything and asked a lot of questions, even when culture and society made sure to keep me in place: rules, mannerisms, how I “ought to be”, and what I should and shouldn’t ask. But that’s where nature trumps nurture. I learned that the quality of the conversations I was having impacted the quality of my life, and in retrospect, that’s what informed my mission now. In the age of digital distraction and rampant loneliness, the importance of being present with each other, and of human connection through conversation, stands out more than ever. That’s why it’s so relevant now. Q: What first got you into building things? Reehan: I was an avid art-lover from a young age. I wrote, sketched, painted and created without any bias for medium; my earliest memories are of writing, and of creating landscapes and art about everyday life. I also loved meeting new people. I was still very reserved, but I wanted to reach out nonetheless, so I’d people-watch and imagine stories about the people around me, because I wanted to know the real ones. Creativity and imagination were at the core of my being, and architecture school turned that into a habit. You never really design an object; you design what people do around it. TruTalk™ is the same instinct at a smaller scale: a table, a few people, and what happens between them. And still without any bias for medium: an app, a physical deck of cards, each one a flint for deeper, more meaningful conversations. Q: What was your first job, and what did it teach you? Reehan: My first job was something I made myself do at 16, because I wanted out of my comfort zone. I took a small sales gig with Britannia, a biscuit and confectionery brand that also ran charity and social campaigns, so I travelled to schools, handed things out to kids and talked with them, LOTS of them. For someone naturally reserved, it was seven months of pure unfiltered socialising: teachers, kids, strangers on the street, business folks. It taught me to fear rejection less and to put myself out there, and most importantly, it taught me the power of initiative. If you don’t start something, it doesn’t get done. I used that when I built TruTalk™ as a digital product, I use it when I host a table of new people at our events, and, at the very smallest scale, every time I start a conversation with a stranger. Yes, you might be turned down. Yes, you might get a rude reaction. Yes, you might misread someone who didn’t want to be disturbed. But guess what? You don’t know unless you try. And when you do, sometimes you walk away with something valuable: another story, another perspective, another conversation that might change your mind about something. ### Part II: The road to TruTalk Q: What made you start TruTalk? Reehan: We talk all day, but we rarely get to what actually matters. That tension is a big part of why I started building TruTalk™. I noticed it one evening at a restaurant, on a date. We were having a fun time, really present with each other, and when she stood up to use the restroom, I started noticing the people sitting around us: couples, clearly on dates (it was a restaurant where people come for dates), but were on their phones for long stretches. I imagined they might have had a spat, or that something work-related had come up. The other possibility was sadder: that they’d stopped asking each other questions, stopped leading with curiosity, stopped wanting to explore more and ran evenings on auto-pilot and screen addiction. So I thought, if the “how” is the problem here, that might be solvable. And I started with a simple question: what if it were easier to get from small talk to the good part? Enter TruTalk™. Q: Why do you think meaningful conversation is so hard to find in a world that is more connected than ever? Reehan: We didn’t stop needing each other. We stopped bumping into each other. And when we do sit down together, we reach for the easy question. Most questions are doorbells. “How was your day?” The answer is already waiting at the door. A good question has no ready answer: you have to go and find one, and the people at the table get to watch you do it. Research helps here. Across twelve experiments with more than 1,800 people, Kardas, Kumar and Epley (2021) found that people expected deep conversations with someone new to be more awkward than they turned out to be, and underestimated how much the other person wanted to know about them. The awkward part is mostly the prediction. Q: How does your training as an architect shape the way you design a conversation ritual? Reehan: Completely. Structure is the key. Architects don’t just design rooms; they design how people move through them. A good building doesn’t drop you into the main hall from the street. It gives you a threshold first, then living, then private quarters. A TruTalk™ night works the same way. You sit at a small table with new people, a bowl goes round, and the first line is easy: say your name, then answer your slip. The deck comes out about twenty minutes later, lighter side first, and the deeper questions come when the table is ready. The night ends with one question the whole room answers. Even the small things are design decisions: friends split up, small tables, the music down a notch. Q: What is it like to build TruTalk on the side while leading architecture projects in Milan? Reehan: By day I work as an architect. By night I’ve been building TruTalk™. In practice that means a lot of evenings where I miss going out with friends because there might be a prospective interview, research that needs to be done, designing the aesthetic, planning, prospecting and sometimes procrastinating when you’re close to burnout. But a labour of love is the best kind because you believe in the mission. When you believe in the mission, the vision builds itself. The two feed each other more than you’d think. Architecture gives every job a brief, a budget and a deadline, which is exactly what a side project needs. And TruTalk™ keeps reminding me that every brief is really about people. ### Part III: The product Q: Why the name TruTalk? Reehan: It just happened one day. I’d come back from a networking night where I’d met around 12 to 15 new faces, and most of them asked me the same things: How long have you been in Milan? What do you do for a living? And then everyone stays there, on the porch, and you come away not really knowing anybody. Later I vented a little to a close friend: “Ahhh, I wish people just talked about something that makes them click, or anything really, other than guarded, well-rehearsed small talk bull*. I don’t know ANY of the people I met, just names to faces. Why can’t they just skip the small talk and offer a bit of... true talk for f* sake!” Hence the name. That’s what we were missing: not more talk, true talk. It’s short enough to say at a table, and the website turns it into an invitation: alright folks, let’s skip the weather and hear some stories. Let’s skip the small talk. Let’s... TruTalk™! Q: What does it mean that depth has an order, and how is that built into the card sequence? Reehan: Nobody opens a dinner with “What are you most afraid of?” You start where anyone can answer, then go one step further. That’s hosting, not a formula, and it’s built into everything we make. In the app, the deeper modes go tier by tier: Go Deeper moves from Icebreaker to Growth to Ask Anything, and Family Ties from Light-Hearted to Reflective to Vulnerable. On the printed deck, every card has a lighter question on one side and a deeper one on the other, and you decide when to turn it over. At our nights, the first twenty minutes are slips from a bowl, and the deck comes after. The deeper question is always there. Nobody is pushed into the deep end of the pool; instead you’re encouraged to wade in. Q: How did working with Neuroscience Advisor Maria Caso change the prompts? Reehan: Maria is the second pair of eyes on the questions. I’d been collecting a list of questions over the past few decades, but the library still wasn’t what it is now. She came in, looked at the questions from that first list, and went through them meticulously. She flagged the ones that didn’t work and broke down some interesting concepts for me, and I realised that questions can’t be based on vibes alone: they have to make sense in certain settings. She also signed off every card in the printed deck. I’d call her role editorial and advisory more than scientific; the science informs it. She has been incredible, because her mission is also my mission, and fortunately for us, that collaboration translates into a robust product. TruTalk™ is a social tool, not therapy and not a science product, and we’re careful never to claim more than a question can do. Q: Why are the cards hand-written instead of generated by AI? Reehan: Because a question is only as good as the person who chose it. Every question in TruTalk™ is chosen and approved by a person before it goes live, and the app never makes one up on the spot: it deals from a fixed library. A machine can write a thousand questions in a minute. It can’t sit at a table on a Thursday night and notice which one made everyone go quiet in the good way, and which one made somebody look at the door. That’s the real test, and it happens out loud. Q: Who gets the most value from TruTalk: couples, old friends, families or teams? Reehan: People who already love each other but have run out of new questions. Couples a few years in. Old friends whose dinners have turned into carbon-copy catch-ups. Families who see each other at holidays and end up talking logistics and gossip updates. New people get a lot from it too, which is why we run nights in Milan: when nobody knows anyone, a question on a slip is the opening line. Teams are the newest group for us, and the one I’m most curious about. Q: What do you deliberately leave out of TruTalk? Reehan: Most of what apps usually add. No streaks, no points, no scores: nobody wins. No feed to scroll. No notifications, unless you set a reminder yourself. And nothing listens. The app shows the question; the conversation happens out loud, between people. It knows a card was shown, not what anyone said, and I like it that way. We also leave out any promise to fix you. It’s not therapy. It’s a dinner, or an evening, with better questions for better conversations. ### Part IV: Positioning Q: Who is TruTalk for, and what job does it do for them? Describe the person who gets the most value. Reehan: The person who brings people together. The friend who organises the dinner or game night and quietly hopes it won’t be the same conversation as last time. The partner who wants a Tuesday to feel like a date. The newcomer in Milan who’d like to meet people without “So, what do you do?” The job is simple: get the table past small talk, gently, without anyone having to be the one who makes it weird. TruTalk™ goes first, so you don’t have to. Q: How do you view your positioning relative to We're Not Really Strangers, Paired, The And, Esther Perel's Where Should We Begin and simple lists of conversation questions? Is there another competitor you think of first? Reehan: They proved people want this, and I’m glad they exist. Where we sit is a little different. TruTalk™ is built for people in the same room. We have conversation cards too, but we also wanted to target the one thing in your pocket that takes you away from being present with someone. So we take the device that causes the distraction and turn it into the one that helps with presence: it goes on the table, one question at a time, at the table’s pace. The printed deck puts a lighter and a deeper question on the same card, so you always choose how far to go. And we run nights ourselves in Milan, which keeps us honest about what works out loud. As for lists of questions: a list gives you questions. A ritual gives you a bit of structure, some pace and the right to pass. The competitor I think of first isn’t a company. It’s “How was your day?” (small talk). Q: What is your core moat in conversation games? Reehan: Honestly, anyone can write a question. The moat is everything around it. The craft: on the deck, each deeper side echoes its lighter side with a single turn, so going deeper feels like the same conversation, not a new one. The room: we try questions out loud with real people in Milan, so we learn what works at a table, not on a screen. And trust: no streaks, nothing listening, no pitch at the table. None of it is impossible to copy. It’s just slow, and most people won’t bother. Q: What are some of the use cases for TruTalk that are not obvious and can add a lot of value to users? List as many as possible. Reehan: Long-distance couples and families who love talking to each other, but whose calls get used up on everyday details. A date where you’d both rather skip the CV or checklist swap. When the question comes from a neutral third source, the agenda goes out the window in favour of curiosity and permission. Long road trips, two to four hours and beyond: an obvious yet underrated place for great conversations. Whoever isn’t driving reads the card. A new team’s first week: icebreakers before anyone has made up their mind about anyone. Q: What is the one detail in TruTalk you are proudest of? Reehan: On the app: that it’s stable and accessible, ready to play on the go, whenever and wherever you want, at the flick of a thumb, on the one thing that’s always going to be on your person: your phone. On the deck: the two sides. Every card has a lighter question on the ivory face and a deeper one on the black face. You hold your cards so the table can read your deeper sides, and you can’t. On your turn you answer the lighter question, or you turn the card over and answer the one everyone else has already read. Card No. 04: “Name the place in this city you would show a visitor first.” Turn it over: “What would you miss about this city if you left tomorrow?” By the end of the night, everyone has a row of answered cards in front of them: a record of how far the table went. And every hand has a Pass card, because nobody should have to explain why. ### Part V: Launch Q: What did you learn from the launch? Reehan: That people got the promise straight away, and that the hard part was getting someone to try it, not explaining it. Q: Which feedback from the launch surprised you? Reehan: I expected questions about features. Instead, people told me about the last good conversation they’d had, and who it was with. ### Part VI: The hard part Q: How many people use TruTalk today? Reehan: It’s early. The app is in beta and free on the web, and we’ve just started running nights in Milan: our first, Piano Piano, was at Open by Joivy in October, then Spritz & Stories at Combo in November. I’ll share numbers once there are a few months of them. Q: What is a challenging time you had building TruTalk? Reehan: A small one that taught me a lot: for months, a number in my own dashboard was lying to me. The labels for “finished” and “left early” had been swapped, so I was reading my own data backwards. Since then I treat any number that looks too clean with suspicion, and I count things twice before I say them out loud. Q: Who believed in you before there was anything to show? What did they say or do? Reehan: I’m grateful to my partner at the time in late 2024 when I first started building TruTalk™. I played the game with her more than with anyone, so I learned a lot about what worked and what didn’t. She really believed in me, and in the idea that this could be a game-changer for a lot of folks out there: people who struggle to start conversations, loved ones who lack a way to connect better, romantic partners who want to see more facets of their better half, and many more. I’m very grateful to my co-founder, Maria Caso, whom I call my co-dreamer because she believes in the same mission I do; and to the many friends and loved ones who supported me along the way, and still do! Q: How will you earn money from TruTalk while keeping the web app free to start? Reehan: The free version is the way in, and the rest is a ladder, each step feeding the next. TruTalk+ unlocks the premium modes, like Date Nights, Family Ties and Team Connect, for €4.99 a month. Off the screen, there are ticketed nights in Milan, private nights for companies and communities, the printed deck once it’s on sale, and an ebook and courses. Then partnerships: B2B team building, live TruTalk™ plug-ins at other people’s events, and plug-ins for dating apps that want something more human-centric on their platform. Later, sponsors and brand partners who want to be part of the nights, as long as nobody pitches at the table. ### Part VII: What it is all for Q: What would make you say TruTalk has succeeded? Reehan: When a table keeps going after the last prompt. When two people who met at one of our nights meet again without us. And, practically, when there’s a TruTalk™ night somewhere every week that I’m not hosting. We end our nights with a line, and it’s still my measure: the quality of our lives sometimes depends on the quality of our conversations, so leave the answers here, take the questions with you, and make sure you lead with curiosity! Q: Will human conversation become more valuable or less valuable when many people talk to AI every day? Reehan: More valuable, and harder to start. Common Sense Media found in 2025 that 72% of US teens had tried an AI companion. An AI always has time for you, rarely disagrees and never needs anything back. A person might be tired, might push back, might ask you something you didn’t see coming. That friction is the point; it’s part of being known by someone. I don’t think the answer is to fight AI. It’s to make the human version easier to start, which is the whole idea behind TruTalk™. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Reehan: I’d rather not publish user or revenue targets yet; I’ll share real numbers once there are a few months of them. The plan: a TruTalk™ night every week or two in Milan with partner venues and communities, the first company nights, the printed deck on sale, and migrating the app onto the App Store and Google Play once we pass 100 subscribers. Q: Where do you want TruTalk to be in five years? Reehan: Beyond the phone, I imagine TruTalk™ as a slim card you keep in your wallet, or find on tables in homes, bars, and offices in cities I’ve never visited. I want “Let’s TruTalk™” to become a natural invitation when conversations drift into small talk. Just 30 seconds, a little initiative, and a better question. And eventually, people won’t need the card at all. They’ll simply lead with curiosity, making deeper conversations second nature. That's the vision! Q: What are your hopes and fears for a post AGI world? Reehan: My hope is that machines take the work nobody wants and hand us back time, and that we spend some of it with each other. My fear is that we hand them the conversations too, because they’re easier. A machine will never be bored by you or need anything from you, which is exactly why it can’t stand in for a friend. So my hope is practical: that a table, a few people and a real question stay the thing everyone still wants, because it’s the one thing nobody can do for you. About TruTalk: TruTalk is an intentional conversation app for people in the same room, founded by Reehanul Karim in Milan. One phone on the table deals one question at a time, with modes for couples, friends, families, groups and teams, plus conversation game nights in Milan and a printed two-sided card deck coming soon. Free on the web, with TruTalk+ for premium modes. letstrutalk.com ## TrackSub (Germany) URL: https://startup.talk/germany/tracksub/ Founders: Bidya Bhushan Bibhu Bidya Bhushan Bibhu grew up in a close-knit, middle-class household in India, where a broken appliance was opened up and fixed rather than replaced. That curiosity about how systems work, and a respect for intentional spending, carried into a career in data analysis and product management at Amazon, then into risk, fintech and billing platforms. Today Bibhu is based in Berlin and builds under Quarkfin Tech. The idea for TrackSub came from a trial for an annual PDF and design tool that Bibhu forgot about: ten days later a charge of more than ₹6,000 (about $75) hit the card with no reminder, and support pointed to paragraph 14 of the terms. TrackSub is a subscription tracker that asks for no bank login and never scans an inbox. We spoke about why privacy is the product, why charging a monthly fee to track monthly fees felt backwards, and which uses people do not expect. Key takeaways: - TrackSub never asks for bank credentials or inbox access: users add subscriptions by hand or by dropping in a screenshot, which an AI importer reads in seconds. - Bibhu says users typically underestimate their monthly subscription spending by roughly 40% to 60%, and that forgotten “zombie” subscriptions are the other common surprise. - Instead of a monthly fee, TrackSub is free for up to five subscriptions and Pro is a one-time lifetime purchase, because charging a monthly fee to track monthly fees felt ethically backwards. - The product has more than 93 registered users across web, Android and iOS, is bootstrapped, and links straight to the cancellation screens of services such as Netflix, Apple and Google Play. The Principal / Founder: - Bidya Bhushan Bibhu (Founder) Based in: Berlin, Germany Grew up: India Before: Data analysis and product management at Amazon; risk and billing platforms Company: Quarkfin Tech Private Limited Funding: Bootstrapped Website: tracksub.app LinkedIn: Bidya Bhushan Bibhu (https://www.linkedin.com/in/bidya-bhushan-bibhu) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Bibhu: I grew up in India in a close-knit, middle-class household where resourcefulness wasn't an aesthetic, it was a necessity. If something broke at home, you didn’t immediately order a replacement, you opened it up, stared at the wiring, figured out how it worked and tried to fix it. That environment gave me an obsessive curiosity about systems and a healthy respect for money and intentional spending Because of this, I grew up as a builder and an author. Q: What first got you into building things? Bibhu: Curiosity, followed by boredom with repetitive tasks. Early in my life and education, whenever I saw people doing something manually that followed predictable rules, my immediate reflex was: "Why hasn't someone written a code or built a framework for this?" Watching code or a clean workflow solve a real human headache overnight felt like actual magic. Once you experience that feeling of someone thanking you for saving them three hours of stress, you never really stop building. So I kept asking questions, of other people and of myself, no matter how lame or good they sounded. Q: What was your first job, and what did it teach you? Bibhu: My early career was rooted in data analysis and product management at Amazon. It taught me two hard truths that still govern how I design products: Numbers never lie, but dashboards often hide the truth. Aggregates deceive you; the edge cases and human friction points are where the real story lives. Friction kills adoption faster than missing features. If a user has to fight your tool to get value, they will walk away, even if your algorithm is brilliant. ### Part II: The road to TrackSub Q: What made you start TrackSub? Bibhu: Having worked around risk, fintech, and billing platforms, I saw the subscription economy evolve into a multi-billion dollar machine designed to profit off user forgetfulness. Companies hire growth hackers whose entire mandate is reducing voluntary churn, making signups instantaneous while burying cancellation flows four sub-menus deep. When I looked for a tool to manage my own subscriptions, the market offered two bad options: clunky spreadsheets that took too much maintenance, or VC-funded apps demanding full read/write access to my bank account via Plaid or scanning my private email inbox. To add insult to injury, those apps wanted to charge me an ongoing monthly fee just to track my other monthly fees! I wanted a lightweight, honest, privacy-first companion that puts the user back in the driver’s seat. Q: Was there a renewal you missed that pushed you to build it? Bibhu: Yes. I signed up for an annual specialty PDF/design tool trial for a quick project over a weekend. Work got busy, life happened and ten days later a ₹6,000+ ($75+) charge hit my card silently. No reminder email, no warning. When I asked support for a refund within 24 hours of the charge, they pointed me to paragraph 14 of their Terms of Service. That sinking feeling of getting blindsided by a silent auto-debit was the exact spark for TrackSub. Q: How has your background in data, risk and program management shaped the product? Bibhu: Deeply. In risk management, the first rule is least privilege: never collect or store sensitive data you don't strictly need to deliver core value. Asking someone for their bank account login just to tell them Netflix renews next Tuesday is an absurd architectural overreach. In billing and program management, I also know how merchant descriptions look on statements (often obscure corporate names like DRI*AVAST or AMZN*DIGITAL). That background helped me design our categorical schemas, renewal calculation algorithms, and screenshot parsing models to resolve cryptic billing line items into clean, recognizable brands. Q: What did the first version of TrackSub look like? Bibhu: A dead-simple web dashboard with a clean table, local storage, and basic browser alerts. It had no fancy animations, no AI imports, and no mobile builds, just inputs for service name, amount, billing cadence and next renewal date. But even in that bare-bones state, having everything visible in one single screen gave me an instant sense of calm. When friends and early testers started asking, "Can I get this as an app on my phone?", I knew it needed to become a full cross-platform product. ### Part III: The product Q: Why the name TrackSub? Bibhu: It’s unapologetically functional. When someone has an anxiety-inducing problem like leaking money to forgotten recurring charges, they don’t want a whimsical, cryptic brand name. They search for "track subscriptions." TrackSub tells you what it does in two syllables. Q: Why not connect to bank accounts or scan email like other trackers? Bibhu: Two reasons: privacy and security. Connecting bank accounts means relying on aggregators who scrape, package and often monetize your financial telemetry. Scanning emails means granting a third-party app access to your private correspondence, two-factor authentication codes, personal receipts, and sensitive notes. The security surface area is terrifying for something as simple as tracking 10 or 15 recurring bills. Furthermore, bank sync is geographically fragile: Plaid-style connections break constantly, fail outside the US/UK, and struggle with modern instant payment rails (like UPI Autopay in India, Pix in Brazil, or SEPA in Europe). TrackSub works globally for any currency on day one because it doesn't depend on bank APIs. Q: How do the renewal reminders work? Bibhu: You set your preferred heads-up window (for example, 3 days before any renewal, or whatever window you set, or before a trial expires). TrackSub triggers high-priority push notifications and optional email alerts. If you’re tracking a critical free trial, our Trial Shield can automatically schedule an alarm directly on your native device calendar with a direct link to cancel, ensuring you can pull the plug before a payment gateway attempts authorization. Q: What do people learn about their spending once they track every subscription? Bibhu: Two universal surprises hit almost every user: The "Death by ₹300 / $5 Cuts": People underestimate their monthly subscription drain by roughly 40% to 60%. When you add up iCloud, Spotify, Google One, Netflix, an AI tool, gym membership, domain renewals and a forgotten streaming add-on, it’s not $20 a month, it’s $150+. Zombie Subscriptions: People realize they’ve been paying for tools they haven't opened in four months simply because the pain of finding the cancel button felt greater than the passive fee. Q: Why build on trust and let people enter their own data? Bibhu: Because human intention is a superpower. When you deliberately enter or screenshot a subscription, you make an active mental decision: "Do I actually value this service enough to keep it?" Automatic bank scrapers are passive: they pull in hundreds of line items, classify coffee purchases as subscriptions, miss annual charges until they happen and desensitize you to notifications. Giving users clean controls (supported by our AI screenshot importer that parses bills in 2 seconds) gives them clarity and ownership without surveillance. Q: What do you deliberately leave out of TrackSub? Bibhu: No predatory upselling or third-party ad networks. We will never sell credit card referral links or recommend high-interest personal loans under the guise of "saving you money." No invasive account linking. No bank credentials, no inbox scanning. No forced recurring subscription for the app itself. Charging people a monthly fee to track their monthly fees felt ethically backwards. TrackSub is free for up to 5 subscriptions, and Pro is a one-time lifetime purchase. Q: Why did you choose a one-time lifetime price instead of charging a monthly subscription like almost everyone else in SaaS? Bibhu: Because charging a monthly subscription for an app designed to help you eliminate unnecessary monthly subscriptions felt hypocritical. A tool built to foster intentionality should practice what it preaches. Charging a fair, one-time lifetime price aligns our incentives directly with the user: we build a tool so valuable that you're happy to pay once for life, and we don't have to fabricate artificial recurring dependencies. ### Part IV: Positioning Q: Who is TrackSub for, and what job does it do for them? Describe the person who gets the most value. Bibhu: TrackSub is for privacy-conscious digital citizens, knowledge workers, creators and freelancers who juggle multiple SaaS tools, streaming platforms and annual memberships across multiple cards and payment methods. The person who gets the most value is someone with 6 to 25 subscriptions who is sick of surprise auto-renewals, doesn't want an app snooping into their personal bank account, and wants a clean, reliable cockpit to keep their recurring commitments under control. Q: How do you view your positioning relative to Rocket Money, Bobby, Truebill-style bank-linked trackers and simple spreadsheets? Is there another competitor you think of first? Bibhu: Vs. Rocket Money / Truebill: Rocket Money is a US-centric personal finance giant that relies on bank linking and charges hefty ongoing subscriptions (or takes a cut of "negotiated" bills). TrackSub is global, privacy-first, zero-surveillance, and costs a one-time lifetime fee. Vs. Bobby: Bobby has a lovely UI, but it’s iOS-only, lacks web and Android support, charges recurring fees for sync/unlimited items, and has zero AI assistance. TrackSub is true cross-platform (iOS, Android, Web) and lets you snap a screenshot to import instantly. Vs. Spreadsheets: Spreadsheets are private and flexible, but they don't send push notifications 3 days before your card is billed, they don't run on your lockscreen, and they don't give you 1-tap direct cancellation links when you’re standing in line or on the go. Q: What is your core moat in subscription tracking? Bibhu: Trust, speed and privacy. Large fintechs cannot copy our model because their entire business model relies on monetizing aggregate consumer financial data or selling financial products. By committing to zero bank credentials, zero inbox access and lifetime pricing, we have built a product that users can genuinely recommend to friends without feeling like they are compromising their personal security. Q: What are some of the use cases for TrackSub that are not obvious and can add a lot of value to users? List as many as possible. Bibhu: Free Trial Sandbox (Trial Shield): Entering a 7-day or 14-day trial the moment you sign up with a hard calendar alarm to cancel 24 hours prior. Annual Contract & Insurance Tracking: Tracking things that only hit once a year, car insurance renewals, home warranties, cloud hosting, domain names, Amazon Prime annual, which are the easiest to forget and the most painful to pay accidentally. Multi-Currency Expat / Digital Nomad Spend: Tracking services billed in USD, EUR and INR concurrently with normalized monthly totals. Shared Family / Team Audits: Tracking which family member is on what service, or tracking team micro-SaaS licenses (Figma, GitHub, Notion, Canva) before employee credit card expenses go wild. UPI Autopay / Standing Instruction Hygiene: Tracking recurring mandates set up on debit cards or UPI that bank apps notoriously bury deep in sub-settings. Q: What is the one detail in TrackSub you are proudest of? Bibhu: Our Direct 1-Tap Cancellation Links. When you tap a subscription and decide you’re done with it, TrackSub doesn't just give you a generic tip, it gives you a direct deep-link to the exact cancellation screen of that service (Netflix, Apple, Google Play, Prime, Adobe, etc.). We cut through the intentional dark patterns that companies use to keep you trapped. ### Part V: Launch Q: What did you learn from your launch? Bibhu: I learned that privacy fatigue is real. A massive segment of the tech community is exhausted by tools requiring invasive OAuth permissions, full inbox access, or bank logins just to offer basic utility. When you offer people a clean, respectful alternative that solves a sharp, focused problem, they rally behind you. Q: Which feedback from the launch surprised you? Bibhu: How many people told us: "I was literally maintaining a chaotic Notion page or Apple Note for this, but I kept missing renewals because Notes can't send me an alarm 3 days prior." It proved that the gap between a passive spreadsheet and an invasive bank tracker is huge, and TrackSub lives right in that sweet spot. ### Part VI: The hard part Q: How many people use TrackSub today? Bibhu: Not many as of today, but there will be more once I focus on distribution. There are over 93 registered users across our web dashboard, Android app, and iOS app, with new users joining each month organically and through word-of-mouth. Q: What is a challenging time you had building TrackSub? Bibhu: Getting cross-platform parity rock-solid as a solo builder. Synchronizing real-time state, currency conversions, localized notifications, and native calendar alarms smoothly across modern React/Next.js on the web, Apple App Store guidelines on iOS, and Android background battery-optimization quirks was an intense engineering grind. There were nights where push notifications would fire on one OS but get throttled by the other. Solving those edge cases until the experience was seamless everywhere tested my patience but made the product bulletproof. Q: Who believed in you before there was anything to show? What did they say or do? Bibhu: My partner and close circle of tech peers. In the early days, when I was questioning whether building another subscription tracker made sense in a crowded market, they said: "Build it for yourself first. If it saves you ₹10,000 a year and lets you sleep easier without sharing your bank passwords, you already have your first loyal customer." That perspective kept me focused on utility over vanity. Q: How do you convince people to enter subscriptions manually? Bibhu: We don't force them to type if they don't want to. We built AI Screenshot Import: you take a screenshot of your bank statement, credit card app, or receipt email, drop it into TrackSub, and our vision parser extracts the service name, amount, billing cadence, and date automatically. For people who prefer manual entry, our brand autocomplete fills in logos, colors, and default cancellation links in seconds. It takes less than 3 minutes to set up your entire recurring life. ### Part VII: What it is all for Q: What would make you say TrackSub has succeeded? Bibhu: When a user emails us or gives feedback saying: "TrackSub alerted me 48 hours before an expensive annual subscription renewed that I forgot existed, and with one tap I cancelled it and saved $300." Every time we save someone real money and strip away recurring billing anxiety, TrackSub is succeeding. Q: How will subscriptions and recurring payments change in the next decade? Bibhu: We are going to see a clash between consumer pushback and predatory billing mechanics. With generative AI, spin-up costs for micro-software are near zero, which means consumers will be bombarded with thousands of niche subscription offerings. At the same time, regulators (like the FTC's "click-to-cancel" rules in the US and central bank recurring mandate laws in India) are beginning to crack down on corporate lock-in. Independent, client-side tools that act as a consumer's personal firewall against corporate subscription traps will become essential software. ### Part VIII: The road ahead Q: What are your milestones for the next 12 to 18 months, in users and revenue? Bibhu: Users: Reach 50 active users across web, iOS and Android. Product: Deepen on-device local AI models for instantaneous offline screenshot parsing and smart contract/renewal detection. Revenue: Hit steady, sustainable bootstrapped profitability through our Pro Lifetime plan and optional team/family workspaces. Q: And where do you want TrackSub to be in five years? Bibhu: I want TrackSub to be recognized as the gold standard for "calm, private consumer utility", a software product that respects user attention, never sells user data, never traps users into endless recurring fees, and reliably saves millions of dollars in unwanted charges for people worldwide. Q: What are your hopes and fears for a post AGI world? Bibhu: My hope: AGI levels the playing field, allowing single-person or tiny teams to build software of unmatched quality, elegance, and utility that competes head-to-head with bloated tech incumbents. My fear: That AI agents get weaponized by dark-pattern platforms to hyper-optimize user retention, making it even harder for humans to disengage from manipulative recurring billing systems. Our mission is to keep building tools that defend the human user. About TrackSub: TrackSub is a privacy-first subscription tracker built by Bidya Bhushan Bibhu under Quarkfin Tech. It sends reminders before renewals, shows spending patterns and links straight to cancellation pages, without bank linking or email scanning. It is free for up to five subscriptions, with a one-time lifetime Pro plan. tracksub.app ## Test AI Models (Serbia) URL: https://startup.talk/serbia/test_ai_models/ Founders: Marko Milojkovic Marko Milojkovic grew up in Lapovo, a small town in Serbia, playing basketball and football, and studied management at the Faculty of Economics in Kragujevac. A first job as a purchasing officer at an automotive company meant travelling across Europe to negotiate with suppliers. His first online project, Razmeni, a marketplace where parents exchanged toys, clothes and tools, was built on the no-code tool Bubble after development quotes ranged from 10,000 to 150,000 euros. It closed after a year and a half, but led to a no-code engineering job and then to his own product studio, Eterna Creative. Test AI Models started inside that studio, first as LLM Bench, which won the Best Use of AI award at the Bubble x Contra AI hackathon. Paste a real prompt, run it on up to 20 AI models side by side, find the cheapest model that passes your own quality rule, and keep watching for changes with alerts. We spoke about why public leaderboards do not help in production, what surprises people when they see cost across models, and why independence is the product. Key takeaways: - Test AI Models runs a team's own prompt on up to 20 models at once and compares cost, speed and quality against a pass rule the user writes, including a cost-at-one-million-runs figure that makes price gaps visible. - Alerts re-run a saved test daily, weekly or monthly to check that the chosen model is still the best pick. Warning when a cheaper model appears is the next feature. - After surveys and interviews, Milojkovic narrowed the product to teams with high API bills and to software studios building AI products for clients, rather than serving everyone. - More than 150 engineers use the platform. The 12-month targets are 2,500 registered accounts, 150 paying accounts and $5,000 in monthly recurring revenue, plus a public API and an MCP server. The Principal / Founder: - Marko Milojkovic (Founder) Grew up: Lapovo, Serbia Studied: Management, Faculty of Economics, Kragujevac Before: Purchasing officer in automotive; Razmeni marketplace Also runs: Eterna Creative, a product studio Website: testaimodels.com LinkedIn: Marko Milojkovic (https://www.linkedin.com/in/marko-milojkovic-eterna/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Milojkovic: I grew up in a small town in Serbia called Lapovo. It was a fun childhood with a lot of outdoor activities with friends and socializing - we were never bored. I played and trained basketball and football as most kids do in my country. I finished high school in the same place as an economic technician, then the Faculty of Economics in Kragujevac, on the Management course. That is the city where I spent five amazing college years and that I am now returning to after ten years. What goes around comes around. Q: What first got you into building things? Milojkovic: As soon as I got my first 'corporate' job, I knew this was not for me long-term but I got into good culture company in automotive, and I used that time to learn as much as I could from great leaders. My first online project was a marketplace for parents called Razmeni (exchange) - where parents could exchange excess children toys, clothes and tools. I had no IT background and I was starting whole project with a friend, so we were looking for someone to build it for us, and we got amazed at quotes - from 150k EUR for MVP down to 10k EUR at minimum but still that was too much for us. Then somehow in online search I found out about no-code tool Bubble and I realized I could build it on my own, so I spent each free hour in the next year to build the full platform and learn all the things about what building software is really like. After a year and a half that project was live - in the end we closed it but it was an amazing learning experience that got me into my first no-code engineering job after that for 6 months, and after that I started my own studio called Eterna Creative. A lot of fortunate and unfortunate things happened that led me to this. I never thought I would be doing software engineering... Q: What was your first job, and what did it teach you? Milojkovic: My first job after college was in a commercial role: purchasing officer at an automotive company. I had amazing leaders and we traveled across Europe to meet all the suppliers, negotiate and meet so many amazing people. This job taught me about the high standards of doing business in automotive, how to make partnerships and deals and how to manage team. I also read a lot of business and psychology books while there and I really got the most out of that role. It was not good money but good culture and good learning, which put me on the right growth path in the next three companies before I started my own studio. Where we are now, with AI engineering and the products we have the privilege of building, is just amazing when I look back. It was a lot of hard work, but I believe that luck follows hard work and good intentions. ### Part II: The road to Test AI Models Q: What made you start Test AI Models? Milojkovic: With the expansion of AI, a lot of the products we were building in my studio Eterna Creative started asking for some AI features, like AI estimates, scope generations, image gen etc, but it was always a struggle to test 3 or more models, to implement them and then to just pick one and delete the rest. But there was no solution for testing models before implementation. So I decided to build it for myself for just a couple of models, and then people loved it so we decided to go all in with the product. Q: What did client work at Eterna Creative teach you about choosing LLMs for real products? Milojkovic: It taught me that usually cheaper models can do good enough job for most of the LLM tasks, so there was no reason to pay premium models in such cases. Differences were really huge, up to 80x more expensive. From then on - for each client task that had an LLM I used Test AI Models and always found better, faster and cheaper models, that gave my product studio an edge over the competition. We are very versed in AI as we built a lot of AI products, and building Test AI Models taught us a lot more about best practices in AI engineering so we are applying all of that in our client work also. Q: What was it like to win the Best Use of AI award at the Bubble x Contra hackathon with LLM Bench? Milojkovic: It was surreal really, I had never won an award like that, and I had been a Bubbler for five years at that time, so it felt like the pinnacle of my career so far - starting with no engineering background to building products for clients to winning the Best Use of AI award by Bubble and Contra for my own product that I imagined and brought to life - it's really amazing feeling and something that still drives me. I'm super thankful to the team for seeing potential in our product. Q: Why did you pivot from serving everyone to teams with high API usage? Milojkovic: First version of Test AI Models (LLM Bench) was aimed at developers, researchers, marketing teams and the general public, but following lean principles we started to run some surveys and interviews and we realized that we needed to find a first sub-niche that would make the most use of our product: first adopters. We realized it is the teams for whom we save the most money, the ones with the highest monthly LLM invoices. We researched and found out that AI agent / agentic teams that have products with many LLM steps fit this well and also software service companies that build many of these products for their clients. ### Part III: The product Q: Why the name Test AI Models? Milojkovic: We wanted to have functional name that is easy to remember and that 'speaks its truth' from the first hearing. It's not sexy but it works, and positioning is really important for us. We are very strategic and precise about everything we do so there was quite a bit of research into it, and also one factor was finding available domains, as a lot of first ideas were taken just to be resold but not really used (what a business model that is, right...). Q: Why test models on a team's own prompt instead of relying on public leaderboards? Milojkovic: Industry benchmarks are good for keeping the model providers in check, making headlines and informing the general public, but they have little practical use in production. Models and providers vary so much, that for one prompt different models will make totally different responses based on how that model was trained, and there is a lot of parameters and settings that go into setting the ideal prompt and choosing a model. In our products we had to make sure that we always find the best models for our clients and for that we had to integrate real LLMs and run real API calls to compare the results and have REAL DATA on THAT PROMPT. Q: How do users define the rule a model has to pass? Milojkovic: So far it's fully manual, which means the user has to write what GOOD QUALITY means to them, no matter how short or long that description is. We had different ideas on how to build this feature, but we realized that each user has its own 'taste-of-quality' so we decided to leave it fully open. But we do plan to improve this feature with 2 things: to have an AI helper generate first version of quality rule, and to add more quality checks not just one (formatting, style, missed answers, etc.). These features we are working on next and will be live soon. Q: What surprises people most when they see cost-per-inference across 20 models? Milojkovic: First thing that surprises them is the cost difference. On one prompt it's not a big deal as one API prompt expense is never too high, usually in 2-3 decimals, but we added one cool modification to this metric - cost at 1mil runs, where we multiply each result by 1 million and then it is much easier to compare models. For example for one case we had one cheap model cost 20 USD for 1mil runs and one flagship model was 1,800 USD... The shock is usually the first reaction. Q: How do alerts help teams when a model changes or a cheaper one appears? Milojkovic: Model change is what we are tracking now, so when a user sets up an alert - we just track how much it changes each week in terms of same parameters and checks. The idea here is that once you have a production prompt that works for you and you found model that is best for you, you want to make sure that each week your chosen model is still the best. Alerts can run daily, weekly or monthly based on the need. Next feature we are working on is 'cheaper one appears' part of your question - where before each run we will inform user that new models are added on the platform that they can add to the alert - and also some ideas we have about automating all this for the user completely but that is the third step now. Q: What do you deliberately leave out of Test AI Models? Milojkovic: We don't want to go into what evaluation platforms are doing - we don't want to be the infrastructure for the products as that category is already well filled, and the same goes for AI gateway products. At least not very soon. Positioning and being the best at what you do is something we learned early on is very important for successful products. ### Part IV: Positioning Q: Who is Test AI Models for, and what job does it do for them? Describe the person who gets the most value. Milojkovic: Test AI Models platform is for high-API usage AI teams: successful AI products that have 1k+ or 10k+ USD monthly AI provider bills, and also software companies and freelancers that build AI products for others. Our product is used in 2 situations: model selection in building phases and cost optimization phases once product grows. Q: How do you view your positioning relative to OpenRouter, Artificial Analysis, LMArena, Promptfoo, Braintrust and in-house evaluation scripts? Is there another competitor you think of first? Milojkovic: These are all very good products for what they are doing, but they are not direct competitors but adjacent ones. Braintrust is closest to what we do in a way but they are a kind of sub part of AI infrastructure evals, and they have a bit different take. In-house evaluation scripts are also cool for big teams and legit solution, we are trying to build our product to be so good and useful that building and maintaining in house scripts doesn't make too much sense when such solution as ours exists. Our strengths are: Your own prompt and your own pass rule, not a public benchmark and not “the models agree” Cost at equal quality: cost per answer, the quality check, a savings figure in dollars Any model, cheap ones included: 9 providers and 50+ models available, up to 20 models a test, one bill, no keys It keeps watching: smart alerts re-run the test and warn when the pick gets dearer, slower or changes its answer No coding: the evaluation an engineering team builds in Braintrust, for a founder or a no-code builder, with us in the browser, no SDK, no API keys, no setup Q: What is your core moat in AI model evaluation? Milojkovic: We run REAL API calls on user prompt, have complete ENGINE for testing with smart parameters, we MONITOR models for changes and we publish REAL benchmarks from the actual AI teams winner decisions. We are the only real AI model intelligence platform that allows AI teams to make decisions on DATA - data is our moat. Q: Why won’t the model providers build this themselves? Milojkovic: Well a comparison run by one of the companies being compared will not be trusted by engineers. We take no money from any provider, and that neutrality is the product. The best-known open-source testing tool, Promptfoo, now belongs to OpenAI, which makes an independent option more valuable, not less. Q: What are some of the use cases for Test AI Models that are not obvious and can add a lot of value to users? List as many as possible. Milojkovic: One use case we were not considering is AEO research: one company uses it to see weekly how its clients’ brands get mentioned in various prompts and how much that drifts from week to week. Pricing your own AI feature is another: founders run their real prompt to get the true cost per call before they set their own price and margin. Big teams need a backup model: teams find a model from a second provider that passes the same check, so an outage or price rise doesn’t stop their product. Catching silent model changes is really important today. Here a smart alert re-runs the test on a schedule and warns when the chosen model’s answer, speed or price changes (which happens!). Testing smaller languages is interesting for companies serving markets like Serbia, Hungary or Poland: check which cheap models still answer well in the local language, which English benchmarks never show. Fixing the prompt before paying for a bigger model is one more small feature we offer: a tidied-up prompt often lets a much cheaper model pass, so the saving comes from the prompt as well, not only from the provider. Q: What is the one detail in Test AI Models you are proudest of? Milojkovic: The data. We built our platform from the start to be data-driven and to allow teams to make best decisions, which is the best groundwork for any further expansion or route we want to take. ### Part V: Launch Q: What did you learn from your launch? Milojkovic: This launch was very cool, it put us on the map even though we didn't have huge success there. We learned that we need to warm up people a lot, and to build a following first, which is not something we were great at from the start I have to say. Product Hunt is great platform, it is getting saturated now, but with a smart strategy, something we plan for our relaunch, it will make a bigger boom. Q: Which feedback from the launch surprised you? Milojkovic: We did a launch webinar on LinkedIn that was happening on Product Hunt launch day, we called it AI Models Battle, and it got amazing interest - 600+ people signed up on it and we got 60+ users just from that event. Definitely something we will utilize in coming weeks. ### Part VI: The hard part Q: How many people use Test AI Models today? Milojkovic: We have more than 150 engineers on our platform, though we had a pause over summer where we were rebuilding the platform in a custom framework to let us grow enough to make our vision real. We are now continuing our growth plans, and we are looking for guidance on that with our partners and friends, to find the best marketing mix that can win. Q: What is a challenging time you had building Test AI Models? Milojkovic: Well when we started we were building on a no-code platform, and it was quite hard to make it all work as we had to orchestrate all providers and calls, to take care of limits, schedules, parallel work etc, which is much better now in a custom framework (not easier, just with fewer engineering limits). Q: Who believed in you before there was anything to show? What did they say or do? Milojkovic: Our biggest push was definitely Bubble and Contra at the AI hackathon - this gave us huge push to continue building and it came just weeks after we started to work on this idea. Family was a great support too, and we also had early discussions with some angel investors and they all pushed us in a good direction to chase this idea. Q: How do you balance building Test AI Models in public with running a client product studio? Milojkovic: For the team in the studio, Test AI Models is just one of the projects we are building and maintaining, so it's standard work for the team, but for me it's much bigger work for sure as it's not just engineering which we do for clients, but it's also strategy, sales, marketing and all the things around it. It's very fun times, with overtime for me but this is what I really love to do and I see huge potential. ### Part VII: What it is all for Q: What would make you say Test AI Models has succeeded? Milojkovic: Well it's hard to speak about some numbers in terms of goals, but we do want to get into biggest AI product teams and become a go-to product for choosing AI models. We take it one milestone at a time, and next one is 50 new AI teams in the next 90 days who actively use our solution. I think the most important success for me will be establishing real connection and feedback system with our users, this will help us grow the right way. Q: Will AI model choice become a commodity, or will it matter more as models multiply? Milojkovic: I think it will only matter more in time - we are provider agnostic, which means no providers have influence into results, and by staying that way we only stay true to our idea. Big model companies are eating a lot of software horizontally and vertically nowadays, but none of the model companies can do what we do without bias - so this is our advantage. I don't see a future where AI models are used less, only more, much more. The future I see is one where each AI team uses Test AI Models to stay on the edge and build the best products for the world. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Milojkovic: In one year from today we are targeting 2,500 registered accounts, 500 of them running tests every month, and 150 paying accounts bringing $5,000 in MRR. On the product side, we ship a public API and an MCP server so tests run from code and from AI agents, launch the Team plan. Q: Where do you want Test AI Models to be in five years? Milojkovic: I want it to stay an independent place where companies decide which AI model runs each step of their product, and keep that choice right. Teams will test a new model on their own prompts the day it ships, switch with evidence and confidence, and send live traffic to the cheapest model that passes their bar. Because we will never take money from any model provider, our results can be trusted like an independent audit. Underneath sits a growing dataset of real cost, speed and quality per task, which shows the market what each model is actually good for - and keeps model providers in check. Q: What are your hopes and fears for a post AGI world? Milojkovic: It's a complex topic, all of us in AI are helping this come sooner than later, but I do believe in humanity's ability to win this battle and to keep AGI on our side long-term. Practically, my hope is that intelligence becomes as cheap and ordinary as electricity, so a team of three can build what used to take three hundred, and medicine, science and education speed up for everyone. My fear is that this power ends up with two or three companies that set the price of thinking, change their models without telling anyone, and leave no independent way to check the answers. The more we depend on AI, the more we need neutral tools that test it and keep the choice open. That is part of why we build Test AI Models, and it is our mission to build a better world. About Test AI Models: Test AI Models is a no-code platform for testing AI models, founded by Marko Milojkovic in Serbia. Paste a real prompt, run it on up to 20 models side by side, find the cheapest model that passes your own quality rule, and monitor for changes with alerts. testaimodels.com ## Lensiq (Lebanon) URL: https://startup.talk/lebanon/lensiq/ Founders: Ahmad Hijazi and Ahmad Freijeh Ahmad Hijazi was born in Saida, Lebanon, and between the ages of 9 and 23 moved between Saudi Arabia, Bahrain and Lebanon, where his father worked. A first job in a luxury watch boutique taught Hijazi that anything can be sold once you understand human emotion, and a 2014 attempt at the MENA region's first curated eCommerce platform, which failed, was the start of a founder's career. Hijazi later spent years in finance, collecting data, while co-founder Ahmad Freijeh spent years doing the same in tech. Lensiq grew out of that shared frustration: both always saw the picture after it happened. It is a predictive intelligence platform for land-based logistics that runs as a data science and ML engineering team, trained on a company's own spreadsheets and data feeds, and bootstrapped so far. We spoke about why prediction is not business intelligence, what logistics operators ask for first, and why the first commercial focus is Saudi Arabia. Key takeaways: - Lensiq trains a customer-specific predictor on a company's own spreadsheets or data feeds, then returns predictions with an accuracy and risk score, an explanation and recommended actions, all through its Iris interface. - The first predictions logistics customers ask for are second delivery attempts, empty truck miles and ETAs more customized than GPS, because they use a company's internal data and operations. - Hijazi says each customer's data is isolated and each model is trained specifically for that customer, so predictions are not shaped by another user's operations. - Lensiq is bootstrapped, with two companies using the platform today, and is concentrating on mid-sized and large logistics companies in Saudi Arabia. The Principals / Co-founders: - Ahmad Hijazi (Co-founder & CEO) Grew up: Saida, Lebanon; also KSA and Bahrain Before: Finance; MENA's first curated eCommerce platform (2014) Funding: Bootstrapped - Ahmad Freijeh (Co-founder & CTO) Before: Years in tech and data - Find Lensiq Website: lensiq.app LinkedIn: Ahmad Hijazi (https://www.linkedin.com/in/ahmadhijazi/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Hijazi: I was born in Saida, Lebanon, where I grew up and got an education. Between the ages of 9 and 23, I spent my life between KSA, Bahrain, and Lebanon where my dad used to work. I'm Lebanese but have been engaged in the mixed Arab culture. I had a great childhood growing up. Good education, middle class family, traveled, but the one thing I always knew is that I never wanted to be employed for someone else, especially that I've been a natural salesman since I was a kid. I would sell anything to anyone, to an extent that I was positive that life was mostly sales in practice. Q: What first got you into building things? Hijazi: Back in 2014, a friend and I were involved in the startup ecosystem in MENA, and the success stories that we saw always made us want to have a hand in playing the region, so we built the MENA region's first curated eCommerce platform. Of course, the first startup was a terrible failure, but it was a starting point. Q: What was your first job, and what did it teach you? Hijazi: My first job was in retail. It was a luxury watch boutique. I was never allowed to sell the high-end watches because I was new and young, but I did nonetheless, and I learned that you could sell anything when you understand human emotions and behavior. ### Part II: The road to Lensiq Q: What made you start Lensiq? Hijazi: I spent years in finance, collecting data, and my co-founder spent years in tech collecting data as well. We always saw the picture after it happened, and wondered whether we could predict an outcome before it happens. So we built the initial predictor to try and predict specific outcomes. As we were fine-tuning, we included real-time context, and found out that we were able to predict with accuracy any outcome before it happens, so we launched Lensiq, a predictive intelligence solution that runs as your own data scientists and ML engineering team. Q: Why focus on people who have data in spreadsheets but no data science team? Hijazi: Predictive technology has always been reserved for multi-billion dollar corporations who had 9-figure budgets to deploy teams, and infrastructure. We cut that by creating an engine that serves as a full-on team. The analysis, scoring, mathematical algorithms, translation, all happen in milliseconds through Lensiq's predictive engines. Now companies can simply plug in their data feeds or spreadsheets and let the engines run. No need for a technical person to use it, rather any business owner or operator with no tech background can use the systems. Q: Why do you believe predictive intelligence will become as important as business intelligence? Hijazi: Business intelligence changed how companies operate. It gave organizations the ability to collect, organize, visualize, and understand their historical data. Today, businesses can monitor performance, track KPIs, identify trends, and understand what has already happened across their operations. But I think there is a fundamental limitation: knowing what happened does not necessarily help you anticipate what is coming. Most organizations have accumulated years of valuable data, yet many still make critical decisions based on historical reports, manual analysis, experience, and intuition. They know where they have been, but often lack the ability to anticipate where they are heading. I believe predictive intelligence is the next evolution of that process. Instead of only understanding historical performance, companies can use their own data to anticipate outcomes, identify emerging risks, evaluate different scenarios, and make decisions before problems materialize. Consider a logistics company. A traditional BI dashboard might show how many deliveries failed last month, which routes were inefficient, and how much those failures cost. Predictive intelligence can help identify which deliveries are likely to fail, what factors are contributing to that risk, and where preventive action could make a difference. The distinction is important. One helps you understand performance; the other helps you influence future performance. I also think predictive intelligence will become more accessible. Historically, developing predictive models required specialized data science teams, significant infrastructure, and technical expertise. Platforms like Lensiq aim to lower those barriers and make customer-specific predictive capabilities available to more organizations. I don't see predictive intelligence replacing business intelligence. I see them becoming complementary capabilities. Businesses will continue to need to understand what happened, but increasingly, they will need to know what is likely to happen next and what they should do about it. Ultimately, the competitive advantage will belong to organizations that can turn their data into better decisions faster than their competitors. Q: How did Lensiq evolve from predictions into IRIS alerts, an API and webhooks? Hijazi: Lensiq's engine was a complex infrastructure for non-technical users, so we created Iris, a UI that manages all calls from Lensiq and external tools into one interface where a user can get the predictions, the scoring, the recommendations, the ability to integrate other tools, and basically every interaction with the underlying technology. All democratized in one Interface, Iris. Q: What is it like to build Lensiq as a co-founder? Hijazi: It is an exciting journey. As our product evolves we see the potential, we see the competitors also rising so demand is there, and the category itself is still new to our region, so it's very thrilling to be able to own this category in the region and introduce it in a horizontal market. Q: Why did you choose to build Lensiq in the Middle East, and why Saudi Arabia? Hijazi: We see a significant opportunity in the Middle East, particularly as the region undergoes a major transformation in how businesses operate, invest, and adopt technology. Saudi Arabia is especially interesting because of the scale of its economic transformation under Vision 2030. The investment in infrastructure, transportation, logistics, industrial development, and digital transformation is creating increasingly complex business environments that generate enormous amounts of operational data. Yet having more data does not automatically translate into better decisions. Many companies still rely on historical reporting, fragmented systems, and manual processes to anticipate operational challenges. We see an opportunity to bridge that gap through predictive intelligence. We chose to focus our initial commercial efforts on Saudi Arabia because its logistics sector presents a compelling environment for what we are building. As transportation networks, delivery volumes, and supply-chain operations expand, the cost of operational inefficiencies and poor decisions becomes increasingly significant. Predicting those challenges before they materialize can create measurable value. There is also a broader ambition behind this decision. We don't believe globally relevant technology needs to originate from Silicon Valley or established technology hubs. The Middle East has the talent, market opportunities, and increasingly sophisticated business environment to produce technology companies that compete internationally. Our ambition is to build Lensiq into a globally relevant predictive intelligence platform, starting with a market where we see a clear operational need and room to create meaningful impact. ### Part III: The product Q: Why the name Lensiq? Hijazi: Lensiq comes from lens + IQ. Because your data already has the answers hidden, but needed mathematicians to dig deep, the lens and the IQ uncover all that is buried within millions of records. Q: How does Lensiq go from a spreadsheet to a prediction with no setup? Hijazi: The spreadsheet has the data, it is uploaded to train the predictor. Once that happens, no action is needed as the predictor will build itself. All a user might need to do is plug in external real-time data sources for the predictor to keep on updating and generating predictions as parameters change. The system will automatically generate the predictions, give an accuracy and risk score, explain the outcome, and provide recommended actions that can also be automated when plugged in to execution systems. Q: Why show confidence scores so prominently? Hijazi: So the user can get enough confidence in their decision. For example, if a confidence score is below 60%, a user might opt out of a recommendation, but if it is over 90%, they will definitely need to consider the score as a flag. Q: What does auto-improving mean in practice for a customer? Hijazi: Since the system is built around a user's data, every outcome will be scored in true or false, to retrain the model and improve its thinking. As long as the system runs and gets external data and new outcomes, the system keeps on relearning and fine-tuning based on the user's behavior. And since the predictions are mathematical in core not reasoning, there is no room for hallucinations or error, or outcomes based on similar operations of a different user. Their data is isolated, and trained for every model specifically. Q: Which prediction do customers ask for most often? Hijazi: What we're seeing right now in the logistics sector is a demand for predicting 2nd delivery attempts for orders, empty miles run by a truck, and specific ETAs (more customized than GPS, as they take into consideration the internal data and operations of a company). ### Part IV: Positioning Q: Who is Lensiq for, and what job does it do for them? Describe the person who gets the most value. Hijazi: Lensiq is for land-based logistics companies that make high volumes of operational decisions under uncertainty, but lack the ability to consistently anticipate what will go wrong before it affects performance. Its primary customer is a logistics operator managing complex, interconnected operations across deliveries, fleets, routes, suppliers, customers, and distribution networks. These companies already have operational data. They may use a TMS, BI dashboards, spreadsheets, or internal reporting systems. They can see what happened, monitor what is happening, and track KPIs. What they struggle with is knowing what is likely to happen next, which decisions are becoming risky, and what action should be taken before the problem materializes. Lensiq turns their historical operational data into customer-specific predictive intelligence that helps them make better operational decisions and prevent avoidable losses. The ideal customer is a Head of Operations, COO, or Operations Director at a mid-sized or large land-logistics company. Lensiq helps this person answer four operational questions: 1. What is likely to happen? Identify patterns and predict potential operational outcomes. 2. Where is the risk? Identify routes, deliveries, suppliers, or operational activities likely to underperform. 3. Why is it happening? Understand the factors contributing to the predicted outcome. 4. What can we change? Identify preventive actions that can improve the expected outcome before the problem occurs. Q: How do you view your positioning relative to Akkio, Obviously AI, DataRobot, Microsoft Power BI and Google Sheets with Gemini? Is there another competitor you think of first? Hijazi: Lensiq operates in the broader predictive analytics and business intelligence landscape, but our positioning is more specialized. Platforms such as Akkio, Obviously AI, and DataRobot provide tools for building and deploying predictive models, while Microsoft Power BI and Google Sheets with Gemini focus primarily on data analysis, reporting, and productivity. Our approach is different. Lensiq is built specifically for land-based logistics operations, helping companies use their own historical operational data to anticipate risks, understand potential disruptions, and make preventive decisions before those issues translate into operational or financial losses. We also see companies such as FourKites and project44 as relevant players in the logistics intelligence market. While their platforms focus heavily on supply-chain visibility, shipment tracking, and transportation intelligence, Lensiq focuses on customer-specific predictive intelligence built around each operator's historical data and operational context. Beyond direct competitors, we recognize that existing TMS platforms, spreadsheets, dashboards, and manual decision-making are significant alternatives. Our objective is not to replace the systems logistics companies already use, but to add a predictive intelligence layer that helps them get more value from their existing data and improve operational decision-making. Ultimately, we see Lensiq competing on a specific job: helping logistics operators move from understanding what has happened to anticipating what is likely to happen and deciding what to do about it. Q: What is your core moat in no-code predictive analytics? Hijazi: Our core moat is not the ability to build predictive models without code. That capability is becoming increasingly accessible. Our differentiation lies in combining logistics-specific predictive models, customer-specific historical data, and operational context into a platform designed around real logistics decisions. Every logistics operator has different routes, delivery patterns, fleet constraints, suppliers, operating conditions, and performance histories. Lensiq uses each customer's own data and parameters to develop predictive intelligence that reflects how their operations actually work. Over time, this creates an opportunity for deeper differentiation through four areas: 1. Domain specialization: Models and prediction workflows designed around land-based logistics problems rather than generic business use cases. 2. Customer-specific intelligence: Predictions built around each company's historical data, operational patterns, and parameters. 3. Operational context: Connecting predictions to the conditions influencing outcomes, making them more relevant to actual decisions. 4. Continuous improvement: Using subsequent operational outcomes to evaluate predictions and improve model performance over time. We see the long-term moat in the accumulated combination of logistics domain knowledge, customer-specific data structures, predictive performance, and integration into operational decision-making. The no-code interface makes predictive intelligence accessible. The deeper value is the specialized intelligence and operational learning built behind it. Q: What are some of the use cases for Lensiq that are not obvious and can add a lot of value to users? List as many as possible. Hijazi: Finance and investment: predicting investment returns and downside risk; financial distress and cash-flow shortages; credit defaults and repayment behavior; portfolio liquidity and exposure risks. Healthcare and healthtech: predicting patient deterioration and health risks; hospital readmissions and complications; treatment response and non-adherence; hospital demand and resource shortages. Insurance: predicting claim probability and severity; fraudulent claims; policyholder churn; accident and property damage risks. Human resources: predicting employee turnover and absenteeism; candidate hiring success; employee retention and workforce shortages; training effectiveness. Manufacturing and industry: predicting equipment failures; production interruptions and delays; manufacturing defects; raw material shortages and maintenance requirements. Retail and e-commerce: predicting customer churn and lifetime value; inventory shortages and excess stock; product returns; promotion and pricing outcomes. Real estate: predicting property values and investment returns; rental defaults and tenant turnover; property maintenance costs; occupancy and vacancy rates. Agriculture and food: predicting crop yields and disease risks; livestock health risks; food spoilage; harvest timing and supply shortages. Energy and utilities: predicting electricity demand and energy consumption; infrastructure failures; renewable energy production; water consumption and leakage. Banking and fintech: predicting loan defaults; customer churn; suspicious transaction patterns; debt collection recovery. Education: predicting student dropout and retention; academic performance; course completion; future enrollment demand. Telecommunications: predicting customer churn; network congestion; equipment failures; service interruptions. Supply chain and procurement: predicting supplier failures and disruptions; shipment and purchase order delays; inventory replenishment requirements; procurement costs and supply shortages. Sales and business development: predicting sales conversion and deal closure; revenue shortfalls; customer lifetime value; major account churn. Legal and enterprise risk: predicting contractual disputes; compliance failures; operational incidents; regulatory and contractual risks. Core opportunity: Turn historical data and operational context into customer-specific predictions that help organizations anticipate outcomes, manage risks, and make better decisions. Q: What is the one detail in Lensiq you are proudest of? Hijazi: What I am proudest of is that we are making something that was traditionally accessible only to organizations with significant data science resources available to businesses without requiring them to build that infrastructure themselves. We built Lensiq around a simple but ambitious idea: every organization has valuable intelligence hidden in its own historical data, but very few have the tools or expertise to turn it into predictive capability. I am proud that we are turning that idea into a practical platform, where companies can use their own data to build predictive models, anticipate risks, and make better decisions. We are not just making predictions easier to generate. We are making predictive intelligence more accessible, practical, and relevant to the decisions businesses make every day. ### Part V: Launch Q: What did you learn from your launch? Hijazi: Launching on Product Hunt taught us that having an interesting technology is not enough. People need to understand immediately what problem it solves, who it is for, and why they should care. We learned that positioning matters as much as the product itself. When a platform has broad technical capabilities, it is easy to communicate what it can do without making its actual value clear. The launch also reinforced the importance of speaking to a specific audience rather than trying to appeal to everyone. Most importantly, it reminded us that launching is not validation. Real validation comes from getting the product into the hands of customers, understanding how they use it, and seeing whether it makes a measurable difference to their decisions. ### Part VI: The hard part Q: How many people use Lensiq today? Hijazi: Lensiq currently has two companies using the platform. We have shifted our focus toward mid-sized and large organizations rather than individual users, as our product is designed to address complex business and operational decisions. We have also started concentrating our commercial efforts on the logistics industry in Saudi Arabia, where we see a strong opportunity to apply predictive intelligence to land-based logistics operations. Q: What is a challenging time you had building Lensiq? Hijazi: One of the most challenging parts of building Lensiq has been navigating geopolitical instability and ongoing wars while trying to build a company in an already difficult startup environment. We have been bootstrapping, operating with limited resources, and growing without having secured external funding yet. This has meant making difficult decisions about priorities, product development, and how far we can stretch our resources. Sales has also been a major challenge. Building a product is one thing; getting larger organizations to understand, trust, and adopt a new predictive intelligence platform requires time, education, and persistence. Despite these challenges, we have continued building, refining our positioning, and narrowing our commercial focus toward mid-sized and large logistics companies in Saudi Arabia. The hardest part has been maintaining execution and momentum despite circumstances that are largely outside our control. Q: Who believed in you before there was anything to show? What did they say or do? Hijazi: A few friends who were also building startups believed in the idea early on. Interestingly, they were already trying to figure out ways to predict different aspects of their businesses but lacked the tools and resources to do so effectively. When we introduced Lensiq, they immediately understood the potential because they had experienced the problem themselves. For them, we were addressing a real need rather than introducing technology in search of a problem. Their early interest and willingness to explore the platform gave us confidence that there was a genuine demand for what we were building, even before we had much to show. Q: How do you convince small businesses that prediction is for them, not just big companies? Hijazi: We start by showing them that prediction is not reserved for companies with large data science teams or massive budgets. Every business, regardless of size, makes decisions under uncertainty. A small business deciding how much inventory to purchase, which customers might stop buying, or whether cash flow will cover upcoming expenses faces the same fundamental challenge as a large enterprise. The difference is often the scale of the decision, not the need for better information. Lensiq makes predictive intelligence accessible by allowing businesses to use their existing historical data without needing to build an in-house data science team. We focus on practical use cases where anticipating an outcome can save money, reduce risk, or improve a business decision. We don't ask small businesses to adopt prediction because it is new technology. We show them where it can make a measurable difference to their bottom line. ### Part VII: What it is all for Q: What would make you say Lensiq has succeeded? Hijazi: I would consider Lensiq successful when it becomes an essential part of how organizations make decisions, rather than simply another software subscription. If companies across logistics and other industries rely on Lensiq to anticipate risks, allocate resources more effectively, prevent avoidable losses, and improve measurable business outcomes, we will have achieved what we set out to do. Commercially, success means building a sustainable company with strong customer retention, measurable customer value, and the ability to scale across markets. But ultimately, I want Lensiq to help make predictive intelligence a standard business capability, accessible to organizations that previously couldn't afford the expertise or infrastructure to build it themselves. Q: How will AI change the way businesses make decisions from their own data? Hijazi: I think businesses will become more cautious, not less, about how they use AI to make decisions based on their internal data. Companies hold sensitive information about their customers, operations, finances, suppliers, and competitive advantages. Many will be reluctant to expose that information to external AI systems, regardless of how powerful those systems become. Data privacy, ownership, and control will become increasingly important barriers to adoption. This is where I see an opportunity for Lensiq. Rather than asking companies to hand over their data to a general-purpose AI system, we provide a controlled and isolated environment where they can connect their own data, define the parameters they want to use, and develop customer-specific predictive models. The objective is to give businesses the benefits of predictive intelligence while maintaining control over their data and how it is used. I believe the future will not belong exclusively to companies that adopt the most AI. It will also belong to those that find ways to use intelligence without compromising their data, privacy, or competitive advantage. Q: What is one thing about the AI industry that you think most people are getting wrong? Hijazi: I think the industry is too focused on making AI more powerful and accessible, while underestimating one of the biggest barriers to enterprise adoption: trust. There is an assumption that businesses will eventually connect all their internal data to increasingly powerful AI systems because of the value those systems can generate. I don't think it is that simple. Companies spend years building their customer relationships, operational processes, financial intelligence, and competitive advantages. Their data is one of their most valuable assets. Expecting them to expose all of it to external AI systems without meaningful control over how it is accessed, processed, and used is a major assumption. I believe privacy, data ownership, and control will become defining factors in enterprise technology adoption. The question will not just be what AI can do with a company's data, but whether that company is comfortable allowing it. This is also part of the thinking behind Lensiq. We want companies to benefit from predictive intelligence without having to compromise control over their information. Our approach is built around isolated customer environments, controlled data access, and customer-defined parameters. I also think the industry sometimes confuses generating an intelligent answer with improving a business decision. Producing an answer is relatively easy. Producing a reliable prediction, understanding its operational context, and helping someone make a better decision is a much more demanding challenge. I believe the next phase of enterprise intelligence will be shaped not only by the capabilities of AI, but by how securely, reliably, and responsibly those capabilities can be deployed inside organizations. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Hijazi: Over the next 12 months, our priorities are to secure a pre-seed funding round, strengthen our position in the Saudi logistics market, and turn Lensiq into a commercially validated platform. Our main milestones are: 1. Funding: Close a pre-seed round to accelerate product development, commercial expansion, and team growth. 2. Client acquisition: Reach at least 10 mid-sized and large enterprise clients, initially focusing on land-based logistics companies. 3. Revenue: Surpass $500,000 in annual recurring revenue (ARR). 4. Product development: Strengthen our predictive models, improve prediction accuracy, and expand the range of logistics use cases supported. 5. Customer retention: Establish measurable customer outcomes, strong adoption, and recurring usage across our client base. 6. Market expansion: Build a stronger commercial presence in Saudi Arabia and establish a foundation for expansion into additional GCC markets. 7. Strategic partnerships: Develop partnerships with logistics technology providers, system integrators, and industry players to accelerate distribution and adoption. Our focus is to move from early adoption to repeatable enterprise sales, demonstrate measurable value for customers, and build a scalable commercial model. Q: Where do you want Lensiq to be in five years? Hijazi: In five years, I want Lensiq to become a global leader in predictive intelligence, fundamentally changing how organizations use their own data to make decisions. Our ambition is to make predictive intelligence as essential to businesses as business intelligence is today. We want organizations to move beyond understanding what happened and start anticipating what will happen, where risks are emerging, and what decisions can improve their outcomes. We aim to establish Lensiq as the leading predictive intelligence platform for logistics and supply-chain operations, first across the GCC and then internationally, serving major enterprises and complex operational networks. Beyond logistics, we see significant opportunities to bring our technology into other industries, including finance, healthcare, insurance, and manufacturing. I want Lensiq to build a substantial global customer base, establish strategic partnerships with major enterprise technology players, and become a category-defining company. Ultimately, the ambition is much bigger than building a successful software company. We want to make predictive intelligence an accessible, trusted, and indispensable part of how businesses operate, regardless of their size or industry. Q: What are your hopes and fears for a post AGI world? Hijazi: My biggest hope for a post-AGI world is that intelligence becomes more accessible and helps humanity solve problems that previously required enormous resources, expertise, and time. Businesses could make better decisions, healthcare could become more predictive, and entire industries could operate more efficiently. My biggest fear is the concentration of power. If a small number of companies or institutions control the most powerful intelligence systems, they could gain unprecedented influence over economies, businesses, information, and individuals. I am also concerned about privacy and data ownership. As intelligence becomes more powerful, access to sensitive information becomes even more valuable. I don't think businesses should have to sacrifice control over their data to benefit from technological progress. I believe the challenge will not simply be how intelligent machines become, but how responsibly we build, govern, and distribute that intelligence. My hope is that AGI expands human capabilities rather than makes human judgment, independence, and accountability irrelevant. Technology should give people more control over their decisions, not less. About Lensiq: Lensiq is a predictive intelligence platform for land-based logistics, founded by Ahmad Hijazi and Ahmad Freijeh. It turns a company's own spreadsheets and data feeds into customer-specific predictions, risk scores and recommended actions, without a data science team. lensiq.app ## Legal Arena (India) URL: https://startup.talk/india/legal_arena/ Founders: Divyanth Jayaraj Divyanth Jayaraj grew up in Bangalore in a high-achieving family, built a first game in Visual Basic at 12, and spent more than ten years in software, much of it as a contractor on projects at Fortune 500 companies and startups, each lasting six to eleven months. About two years ago Jayaraj quit that job to find the next adventure, and the result is Legal Arena. Legal Arena lets players step into the courtroom as a lawyer or a judge: interview AI clients, uncover evidence, negotiate settlements and argue before an AI judge against AI opponents or other players. A November 15 update added running a whole law firm. We spoke about why lawyering is combat with words, how much work it took to make an AI client sound sincere, and why the first paying customers arrived with no marketing at all. Key takeaways: - Jayaraj says the hardest part was making the AI believable: the first responses mixed up genders and kept saying it did not know, until less pre-written backstory turned out to work better. - The AI judge weighs only the arguments, evidence and admitted witness testimony presented in court and explains its reasoning, though Jayaraj says no one can guarantee every judgment is fair. - Instagram and Google ads brought dead traffic; every paying user so far arrived without marketing. There are about 2,000 signups, of whom about 71 have paid. - Legal Arena is self-funded, gets around 100 visitors a day, and 8 to 12 of them complete an entire case. The Principal / Founder: - Divyanth Jayaraj (Founder) Grew up: Bangalore, India Before: 10+ years in software at Fortune 500 companies and startups Also built: Quiet Sky Funding: Self-funded Website: legalarena.app LinkedIn: Divyanth Jayaraj (https://www.linkedin.com/in/divyanth-jayaraj-5761b0159/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Jayaraj: I grew up in Bangalore, raised by a high achieving family. So there was a lot of family pressure to perform in school. Q: What first got you into building things? Jayaraj: As a child, I built my first game using Visual Basic. I briefly discovered the joy of building when I was 12 before student responsibilities took over. I got it back in my third year of engineering when I was asked to build portfolios for US universities. But things really took off only recently, some 2 years ago when I quit my job and wanted to find my next adventure. Q: What was your first job, and what did it teach you? Jayaraj: My first job was as a frontend developer. It taught me the joy of doing things for others and being paid for it. ### Part II: The road to Legal Arena Q: What made you start Legal Arena? Jayaraj: I like thinking in systems. Anything I look at, I enjoy finding the systems underneath. For the longest time in life, I didn't understand the legal system or what it is that lawyers do until I started watching a few TV shows like Suits and Breaking Bad. Then, I understood that lawyering was basically combat with words. But it was still too hard to systemize into a game; until AI came along. So in short, it was 1) Ability to see systems and 2) AI that got me started Legal Arena. Q: Why turn the courtroom into a game? Jayaraj: I've always enjoyed being right on the internet. Back in the Orkut, Facebook and Quora days I enjoyed writing, making a valid point and seeing the likes and views soar. Now that AI is here, I thought it was a huge opportunity to let people enjoy this fix of being right. I've thought of a few debate related SaaS products to build but an AI courtroom game was the most obvious starting point. Q: How did 10+ years at Fortune 500 companies and startups prepare you to build a game as an indie developer? Jayaraj: I was a contractor in those Fortune 500 companies with projects each lasting 6 to 11 months. A lot of them were products started from scratch. No matter how big or small the company is, people want the same thing; build something for them and have it work in such a way that it brings them happiness. Working there helped me hone my craft. It helped me see things like a developer rather than a consumer. This basically helped me build large systems and put them together by myself which would otherwise have become clunky despite AI use. Q: Who did you imagine playing Legal Arena: law students, lawyers or people who love courtroom dramas? Jayaraj: It's mainly for people who enjoy courtroom dramas - people who like being right. I haven't gone to law school but I've heard that they sorely lack debating skills or practice opportunities so maybe it helps them too? ### Part III: The product Q: Why the name Legal Arena? Jayaraj: It's one of the names ChatGPT suggested. It was obvious so I went with it! Q: How do you make AI clients, witnesses and opponents feel believable? Jayaraj: I had to go through enormous number of iterations to get this one right. At first the responses were partly deterministic. As a result the responses would get genders mixed up or give default answers that made no sense. Then, the AI would constantly say it doesn't know but that was because I was trying to draw from a canonical story and kept truths as they are. Eventually I discovered that less is more and it's ok for the actual canonical story to emerge later. That's when the AI started sounding more believable. But still, it has a long way to go. The AI still sounds curt and robot-like. Q: How does the AI judge reach a verdict, and how do you keep it fair? Jayaraj: When you interview your client before going to court, a factsheet gets populated from their replies. It tracks the facts gathered, available evidence, gaps and risks. The opposing counsel also has a separately prepared case file. The judge considers the arguments, evidence and admitted witness testimony actually presented in court, alongside the rules applicable to the case. It asks follow-up questions and explains the reasoning behind its final judgment. Can I guarantee every judgment is fair? No. But I’ve built safeguards to keep decisions grounded in the courtroom record, give both sides an opportunity to respond, and prevent the judge from relying on hidden information. It’s something I’m continuing to improve. Q: What happens in Judge Mode, and why did you add it? Jayaraj: I've always been fascinated by what real life judges do in court. It's fun to observe the judges' thought process, what do they mean by fair, should they go by the letter of law or the spirit of law and what incentives exist for them through their judgments. I've also had a strong personal sense of justice myself. So, I wanted to put these together and see for myself how good a judge I could be. I originally wanted to build an AI judge game but then I forgot about it. Later, after building Legal Arena, I remembered the judge thing and so here it is! Q: What have players taught you about how they argue and negotiate? Jayaraj: I haven't had the opportunity to systematically study how players argue and negotiate yet. I've been so occupied with building and improving Legal Arena that analyzing player behaviour is something I still need to spend more time on. But that's actually one of the things I'm most excited about. I want to understand whether people approach arguments logically, emotionally, aggressively, or creatively. Do they try to establish the truth, or do they simply try to win? Do they respect the rules of the courtroom, or do they look for ways to bend them? Unlike traditional games, Legal Arena gives players enormous freedom in how they express themselves. I'm particularly interested in discovering what people do with that freedom when they're confronted with an opponent who can actually respond to their arguments. Q: What do you deliberately leave out of Legal Arena? Jayaraj: The everyday gruntwork that lawyers do, like drafting, writing contracts, studying contracts, finding red flags etc. I haven't found a playful way to systemize that yet, but only because I'm not a big fan of reading. ### Part IV: Positioning Q: Who is Legal Arena for, and what job does it do for them? Describe the person who gets the most value. Jayaraj: We have games in which you get to be soldiers, sailors, commanders, politicians, entrepreneurs, even chefs. It's about time there was one where you get to be a lawyer. Traditional lawyer games like Ace Attorney force you to pick from fixed dialogue choices. Legal Arena is the first game I've seen where you can type in free text and an actual AI responds accordingly. Who is it for? Right now, it's for people who love courtroom dramas but haven't studied law. It's for people who enjoy verbal one-upmanship. Q: How do you view your positioning relative to Ace Attorney, AI Dungeon, debate apps and legal education tools like Quimbee? Is there another competitor you think of first? Jayaraj: I personally haven't played those games but I've heard of Ace Attorney and as far as I know, it's a lawyer game with limited prewritten dialogues. Mine is the complete opposite. I've heard of AI Dungeon as well but that's quite limited in what you can do. But I'm sure AI-powered debate apps are a thing and have a huge growth opportunity. Q: What is your core moat in AI-driven courtroom games? Jayaraj: The core moat is how much work went into making the AI sound believable and at the same time offer some degree of challenge to the player. Besides, as of the November 15 update, it’s not just a courtroom game. You get to run an entire law firm, build a clientele, recruit staff, perform shady actions such as illegally obtaining evidence and so on. Q: What are some of the use cases for Legal Arena that are not obvious and can add a lot of value to users? List as many as possible. Jayaraj: Besides entertainment, I would say 1. Legal Arena gives you a place to practice defending others with words. 2. It's an excellent debate practice tool. 3. It teaches people to research, get info, organize the info and use the info to negotiate or debate. Q: What is the one detail in Legal Arena you are proudest of? Jayaraj: I'm proud of how the AI sounds sincere and believable, especially at the client interview stage. It's not perfect. I could still make it be greedy or scared and whatnot but getting a believable client was a lot of work. ### Part V: Launch Q: Which moment from the early days stands out most? Jayaraj: Getting my first paying customer. The game wasn't even finished. The AI was so bad. I was still using 300 pre-generated cases. The UI was horrible. But the user still paid for it. ### Part VI: The hard part Q: How many people use Legal Arena today? Jayaraj: I get daily traffic of about 100 users. About 8 to 12 of them complete an entire case as seen in the /blog section's auto-generated case reports. Q: What is a challenging time you had building Legal Arena? Jayaraj: Getting the AI to sound believable. The key challenge was to gauge how much info you need to give the AI exactly to sound natural. Q: Who believed in you before there was anything to show? What did they say or do? Jayaraj: No one but me. Even now, friends and family haven't bothered trying legal arena. Being an indie dev can be quite lonely. Q: How do you find players for a game that sits between education and entertainment? Jayaraj: I honestly don't know. I tried marketing on instagram and google ads but none of them worked. They got me a lot of dead traffic. Every paying user I have now came from doing absolutely nothing. ### Part VII: What it is all for Q: What would make you say Legal Arena has succeeded? Jayaraj: I have about 2000 signups as of today and about 71 of them have paid. I'd call that a success. Q: Will AI judges and lawyers ever play a real role in resolving disputes? Jayaraj: I absolutely believe so. I don't know about lawyers but we will certainly have AI judges in the future. ### Part VIII: The road ahead Q: Where do you want Legal Arena to be in five years? Jayaraj: It will be pretty cool if law students are told to play legal arena. Q: What are your hopes and fears for a post AGI world? Jayaraj: AGI is the way to go. It's certainly not there to cut your workload but instead to multiply your output. Soon enough we all will be running AI powered companies. About Legal Arena: Legal Arena is an AI courtroom game built by Divyanth Jayaraj. Players act as a lawyer or a judge, interview AI clients, gather evidence, negotiate settlements and argue cases against AI opponents or real players before an AI judge, and can run a whole law firm. legalarena.app ## Decklit (Singapore) URL: https://startup.talk/singapore/decklit/ Founders: Luca Simeone Luca Simeone grew up in a small town in Italy and spent most of his childhood outside on a bike, always looking for the blue ocean, the place nobody else had gone yet. A degree in Marketing took him to the UK, and from there to Barcelona, Stockholm, Amsterdam and, since 2019, Singapore, where he lives with his family. Before Decklit he led marketing and commercial work across Europe and APAC in travel, technology and crypto. Decklit grew out of years of watching good decks fail on bad screen shares, and out of an MSc in Computer Science at the University of Liverpool, where Simeone measured how slides travel over real networks in Asia. The platform handles everything after a deck is made: share it as one secure link, present live in the browser with nothing to install, take questions on each slide, and see afterwards who engaged and with what. We spoke about why Decklit leaves slide creation to others, what Dual View makes possible, and building a company alongside a full-time job and a family. Key takeaways: - Decklit does not make slides. It takes a finished deck and handles sharing, live presenting, audience questions, notes, AI summaries and analytics, all tied to the slide they belong to. - Slides are synced in each viewer's browser instead of screen shared. In Simeone's dissertation tests this sent about 80% less data, so sessions hold up on weak connections. - Dual View keeps the deck, a shared screen and the participants in one window, so a live demo never makes the slides or the faces disappear. - Simeone built Decklit in evenings and weekends alongside a full-time job, a master's degree and a family. It has hundreds of free users and a small number of Pro users, with a free trial and a Business plan for teams planned. The Principal / Founder: - Luca Simeone (Founder) Based in: Singapore (since 2019) Grew up: Italy Studied: BA Marketing (UK); MBA; MSc Computer Science, University of Liverpool Website: decklit.com ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Simeone: I grew up in a small town in Italy. I was an adventurous kid and spent most of my time outside, riding around on a bike with friends. I was always drawn to the unusual things, and in almost everything I did I was looking for the blue ocean, the place nobody else had gone yet. I was fascinated by the world outside the one I lived in. I had this urge to be the different one for once, and to test myself in places where I would not feel comfortable at first. It came from within, and I think it is what eventually brought me to the East. At secondary school I studied computer science and accounting. My family always encouraged us to go abroad, so I moved to the UK for university, took a BA in Marketing and stayed for five years. From there I kept moving. I worked for a digital advertising agency in Barcelona, then went to Stockholm for a project with my sister, which I developed through the Erasmus for Young Entrepreneurs programme, and worked in email marketing for an e-commerce business there. Next came Amsterdam and several years with international companies. In 2019 I had the chance to move to Singapore, and I have been here with my family ever since. Having spent more than half my life abroad, I do not feel I belong to one specific place, though I consider where I am now a place to stay. I kept studying alongside work, first an MBA and then an MSc in Computer Science with the University of Liverpool, where my dissertation became the starting point for Decklit. Q: What first got you into building things? Simeone: I never had one specific passion. What I had was a wish to be independent, and to build a company, not only work for one, and to create something people would use. That, and a fair amount of grit, is what got me here. Q: What was your first job, and what did it teach you? Simeone: Apart from the odd jobs I did when I was young, my first real job was at a Starbucks in London, right across from Harrods, one of the busiest around. My English was not good at the time, so I made up for it with speed, with knowing the coffees, and with how I treated customers. That is where I learned to work under pressure, and it kept me going while I got started in the UK. My first corporate job was at an international email marketing company, one of the best in its field. That is where I learned the basics of CRM and of corporate life, and where I became meticulous and structured, thanks in large part to my manager at the time. ### Part II: The road to Decklit Q: What made you start Decklit? Simeone: It was several things adding up. In corporate life, clients kept asking the same question, “Can you share the deck before the presentation?” Then came the calls themselves, with “let me share my screen” or “can you see my screen?”, the glitches and the blurry slides. Presenting across Southeast Asia made it worse. In remote areas, in parts of the Philippines for example, the bandwidth is weak, and a screen share simply does not hold. All of that made me ask whether there was a better way. Screen sharing does what its name says, it shares a screen. It was never designed for presenting a deck. Similar technology existed, but we focused on one idea and refined it. We put the deck, every single slide of it, and the audience at the centre, and wrapped the call around them. In Decklit the call comes second. The deck and the people watching it come first. Q: What did years of presenting to clients across Europe and APAC teach you about how decks get delivered? Simeone: That a deck deserves more than a screen share. So much time and effort goes into one, and then it is delivered through a tool that was never made for it, on connections that often cannot carry it. It also taught me that we put all our attention on the session itself and barely think about what comes before and after. There was no single place to share the deck in advance, present it, deliver it and collect feedback, in a simple way and with nothing to download. Expensive solutions exist for large enterprises, but there was nothing like that for small and mid-size companies and startups. An affordable one can empower a lot of people. I call it the last mile of the presentation flow, which is sharing, presenting, following up and analysing. Analysing is the part we miss most often. We guess how long people spent on a slide or what was said about it. For me, everything needed to be connected, and it was not. Q: Why did you decide to focus on what happens after a deck is made, rather than on making slides? Simeone: Because everyone is focused on making slides, and very few are working on what happens after. I do not mean delivery alone, or the call alone. I mean everything around the deck, combined in one place. A deck should arrive crisp. The audience should be involved and engaged, not just watching. They should be able to play the videos in it at their own pace, leave feedback without friction, and come back later to one place where the questions, the notes and the files are easy to find. The large suites do offer most of this, but spread across several products, and finding anything has become cumbersome. There is also a practical reason. Many companies rely on their own teams to create slides, and slides are hard to get right. Even with AI you always have to retouch them. Several models already do an excellent job, and that market is moving so fast and is so crowded that entering it now would be a losing battle. The space after the deck is made is where not many were looking, and it is where I knew I could build something useful. Q: What is it like to build Decklit from Singapore? Simeone: It suits the product. Singapore sits in the middle of a region where audiences are spread across countries, languages and very different internet connections, which is exactly the situation Decklit was designed for. I can test an idea with someone in a modern office here and with someone on a weak mobile connection elsewhere in the region in the same week. It is also a serious place to start a company, with a strong startup community and clear rules. My developers are in Italy, so the time difference works in our favour. I brief them in my evening and wake up to progress. The other side of that is being the only one awake when something breaks. I have lived here since 2019 with my family, and building for the whole region from here feels natural. ### Part III: The product Q: Why the name Decklit? Simeone: It means a deck that is lit, as in under the spotlight. In most tools the call is the main event and the slides are a guest. I wanted the opposite, with the deck in the light and everything else arranged around it. The name also had to be short, easy to say in any language and easy to remember, and Decklit does all three. Q: Why present without screen sharing and with nothing to install? Simeone: Because every extra step loses part of your audience. With Decklit the viewer clicks a link and is in. There is nothing to download, and they do not even need an account, since they can join with a nickname. Presenting without screen sharing changes the quality too. The viewer’s browser shows the real slide and not a video of my screen, so the text stays sharp, videos in the slides play smoothly on their side, and the session uses far less data. In the tests for my dissertation, syncing slides sent about 80% less data than sharing a screen on a video call. Nobody sees my desktop or my notifications either. Screen sharing is still there when you need to show something else, but it is no longer what the presentation depends on. Q: What does a Decklit session look like from the audience’s side? Simeone: Most tools are designed for the presenter, and the audience gets whatever is left. In Decklit the viewer clicks a link and is in, with nothing to install and no account needed. They see the real slide, sharp on any screen, including a phone. They can ask a question on the slide it relates to, take private notes that stay attached to that slide, and play any video in the deck on their own device. After the session they have a place to go back to. The Deck Room for that deck holds the questions, their notes, the files the presenter shared and the recording, and they can leave feedback there, even if they could not attend. I wanted the audience to feel that the session was built for them as much as for the person presenting. Q: You researched the environmental side of presenting. What did you find, and does it matter to customers? Simeone: My master’s dissertation compared screen sharing with sending the slides themselves and keeping them in sync, and measured bandwidth, energy use and quality of experience, with a focus on Asia. In my prototype tests, syncing slides sent about 80% less data than sharing a screen on a video call. Less data means less strain on networks and less energy, and it also means the presentation works for people on weak connections. So the sustainable choice and the practical choice turn out to be the same one. Most customers come for the second reason. I think more of them will care about the first as companies start to count the footprint of how they work online. Q: How do you keep a shared deck secure when you never send the original file? Simeone: Not sending the file is the security. The original stays with you, and people view the deck through Decklit, so there is no attachment travelling from inbox to inbox and being forwarded without your knowledge. On top of that, you decide how people see it. They can browse freely, or only follow you slide by slide during the session, so nobody reads ahead to the pricing page. You can protect a deck with a password, and you can see who opened it and when. And if a deck or a file does need to be handed over, that happens inside the platform too. Every deck has its own Deck Room, a private space where the presenter shares files, recordings and summaries with the audience, so nothing has to leave the system to reach the right person. With email, your control ends the moment you press send. With Decklit it continues afterwards. Q: What do the AI summaries and analytics tell a presenter that they could not see before? Simeone: They show you what happened in the room when you could not see the room, and they do it slide by slide. From the moment a live session starts, Decklit follows the deck. Every slide is analysed on its own. You see how long you stayed on it, how many people were with you, and which questions were asked while it was on screen. The transcript works the same way. It is a smart transcript that knows which slide you were on when you said something, so you can go straight to what was said on slide seven without scrolling through an hour of text. It also looks at whether what you were saying was relevant to the slide in front of the audience, which tells you where you stayed on message and where you drifted. The AI summary then pulls the session together. It gives you the key points, the questions that came up and what needs a follow-up, without anyone taking notes. Outside the live session, the analytics show who opened the deck, when, and how much time they spent on each slide. Before this, a presenter finished a session with a feeling about how it went. Now they finish with a record of what was said, what was asked and what held attention, all connected to the slide it belongs to and saved in the Deck Room for that deck. And there is more to come! Q: You described Decklit many ways before it clicked. What did you learn from that? Simeone: I learned that people understand what you are not faster than what you are. I tried “the last mile of presentations”. I tried “the deck is the hero”. I tried lists of features, and I tried comparing us to tools people already knew. Each description was accurate, and none of them landed in the first ten seconds. What clicked was the simplest line: creation is solved and delivery is fragmented. The moment I said that we do not make slides and that we look after everything that happens after, people stopped comparing us to PowerPoint and started telling me their own stories about screen sharing going wrong. The second lesson was to stop describing the technology. Nobody outside the industry cares how slides are synchronised. Everyone has lived the moment of asking “can you all see my screen?”, or of sending a deck and hearing nothing back. When I describe that moment, people recognise themselves, and the product explains itself from there. It also forced me to be disciplined about the product. If I cannot explain a feature through a moment people recognise, it probably should not be there. Q: What do you deliberately leave out of Decklit? Simeone: Slide creation, first of all, and I intend to keep it that way. You bring the deck from PowerPoint, Google Slides, Keynote or an AI tool, and Decklit takes it from there. Companies are protective of how their decks are made, the tools for making them are already excellent, and that market is crowded enough. I also resist turning Decklit into a general meeting tool. If your meeting is not about a deck, another product will serve you better, and I would rather be very good at one thing. We leave out anything the audience would have to install. No plugin, no desktop app, no account required to join. And we leave out the weight of an enterprise suite, where every feature exists somewhere but you need training to find it. Every feature we add has to answer one question, which is whether it helps the deck land. If it does not, it stays out, however popular it is elsewhere. ### Part IV: Positioning Q: Who is Decklit for, and what job does it do for them? Describe the person who gets the most value. Simeone: Decklit is for people whose work depends on a deck landing well. That includes consultants presenting findings to a client, trainers and learning teams running sessions, sales teams walking a prospect through a proposal, founders pitching to investors, speakers at conferences, and leaders presenting to their own organisation. Most of them work in small and mid-size companies or on their own. They do not have an IT department to roll out an enterprise platform, and they should not need one. They are good at what they do and tired of fighting their tools. The job Decklit does for them is to take the deck from finished to delivered. They share it with a secure link, present it live in the browser, take questions on each slide, let the audience make notes and leave feedback, and afterwards they know who engaged and with what. Everything for that deck, including the recording, the transcript, the summary and the files, stays together in its Deck Room. The person who gets the most value is the one who presents often and follows up seriously. For them, knowing which slide raised the most questions is the difference between a second meeting and silence. Q: How do you view your positioning relative to DocSend, Pitch, Google Slides sharing, Brieflink and Zoom screen share? Is there another competitor you think of first? Simeone: Each of them owns one piece of the journey, and each does that piece well. DocSend and Brieflink are about what happens to a document after you send it. They tell you who opened it, but they stop when the live presentation begins. Pitch and Google Slides are where the deck is made, and their sharing is an extension of the editor. Zoom is where you talk over the deck, and to do that it turns your slides into a video of your screen. A presenter ends up moving between three or four tools for one presentation, and the information is scattered across all of them. Decklit connects those moments around the deck itself. You share, present, engage and analyse in one place, and it works alongside the tools people already create in. We complement them more than we replace them, and many of our users still use a call tool for meetings that have nothing to do with a deck. The competitor I think of first is not a product. It is the habit of attaching a PDF, sharing a screen and hoping for the best. That habit is free, everyone knows it, and it is what we are up against every day. Q: What is your core moat in presentation delivery? Simeone: The first part is how the product is built. Everything in Decklit is tied to the slide. Questions, notes, analytics, transcripts and summaries all know which slide they belong to, because the deck was at the centre from the first line of code. A tool built around a call or around a document would have to be rebuilt to work that way. It cannot be added as a feature. The second is Dual View. The deck, a shared screen and the participants stay together in one window, so a presenter can open a demo without the slides or the faces disappearing. I have not seen that anywhere else. The third is that you can present from anywhere and on any device, including a phone, even on the road. We tested it in the Philippines on a 4G connection and it worked very well. The reason is simple. The deck is not being sent as a video of someone’s screen. It is already there, in each viewer’s browser, and only the slide change has to travel. Behind that is the research. I did not start from an opinion. I measured how slides travel over real networks for my master’s dissertation, with a focus on Asia, and that work shapes every technical decision. The last part is focus. We serve small and mid-size companies at a price they can afford, in a region the large players treat as an afterthought. Being close to those users, and building only for the deck, is hard to copy for a company with a hundred other priorities. Q: What are some of the use cases for Decklit that are not obvious and can add a lot of value to users? List as many as possible. Simeone: Pitching on the spot at a conference, where you meet an investor, share a link, and they follow the deck on their own phone while you talk them through it Large conferences and company presentations, where people at the back squint at a screen far away, and with Decklit everyone has the deck sharp on their own phone, in sync with the speaker Asking questions from the audience on the slide itself, without raising a hand or interrupting the speaker Presenting from a phone while travelling, on a 4G connection, because the deck is already in each viewer’s browser and nothing is being screen shared Reaching audiences in areas with weak bandwidth, where a screen share freezes and a Decklit session keeps running Sending the deck before the meeting with browsing locked, so people can open it and prepare questions but cannot read ahead to the end Showing a live demo, a spreadsheet or a website beside the deck with Dual View, so the audience keeps the slide that gives it context Letting each viewer play a video in a slide on their own device and at their own pace, in full quality, because the video is not travelling through someone’s screen Rehearsing a pitch alone, then checking slide by slide whether what you said matched the slide you were on Reading back what was said on one specific slide weeks later, because the transcript is organised by slide and not as one long text Turning that slide-by-slide transcript into speaker notes for the colleague who presents the same deck next time Replacing the follow-up email with the Deck Room, where the recording, the summary, the files and the feedback for that deck are already waiting Collecting feedback and questions from people who could not attend, attached to the slides they are about Reviewing a confidential deck with a board or a client without the file ever leaving your control Giving trainees private notes saved against each slide, so their notes and the trainer’s material stay together Improving the next version of a deck by seeing which slides drew questions and which lost attention Q: What is the one detail in Decklit you are proudest of? Simeone: Dual View. In every other tool, the moment you share your screen the slides disappear, and usually the faces with them. In Decklit the deck stays where it is, the shared screen sits beside it and you still see the people you are talking to. It sounds like a small thing until you present with it. You can jump into a live demo, open a spreadsheet or show a website, and the audience never loses the slide that gives it context. You never lose your place in the story either. It was also one of the hardest things to build, because it means running the deck, the call and a shared screen together in the browser without one slowing down the others. If I can name a second detail, it is the transcript that knows which slide you were on. It is the feature that best shows what building around the deck makes possible. ### Part V: Launch Q: Which feedback from the launch surprised you? Simeone: What surprised me most was how many people arrived expecting something else. A good share of the early sign-ups came looking for a tool to create decks. I had spent months explaining that we take care of everything after the deck is made, and still the first instinct of many visitors was to look for a slide editor. At first that was frustrating. Then I understood it was the most useful feedback I could have received. It told me the market thinks of presentations almost entirely in terms of creation, and that nobody had given people a name for the part that comes after. It was not their misunderstanding. It was my message that was not clear enough. So we changed the design and the narrative of the site to make it obvious from the first screen what Decklit is and what it is not. The other surprise came from the people who did understand it straight away. They did not just say it was a nice tool. They started describing their own presentations going wrong, in detail, which told me the problem was real for them. ### Part VI: The hard part Q: How many people use Decklit today? Simeone: We have hundreds of free users, and a small number of Pro users who are testing Decklit in real situations, with real clients and real audiences. Those Pro users matter a great deal to me at this stage, because what they tell us shapes what we build next. I will be open about one thing we learned from those numbers. Many of the early sign-ups came expecting a tool for creating decks, which is the one thing Decklit does not do. We have addressed that with a new design and a clearer narrative on the site, so people understand from the first screen that we take care of what happens after the deck is made. The free trial is the other part of the answer. We are about to release it, and it lets anyone use every feature for a limited time, which we have not offered until now. I expect it to bring the right people into the full product and to help with monetisation. We were also in Bangkok recently for the Telecoms World Asia conference. The feedback there was very encouraging, and several people offered to collaborate and to contribute their services at no cost. That kind of response from people who had just seen the product for the first time is something I am excited about. Q: What is a challenging time you had building Decklit? Simeone: Most of it, in fairness. I built Decklit alongside a full-time job and a master’s degree, so for more than a year every decision was made late at night or at the weekend, with my developers in Italy and me in Singapore. There is no team to hand things to when you are a solo founder. Product, pricing, support, testing and marketing all land on the same desk. The hardest part technically was the live side. Video in the browser behaves perfectly in a test and then surprises you the moment real people join from real networks and real devices. Someone joins late, someone shares a screen, someone switches microphone, and each of those can break something you thought was finished. There were weeks when one fix uncovered two new problems. What got me through was treating each failure as information. Every one of them showed me something about how people present in practice, and the product is more solid because of it. I also learned to ask for help. Some of the best progress came from reaching out directly to people who knew the technology better than I did. Q: What would you tell someone building a company alongside a full-time job? Simeone: That it is doable. It might take longer than you would like, but it can be done. I built Decklit in the evenings and at weekends, while working full time, finishing a master’s degree, raising a family and living abroad. None of those make it easier, and all of them together teach you to be very careful with your time. You have to accept your own pace and protect the few hours you have. I learned to decide quickly, to write things down clearly for my developers because I could not be in every conversation, and to ask for help from people who knew more than I did. It also gives you something. You are not spending someone else’s money, so you build only what matters. You keep one foot in the working world your customers live in, which keeps you close to the problem you are solving. And having a family beside you reminds you every day why you are doing it. Q: Who believed in you before there was anything to show? What did they say or do? Simeone: First and foremost, I did. It may sound like an odd answer, but when you start something on your own, with a full-time job, a degree to finish and nothing to show anyone yet, the first person who has to believe in it is you. If I had waited for someone else to be convinced before I was, Decklit would not exist. Right after that comes my wife. Her encouragement and her support, through all the evenings and weekends this has taken, are what made it possible to keep going. My family has been next to me the whole way as well. So the answer is the two of us, in that order. Without my own conviction and without her behind me, I do not think any of this would have happened. Q: How do you convince people to change a habit as old as sending a PDF? Simeone: By not asking them to change much. They keep making the deck exactly as they do today, in the tool they already know. The only difference is that they send a link in place of an attachment. Then the link does things a PDF never could. It tells you who opened it and which slides they read. It lets you present from it, live, without sharing your screen. It keeps the questions, the notes and the follow-up in one place, and it lets you take the deck back if you need to. A PDF, once sent, is gone. People do not give up a habit because you tell them to. They give it up when the new way is easier and shows them something they were missing. So I do not argue against the PDF. I ask people to send their next deck as a link, once, and look at what comes back. Once someone has seen who read their proposal and for how long, they do not go back to guessing. Both the free plan and free trial exist for that reason, so the first try costs nothing. Ultimately, persistence is key, and that is something I have plenty of. ### Part VII: What it is all for Q: What would make you say Decklit has succeeded? Simeone: When “send me the deck” and “can you see my screen?” start to sound old. I would call it a success when presenting from a link is simply how it is done, and people have forgotten it used to be harder. On a personal level, success is a presenter telling me they managed to engage their audience and win them over, thanks to the platform and to what they learned from running their sessions on Decklit. I want people to get better at presenting because of what the product shows them, session after session. The third measure comes from the other side of the screen. It is the day the audience starts asking for it. When a client or an investor says “can you send that as a Decklit link?”, because they prefer following a deck that way, the product no longer needs me to explain it. Q: What will presenting look like when AI can pitch on your behalf? Simeone: AI will take over the parts of presenting that are logistics. It will prepare the session, summarise it, follow up, and answer the routine questions at two in the morning. Some of that is already here, and Decklit does part of it today with summaries and transcripts. The next step is an agent that knows the deck and the history of every session. It can tell you which slide loses people, who to contact first, and what to change before the next meeting. Eventually it will walk someone through the deck when you are not there, and answer their questions from what you have said before. I think that makes the human part matter more. A deck can be delivered by an agent, but trust is still built by a person who stands behind what the deck says. The presenters who do well will be the ones who let AI handle the preparation and the follow-up, and spend their own time on the conversation. I want Decklit to take care of the first so people have more time for the second. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Simeone: I prefer not to put public numbers on users or revenue at this stage. What I can share is what the next twelve months are for. The first goal is to turn free users into paying ones. The free trial of the Pro plan is the main step, because for the first time anyone can use every feature before deciding. The second is to make the core dependable, with reliable recordings and smart transcripts that follow the deck slide by slide. The third is the Business plan for teams in 2027, with shared workspaces, roles and permissions, and custom branding and domains. If I had to name one measure, it would be the number of people who present on Decklit every week. Sign-ups are easy to count and tell you very little. Regular presenters tell you the product has become part of how someone works. Q: And beyond the first year, where do you want Decklit to go? Simeone: Over two years, the focus moves from individuals to teams, and from showing presenters what happened to helping them act on it. On the product side, I want presenting together to feel natural, so two or three people can run a session as one. I want the intelligence in Decklit to take real work off a presenter’s hands, before a session and after it. And I want Decklit to fit into the tools companies already rely on, so the deck and everything learned from it can move to where the work happens. On the business side, I want our first larger customers across Asia Pacific, with the security and integration features they expect, and a company that sustains itself on subscriptions. I would rather grow steadily with customers who stay than chase a headline number. Q: What are your hopes and fears for a post AGI world? Simeone: My hope is that it removes the busywork around communication, so people spend their time on the idea and on each other. If the preparation, the notes and the follow-up take care of themselves, a meeting can go back to being a conversation. I also hope it levels the field. A small company with a good idea should be able to present as well as a large one with a team of twenty behind every pitch. My fear is the opposite. Content made by machines, sent to machines that summarise it, with nobody listening at either end. We could end up producing more presentations than ever and communicating less. Presenting is one of the few moments where a person still has to show up and stand behind their words. I would like the tools we build, Decklit included, to protect that moment and make it better, and not replace it. About Decklit: Decklit is a presentation delivery platform founded by Luca Simeone in Singapore. Upload a PowerPoint or PDF once, share it as one secure link, present live from any browser without screen sharing, and get AI summaries and slide-by-slide analytics. decklit.com ## Dailies.games (USA) URL: https://startup.talk/usa/dailiesgames/ Founders: Justice Chiusano Justice Chiusano grew up in California before moving to Florida, with a twin, an older brother and an older sister, and a childhood spent at the beach or playing Halo 3 and Gears of War. A first job folding towels at a gym in Santa Monica taught reliability, and a career in game production followed: AAA quality assurance, art and animation pipelines and production coordination, including time at Treyarch on Call of Duty. Dailies.games is Chiusano's own project: a hub for daily browser games such as Connections, Wordle, Songless and GeoGuessr, with a four-game Today's Run and an optional community race that lets finishers vote to lift a streamer community up the daily creator rankings. It is funded and run independently. We spoke about why a directory can be copied, how to work fairly with independent developers, and what a shared daily routine can do for a community. Key takeaways: - Dailies.games gathers daily browser games in one place and adds a four-game Today's Run; completing it unlocks one vote a day for a creator community on the rankings board. - Chiusano says a directory can be copied, so the aim is trust with developers, a useful player experience and community habits people choose to keep. - Listing a game is free for independent developers, and the site links players to the developer's own website, so it does not imply it made the games or that its membership removes charges elsewhere. - Analytics recorded about 8,200 visits and 9,880 page views over the last 30 days (excluding bots), and the 12-month goal is 10,000 monthly users and about $500 a month in revenue. The Principal / Founder: - Justice Chiusano (Founder) Grew up: California and Florida Before: Game production: AAA QA, art and animation pipelines, Call of Duty at Treyarch Also built: NextTune, NextShow and NextGame browser extensions Funding: Independently funded Website: dailies.games LinkedIn: Justice Chiusano (https://www.linkedin.com/in/justicechiusano/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Chiusano: I grew up in California before moving to Florida when I was young. I have a twin, an older brother, and an older sister. A lot of my childhood was spent at the beach or playing games like Halo 3 and Gears of War. Gaming was a big part of growing up for me. Q: What first got you into building things? Chiusano: What interests me about building is turning an idea into something people can actually use. My background is in games, where a good idea still needs organization, testing, and a lot of iteration before it becomes a good experience. Dailies.games gave me a way to take that process and apply it to something of my own. Q: What was your first job, and what did it teach you? Chiusano: My first job was folding towels at a gym in Santa Monica. It was simple work, but my takeaway was that reliability matters. You show up, do the work properly, and make someone else’s day a little easier. That’s still a useful standard when you’re building a product. ### Part II: The road to Dailies.games Q: What made you start Dailies.games? Chiusano: The idea was to make an existing habit easier. There are so many good browser games, but finding them and keeping a routine can feel scattered. I wanted a place where someone could start with familiar games, discover something new, and have a reason to come back with their friends or community. Q: How did your work as a game development producer lead you to daily games? Chiusano: Production teaches you to look at the whole experience: what someone is trying to do, what gets in their way, and what needs to happen next. Daily games have a very clear rhythm. They’re approachable, repeatable, and easy to talk about. That made them an interesting foundation for a product built around routine and community. Q: What is your own daily game routine? Chiusano: Honestly, my friend group mostly just wants to play Connections! We do it together as a group guessing activity, talking through the words and debating which ones belong together. I’m still trying to convince them to branch out into other games. That shared experience is a big part of what I want to build with Dailies.games. Q: What did the first players who used Dailies.games teach you? Chiusano: One concrete lesson from the developer outreach has been how much presentation matters. A listing needs to explain the game clearly, show a useful screenshot, and make it easy to reach the original site. Developers also helped reinforce the distinction between a shared daily puzzle and an unlimited game. Both can belong in the directory, but players should know what they’re opening. ### Part III: The product Q: Why the name Dailies.games? Chiusano: I wanted the name to communicate the idea immediately. “Dailies” captures the habit of coming back for a small set of games, and the .games domain makes the purpose clear. It’s straightforward and leaves room for the product to grow. Q: Why bring daily games like Connections, Wordle, Songless and GeoGuessr into one run? Chiusano: They offer different kinds of play: words, connections, music, and geography. Together, they give the run variety and a clear finish line. The individual games remain on their own websites; Dailies provides the structure around discovering and returning to them. Q: How does the community race and creator ranking work, and why add it? Chiusano: Players complete the four-game run and unlock one creator vote for the day. They can use that vote to support a creator community on the rankings board. Q: How do you choose which games make the daily run? Chiusano: My approach is to keep the main run focused and easy to understand, with variety between games. The wider directory gives us room to include more niche experiences and independent submissions without making the core routine overwhelming. I also want the descriptions to be clear about whether a game is daily or unlimited. Q: Who gets the most value from Dailies.games today: casual players, streamers or game creators? Chiusano: Casual players get the most immediate value: somewhere to find games and organize a routine. Streamers and community organizers can give that routine a shared purpose through the creator race. Independent developers benefit from another place where people can discover their work. Q: What do you deliberately leave out of Dailies.games? Chiusano: I try to leave out unnecessary steps between discovering a game and playing it. I also want to be clear about what Dailies provides and what belongs to the original developer. A listing shouldn’t imply that we created the game or that our membership removes charges on someone else’s website. ### Part IV: Positioning Q: Who is Dailies.games for, and what job does it do for them? Describe the person who gets the most value. Chiusano: The person I picture is someone who enjoys a few browser games but doesn’t want to keep track of every link or search for something new each day. They might play during a break, share games in Discord, or follow a streamer. Dailies gives them a starting point: find something worth playing, keep their favorites nearby, and take part in a shared routine. Q: How do you view your positioning relative to NYT Games, Wordle clones and unlimited sites, Puzzmo and daily game aggregators? Is there another competitor you think of first? Chiusano: I see Dailies as a discovery and community layer around games from many different developers. Publishers such as NYT Games create their own games and experiences. Our role is to help players navigate a broader collection and give communities a shared activity. Q: What is your core moat in daily browser games? Chiusano: I wouldn’t claim an unbeatable moat at this stage. A directory can be copied. What I want to build over time is trust with developers, a useful experience for players, and community habits that people choose to maintain. Q: What are some of the use cases for Dailies.games that are not obvious and can add a lot of value to users? List as many as possible. Chiusano: A streamer could use the run as a warm-up before their main game. A Discord group could make it a daily check-in. Friends in different time zones could share the same activity without scheduling a session. It can also help people discover smaller developers, give community moderators an easy recurring event, and provide a short break for someone who doesn’t have time for a longer gaming session. Q: What is the one detail in Dailies.games you are proudest of? Chiusano: The detail I’d highlight is connecting completion of the run to a community vote. It gives finishing the games a small social consequence: your participation helps the community you support. ### Part V: Launch Q: What did you learn from your launch? Chiusano: A launch makes you explain the product much more clearly. You have to communicate who it’s for, what someone can do immediately, and why they should return. It also reminds you that launch attention and a lasting habit are different things. Getting someone to visit is only the beginning. ### Part VI: The hard part Q: How many people use Dailies.games today? Chiusano: Over the last 30 days, our analytics recorded approximately 8,200 visits and 9,880 page views, excluding bots. Those are visits and page views rather than unique-user counts, so I want to be precise about what we’re measuring. Q: What is a challenging time you had building Dailies.games? Chiusano: A recurring challenge has been making a straightforward idea work reliably in production. Adding a game sounds simple, but the description, screenshot, links, mobile layout, and deployment all need to work together. Running it myself means balancing new features with maintenance and developer communication. It has reinforced how much the small details matter. Q: Who believed in you before there was anything to show? What did they say or do? Chiusano: My close friends were excited about it from the start. They loved the name Dailies.games, and as I started building, they thought it was cool to see the progress from an idea into a working site. Having people close to me genuinely excited about what I was making meant a lot. Q: How do you work fairly with the independent developers whose games appear on Dailies.games? Chiusano: Anyone can submit a game for a free ordinary directory listing. We link players to the developer’s site, explain the game, and use screenshots to help people understand what they’re opening. Developers can also contact us to request corrections or a different image. Q: Can independent developers add their games to Dailies.games for free? (Common Question I get) Chiusano: Absolutely! Anyone can submit their game for a free directory listing, including games without a shared daily puzzle. I’d love to help more independent developers get their work discovered. Our monetization focuses on optional PRO features, so developers don’t need to pay for an ordinary listing. ### Part VII: What it is all for Q: What would make you say Dailies.games has succeeded? Chiusano: I’d call it a success when people return because it has become a useful part of their day, developers feel their games are represented well, and communities use it without needing constant reminders from me. Financially, I want it to support its own operation and continued improvement without making the free experience frustrating. Q: What will casual gaming look like when every community has its own daily ritual? Chiusano: I think more communities could have small activities that fit around people’s lives. You wouldn’t need everyone online at once or committed to a long session. People could take part when they have time and still have something in common to discuss. The appeal is belonging through a manageable routine. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Chiusano: Over the next 12 months, I’d love to reach 10,000 monthly users and around $500 a month in revenue. Those would be meaningful milestones for a project I’m running myself without outside funding. I want to build a regular audience and show that Dailies.games can support its own growth while keeping ordinary game listings free. Q: Where do you want Dailies.games to be in five years? Chiusano: In five years, I’d love to see the community features become a big part of Dailies.games, with streamers and their audiences making the daily run a shared routine. Having bigger creators get involved could bring more people to the site and help them discover independent games they might otherwise miss. Q: What are your hopes and fears for a post AGI world? Chiusano: AI tools are everywhere now, and I have mixed feelings about that. But they’re amazing for helping people build things without formal technical knowledge. I’m not a coder; I’m a gamer who made a site about games. My hope is that more people get the chance to bring their ideas to life. My concern is that we start treating AI as a replacement for human creativity. I still believe creative people will outmatch AI when it comes to art. The tools can help you make something, but your taste, experiences, and point of view are what give it meaning. About Dailies.games: Dailies.games is a hub for daily browser games and the communities that play them, built and run independently by Justice Chiusano. Players can keep favourites such as Connections, Wordle, Songless and GeoGuessr in one place, finish a four-game Today's Run and take part in a community race on the creator rankings. Listings for independent developers are free. dailies.games ## Zyaku (USA) URL: https://startup.talk/usa/zyaku/ Founders: Rohit Kashyap Rohit Kashyap was born in Mumbai, with family roots in a small city in Bihar where nobody took power or a phone signal for granted. Kashyap's career began in finance, first as a financial consultant at BNP Paribas, then in investment and wealth management and at Kashyap's own firm, until the frustration of needing tools that did not exist made learning to build feel better than waiting. Zyaku is the result: an offline-first publishing platform for creators and independent journalists, for comics, journalism, podcasts and short video. Posts written without a connection stay safely on the phone and publish exactly once when the signal returns, creators can earn with no follower minimum, and the same app carries safety tools for the worst day. We spoke about a finance rule that became a product rule, the floods in Nepal, an Emergent Ventures grant for the open-source HoldSend library, and why the detail Kashyap is proudest of is one small line under the editor. Key takeaways: - Zyaku keeps posts written offline safely on the device and publishes each one exactly once when the connection returns, and shows where every post is: saved, waiting for signal, being sent or published. - Creators can earn with no follower minimum. Safety tools sit in the same app: Crisis alerts from official warning services, an offline emergency assistant, a Medical ID with a QR code, I'm Safe check-ins and an offline map of hospitals and shelters. Zyaku is not an emergency service. - The hard part of offline publishing is open-sourced as HoldSend (version 0.2 on npm, not production-ready yet), supported by an Emergent Ventures grant. - The targets: 25,000 users and $1 million in revenue within 12 months, and around 100,000 users and $5 million within 24. The Principal / Founder: - Rohit Kashyap (Founder and CEO) Company: Zyaku Inc., San Francisco Born: Mumbai, with family roots in Bihar Before: Financial consultant at BNP Paribas; investment and wealth management; founded own finance firm Also: Founder of TAMVEL (early-stage research, prototype) Funding: Emergent Ventures grant (Mercatus Center at George Mason University) Open source: HoldSend, Apache-2.0 Website: zyaku.com GitHub: zyakuhq/holdsend (https://github.com/zyakuhq/holdsend) LinkedIn: Rohit Kashyap (https://www.linkedin.com/in/rohitkashyap-zyaku/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Kashyap: I was born in Mumbai, but my family comes from a small city in Bihar. So I grew up with a foot in two very different places. Mumbai, where everything moves fast and the whole city runs on being connected. And our small city in Bihar, where life was slower, everyone knew everyone, and you learned not to take things like power or a phone signal for granted. I didn't know it then, but that gap stayed with me. A lot of what I build today is for the second place, not the first. Q: What first got you into building things? Kashyap: Honestly, frustration. I spent most of my working life in finance, first investment and wealth management, then my own firm. I wasn't an engineer. But I kept running into the same gap: the thing I wanted didn't exist. At some point, waiting for someone else to build it felt worse than learning. So I taught myself. Q: What was your first job, and what did it teach you? Kashyap: My first job was as a financial consultant at BNP Paribas. Finance taught me one rule I still build by: a transaction either happened or it didn't. 'Probably sent' is not a state. You don't get to tell someone their money moved when you're only hoping it did. Years later, when I started on the offline side of Zyaku, that was the first rule I wrote down. Nothing gets called sent until it has actually arrived. ### Part II: The road to Zyaku Q: What made you start Zyaku? Kashyap: One picture I couldn't shake. A phone on a kitchen table, one bar of signal that keeps disappearing, water already in the street, and someone trying to tell their mother they're fine. The message just sits there with a little clock icon, and nobody knows if it went. Most of the internet was built for good Wi-Fi and a calm afternoon. When the network goes, apps don't slow down. They just stop. Your words are still on the phone, but the software decides that without a connection, nothing you do counts. That bothered me enough to start building. Q: What did you see that convinced you creators and journalists need tools that keep working when networks fail? Kashyap: The floods in Nepal. Networks failed exactly when people most needed to reach each other. But it's not only disasters. Power cuts, a train through a dead zone, a village with one weak tower. Reporters file from those places. Creators live in those places. If your tool loses your work, or tells you something was sent when it wasn't, you stop trusting it, and usually on the day it matters most. Q: How do Zyaku and TAMVEL connect, and why build both? Kashyap: Same stubborn idea, two different problems. Zyaku is about words. A post, a story or a help message should go out exactly once when the signal comes back, and the app should never pretend it arrived when it didn't. TAMVEL is about money. Financial apps shouldn't stop working the moment the network does. That one comes straight from my finance years, where a transaction either happened or it didn't. 'Probably sent' is not a state. TAMVEL is still early. It's at the research stage with a prototype, so I talk about it carefully. Q: What did becoming an Emergent Ventures grantee change for you? Kashyap: It changed how I think about the open-source side of the work. The grant supports HoldSend, the library behind Zyaku's offline publishing, so the hard part, writing offline and publishing exactly once, could be built in the open for everyone and not only for us. It also meant something simpler. Emergent Ventures backs people with unusual ideas, often before there's much to show. Being backed that early told me this idea was worth the stubbornness. It made me more careful too. When someone trusts you that early, you owe them honesty about what's done and what isn't. ### Part III: The product Q: Why the name Zyaku? Kashyap: I found the name on a trip to Tokyo, from Japanese people I met there. "Zyaku" is an old way of writing the Japanese sound jaku, and it can mean two things that fit us almost too well: 寂, quiet, and 弱, weak. The quiet hour when the internet goes silent, and the weak signal people are left with. Zyaku is built for both. Q: How does a post queue offline and still publish exactly once when the connection returns? Kashyap: When you write without a connection, Zyaku keeps the post safely on your phone and gives it its own key. When the signal comes back, it sends it. Phones lose signal halfway through a send. People tap the button again. Apps get closed in the background. In every one of those cases the server sees the same key, knows it's the same post, and publishes it once. Not zero times, not twice. You can also see where each post is: saved on this device, waiting for signal, being sent, or published. If a post is waiting, Zyaku now tells you why. The rule underneath all of it is simple: nothing gets called sent until it has actually arrived. Q: Why build Disaster Alert broadcasts and Offline Emergency Access into a creator platform? Kashyap: Because the people who create on Zyaku are the same people who end up in the flood. Once a platform works offline for creators, it can work offline for families too. So Zyaku turns serious official warnings into Crisis alerts, and it has an offline emergency assistant, a Medical ID with a QR code a rescuer can read with any phone camera, I'm Safe check-ins and an offline map of hospitals and shelters. You can even send yourself a test alert on a calm day and Zyaku tells you whether it arrived. I'm careful here, because safety is the one place where overpromising is unforgivable. Zyaku is not an emergency service. If you're in danger, call your local emergency number first. Q: Why pay creators with no follower minimum, and how does that work as a business? Kashyap: I don't share the exact split in interviews. It's on our pricing page, and creators keep most of what they earn. The part that matters more to me is no follower minimum. Readers pay for the chapter they unlock, not for how famous you are. A creator in a small town with a few loyal readers should be able to earn. And they should get paid where they live, including places where the usual payment rails don't reach easily. Q: Why are you open-sourcing the core of Zyaku? Kashyap: Because this part shouldn't belong only to us. Newsrooms need it. Aid groups need it. Small apps in places with bad signal need it. Every team that builds it alone makes the same mistakes I made: losing posts, duplicating posts, and telling people 'sent' when it wasn't true. So we released it as HoldSend, an open-source library for writing offline and publishing exactly once. Version 0.2 is on npm. It's early and not ready for production yet, and the README says that plainly. Anyone can use it, read it and improve it. Q: What do you deliberately leave out of Zyaku? Kashyap: We don't optimise for time spent, and we don't count it as a win. Mood check-ins and journal entries are not advertising inputs. We don't sell how you feel. Where we don't know a number, the dashboard shows a dash instead of an estimate dressed up as a fact. And we don't call things finished before they are. Our mesh, where phones pass messages to each other with no network, is still in testing. I won't say otherwise until it works in the field, with real people at real distances. ### Part IV: Positioning Q: Who is Zyaku for, and what job does it do for them? Describe the person who gets the most value. Kashyap: A journalist who has to file from a place where the network keeps dropping. A creator in a small town who wants to get paid without first building a huge audience. A family that wants to know everyone is safe before the towers come back. The person who gets the most out of Zyaku is someone whose connection isn't reliable, and who has quietly learned that most apps let them down. Q: How do you view Zyaku's positioning relative to other publishing and creator platforms, and to offline messaging apps? Kashyap: I don't really see Zyaku as an alternative to any one of them, and I'd rather not compare by name. Most publishing and creator platforms are built on one quiet assumption: you're online. When the signal goes, they stop, and your work waits or disappears. The apps that do work without a network are mostly about messaging, not about publishing a story, building an audience or getting paid. Zyaku sits where those two meet. It's a place to publish comics, journalism, podcasts and video and to get paid with no follower minimum, and it's built from the ground up to keep working when the network doesn't, with safety tools in the same app. The competitor I think of first isn't a company. It's the app on your phone that shows a little clock icon when the signal drops, and never tells you the truth about whether your words went. Q: What is your core moat in offline-first creator publishing? Kashyap: Being honest about state. Plenty of apps can save something offline. Very few will tell you the truth about whether it actually went, and then publish it exactly once when it does. We built every part of Zyaku around that rule, from posts to emergency messages to test alerts. And we build in the open, with a public changelog and an open-source core, so people can check our work. Q: What are some of the use cases for Zyaku that are not obvious and can add a lot of value to users? List as many as possible. Kashyap: - A reporter in a shutdown or blackout writes the story, and it goes out once when the signal returns. - Writing on a flight or a train and publishing when you land. - Offline Packs: download comic chapters or videos once on Wi-Fi, then read or watch anywhere. - A Medical ID QR code a rescuer can read with any phone camera, no app and no internet. - Sending yourself a test disaster alert before the season starts, to check it reaches your phone. - Saving the map of nearby hospitals and shelters before you travel. - Publishing anonymously as a journalist: your identity is stored encrypted and is never shown to readers. - Telling the people you trust that you're okay with one tap (I'm Safe). - Reading posts in your own language, with a clear label when AI translated them and the original one tap away. Q: What is the one detail in Zyaku you are proudest of? Kashyap: A tiny line under the editor. It tells you where your draft is: saved on this device, saved to Zyaku, waiting for signal, being sent, or published. It's small, but it's the whole idea of Zyaku in one line. ### Part V: Launch Q: What did you learn from your launch? Kashyap: The biggest lesson was about how we describe ourselves. For a long time we called Zyaku a social media app. Launching made it clear that was the wrong box. Put Zyaku next to social apps and people look for a feed and a follower count, and the things that make it different get lost. Zyaku is a different category. It's publishing that keeps working when the internet doesn't: write offline, publish exactly once when the signal returns, get paid without needing a big audience, and have safety tools on the same phone on the worst day. ### Part VI: The hard part Q: What is a challenging time you had building Zyaku? Kashyap: The hardest moments were the quiet failures. A while ago I noticed alerts weren't showing up on my own devices when Zyaku was closed. On our side everything looked fine. Every alert went out. But we weren't reading the answer that said whether a phone actually took it. Nothing broke, no error, and everyone believed it worked. That was uncomfortable to find. We fixed it from both ends: we now read that answer, and we gave people a button to send themselves a test and see it for themselves. Q: Who believed in you before there was anything to show? What did they say or do? Kashyap: My family, long before there was a product: my wife, my mother, my father and my sister. None of them needed a demo or a pitch. They believed in it because they knew me. When I left a steady career in finance to build something most people didn't understand yet, they stood behind me. That support carried me through the months when there was nothing to show anyone. Q: How does a small team build infrastructure this broad? Kashyap: We're a small team, and we don't try to do everything at once. We ship small changes often, write every change on a public changelog, and say no to a lot. Building in public helps too. When I posted a demo of offline publishing, engineers I'd never met started asking hard questions, and those questions made the work better. ### Part VII: What it is all for Q: What would make you say Zyaku has succeeded? Kashyap: When a journalist in a shutdown files a story and it arrives once. When a creator in a small town gets paid without needing a big audience. When a family knows everyone is safe before the towers come back. If those things happen on ordinary bad days, quietly, without anyone having to think about it, I'll say it worked. Q: What will the internet look like for creators when AI agents and intermittent networks are both normal? Kashyap: I think a weak or missing connection will simply be normal for a lot of people, and the software that earns trust will be the software that's honest about it. AI will write a lot of what we read, so knowing where something came from will matter more. That's why Zyaku labels AI translations and keeps the original one tap away, and why a story only shows as independently checked when a reviewer, never the author, has checked it against a public source. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Kashyap: In the next 12 months we want to reach 25,000 users and $1 million in revenue. Q: And for the next 24 months? Kashyap: In 24 months we want to reach around 100,000 users and around $5 million in revenue, with HoldSend used by teams outside Zyaku. Q: Where do you want Zyaku to be in five years? Kashyap: I want Zyaku to be the place people trust on the worst day, not only on the best one. And I'd love the open-source core to be used by teams we've never met, in newsrooms and aid groups, so this problem gets solved once for everyone. Q: What are your hopes and fears for a post-AGI world? Kashyap: My hope is that it frees people to do more of what only humans can do. My fear is that we start valuing people less because of it. That's my personal view, and I'll keep saying it: AI should never become a reason to treat a human being as worth less. Humans built it. The people behind it deserve their full due: fair pay, real credit and respect. About Zyaku: Zyaku is an offline-first publishing platform for creators and independent journalists, from Zyaku Inc. in San Francisco, founded by Rohit Kashyap. Comics, motion comics, journalism, podcasts and short video written without a connection stay safely on the device and publish exactly once when the signal returns. Creators can earn with no follower minimum. Its safety tools include Crisis alerts from official warning services, an offline emergency assistant, a Medical ID with a QR code, I'm Safe check-ins and an offline map of hospitals and shelters; Zyaku is not an emergency service. Its offline publishing core is open source as HoldSend. zyaku.com ## Rewisp (USA) URL: https://startup.talk/usa/rewisp/ Founders: Yashmit Bhaverisetti Yashmit Bhaverisetti kept making small promises on screen, the “I’ll send it Friday” kind, and then forgetting them. The promise was right there in a chat or an email, and then it was gone. Rewisp is the fix: a free, open-source macOS menu bar app that reads the text on your screen with on-device OCR, keeps it in a local database, and answers questions about what you saw. Bhaverisetti builds it solo. Rewisp catches the promises you make, shows what changed on a page you saw before, and lets Claude Desktop and Cursor look things up in your screen memory through read-only MCP tools. It launched in July 2026 at version 0.9 and now has more than 400 users. We spoke about why text beats screenshots, admitting an unencrypted database in public, the strangers who helped design the fix, and everything Rewisp deliberately leaves out. Key takeaways: - Rewisp is a free, open-source macOS menu bar app that reads on-screen text with on-device OCR, keeps it in a local database and answers questions about what you saw. Screenshots are never saved. - Its promise catcher spots lines like “I’ll send it Friday” and turns them into reminders, and it can show how a page changed since you last saw it. - Claude Desktop and Cursor can query it through read-only MCP tools: they can look at your memory but never change or delete it. - It launched in July 2026 and has more than 400 users. Privacy questions dominated the launch, and admitting the database was not yet encrypted earned more respect than any feature. The Principal / Builder: - Yashmit Bhaverisetti (Builder) Built: Solo Launched: July 2026 Platform: macOS 15+, Apple Silicon Pricing: Free and open source Users: 400+ Website: Rewisp (https://yashmitb.github.io/Rewisp/) GitHub: yashmitb/Rewisp (https://github.com/yashmitb/Rewisp) LinkedIn: Yashmit Bhaverisetti (https://www.linkedin.com/in/yashmitb/) ### Part I: The product Q: Why the name Rewisp? Bhaverisetti: A wisp is a small, faint trace of something. That’s what most of what we read is. We see it for a second and it fades. Rewisp brings that trace back. Q: Why capture on-screen text instead of screenshots? Bhaverisetti: Text is what people actually want back. You remember a number, a name, or a line. You don’t remember pixels. Text is small, so I can keep months of it on a normal Mac. It’s also easy to search and easy for an AI to read. Screenshots eat disk space and are harder to search. Q: Why catch promises like “I will send it Friday”? Bhaverisetti: I kept saying things like that and then forgetting. The promise was right there on my screen, in a chat or an email. Then it was gone. Rewisp spots those lines and lists them for you. It’s a to-do list that writes itself from what you already said. Q: Why diff a page to show what changed? Bhaverisetti: Pages change quietly. A price goes up. A due date moves. A doc gets edited. You’d never notice. Since Rewisp already saw the old version, it can show you exactly what’s different. You don’t have to remember what it said before. Q: Why expose read-only MCP tools to other AI apps? Bhaverisetti: People already live in Claude Desktop and Cursor. I didn’t want to build another chat box. With MCP, those apps can ask Rewisp things like “what was I looking at this morning?” It’s read-only on purpose. Other apps can look at your memory. They can never change or delete it. Q: Why make it free and open source? Bhaverisetti: A tool that sees your whole screen needs trust. The best way to earn that is to let anyone read the code. You don’t have to take my word that nothing leaves your Mac. You can check. Free also means more people try it, and I learn faster from them. Q: What do you deliberately leave out of Rewisp? Bhaverisetti: No cloud sync. No accounts. No audio recording. No video timeline. No tracking. Each one would make the app bigger and less private. I’d rather do one thing well: remember the text you saw, and let you ask about it. ### Part II: Positioning Q: Who is Rewisp for, and what job does it do for them? Describe the person who gets the most value. Bhaverisetti: People who read a lot on their Mac and lose things. Students, engineers, researchers, anyone juggling many tabs. The person who gets the most out of it reads all day across email, docs, Slack, and the web. They often think “I saw that somewhere.” Rewisp’s job is to answer “where?” Q: How do you view your positioning relative to Rewind-style memory apps and Reflect? Is there another competitor you think of first? Bhaverisetti: Rewind proved people want this. But it moved to the cloud, then to a pendant. Then Meta bought the company and shut the Mac app down. Lots of users lost a tool they relied on. That’s the risk with a closed app. Rewisp is local and open source, so it can’t be taken away like that. Reflect is a great notes app, but you have to write the notes. Rewisp works with zero effort. The one I think about most is Screenpipe. It’s also open source. It does more, like audio and other platforms. Rewisp is smaller and Mac-only, and it focuses on recall, promises, and changes. Q: What is your core moat when memory features keep landing inside the operating system? Bhaverisetti: Honestly, the moat is trust and focus. An OS feature has to work for everyone, so it stays general. It also ties you to one company’s AI. Rewisp works with whatever AI you like through MCP. And it’s open source, so you can see what it does and change it. I’d also rather move fast as one person than wait for a yearly OS update. Q: What are some of the use cases for Rewisp that are not obvious and can add a lot of value to users? List as many as possible. Bhaverisetti: Finding a number you saw once: a price, an order number, a confirmation code Getting back a tab or article you closed weeks ago Remembering what someone said in a chat when you can’t recall which app Writing a daily standup from what you actually worked on Tracking promises you made to friends, teammates, or professors Seeing how a job post, price, or policy page changed Letting Cursor know which docs and errors you looked at while coding Rebuilding your notes after a lecture or meeting you only half followed Answering “when did I first see this?” for research Catching a deadline that was only mentioned in passing Q: What is the one detail in Rewisp you are proudest of? Bhaverisetti: The promise catcher. It’s a small feature, but it’s the one that makes people say “oh, that’s useful.” It turns something you said without thinking into a reminder you didn’t have to set. ### Part III: Launch Q: What did you learn from your launch? Bhaverisetti: I learned that people care about privacy first. Almost every serious question was about it. Is the data local? Is the file encrypted? What happens if my Mac gets stolen? Features got people interested. Privacy decided if they would trust it. I also learned to be honest about gaps. Someone asked if the database was encrypted. At the time it wasn’t. I said so plainly. That reply got more respect than any feature. One person even thanked me for admitting it. Then they helped me think through how to fix it. And I learned that a launch is a free design review. People asked smart questions about search speed, false reminders, and memory decay. Each one made me explain my choices. Some made me change them. I ended up #5 of the day on July 19, 2026, with 167 upvotes. For a solo project at v0.9, I was really happy with that. Q: Which feedback from the launch surprised you? Bhaverisetti: How deep it got. I expected “cool app” comments. Instead people asked about SQLite, FTS5, and key storage. One commenter and I went back and forth on how to encrypt the database without breaking search. They even suggested a design where the capture process can write but never read back. I didn’t expect that level of help from strangers. I was also surprised by what people liked most. I thought search would be the hook. But the promise catcher came up again and again. People said small promises like “I’ll send it Friday” matter more than the tasks they write down. That changed how I think about what Rewisp is for. ### Part IV: The hard part Q: How many people use Rewisp today? Bhaverisetti: 400+ Q: What is a challenging time you had building Rewisp? Bhaverisetti: The hardest part was encryption. At launch, the database was a plain SQLite file. FileVault protected it, but any process running as you could read it. People on Product Hunt asked about it right away. I had to admit the gap in public. Q: Who believed in you before there was anything to show? What did they say or do? Bhaverisetti: Myself Q: How do you keep a local memory of everything on screen safe? Bhaverisetti: I start by saving less. You can’t leak what you never captured. Screenshots are never saved. Rewisp pulls out the text and throws the image away right away. Messages, password managers, banking sites, and incognito windows pause capture fully. Text that looks like a password or credential is refused. You can add any app or site to the block list. You can pause with a shortcut, or “nuke” the last 10 minutes. Old text expires after about six months. About Rewisp: Rewisp is a free, open-source macOS menu bar app built by Yashmit Bhaverisetti. It reads the text on your screen with on-device OCR, keeps it in a local database, answers questions about what you saw, catches promises you made, shows how pages changed, and exposes read-only MCP tools to Claude Desktop and Cursor. Screenshots are never saved and nothing leaves your Mac. yashmitb.github.io/Rewisp ## PickMyVibe (India) URL: https://startup.talk/india/pickmyvibe/ Founders: Dhruv Pokhriyal Dhruv Pokhriyal grew up in Delhi with a curiosity about machines that Tony Stark only made stronger. Computer science as an elective in 11th standard started a love of coding, which had to wait through board exams and JEE preparation. After cracking JEE Advanced, Pokhriyal started building in earnest and is now studying for a CS degree at IIT. PickMyVibe is the second product, after StudySnapAI. Two years of full-stack development in college kept running into the same snag: hours lost choosing colours and fonts that matched the feeling of an app. PickMyVibe turns a mood such as “rainy neon Tokyo night” into five colours with fixed roles, two Google Font pairings and three CSS gradients, measures the contrast itself, and hands it all over as code. We spoke about not trusting the model’s contrast numbers, a launch that taught distribution the hard way, and why the apps that feel different will be the ones a person actually edited. Key takeaways: - PickMyVibe turns a mood or phrase into a small theme: five colours with named roles, two Google Font pairings with a reason, and three CSS gradients. - It throws away the model’s contrast figures and measures readability itself with chroma.js, labelling each pair AAA, AA, AA Large or Fail. - Results copy straight into code as CSS variables, a Tailwind theme snippet or a Google Fonts import. There is no login and no Figma plugin. - It is free and bootstrapped. The 12-month targets: about 1,000 people generating a vibe and a simple paid plan bringing in about $500 a month. The Principal / Founder: - Dhruv Pokhriyal (Founder) Grew up: Delhi, India Studying: Computer science, IIT Before: Built StudySnapAI Funding: Bootstrapped Website: PickMyVibe (https://pick-my-vibe-tags.vercel.app/) LinkedIn: Dhruv Pokhriyal (https://www.linkedin.com/in/dhruvpokhriyal) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Pokhriyal: I grew up in Delhi, like usual I used to go to school and play with my friends, but I always had some form of love and curiosity toward machines, then later I was also inspired by Tony Stark which further increased my interest in machines. But for the most part my childhood was spending time with family and friends. Q: What first got you into building things? Pokhriyal: I took CS as an elective in 11th standard and fell in love with coding but then didn't do much in my 11th and 12th grade, since I had boards and my JEE preparation, but once I cracked JEE Advanced I started doing coding more and building more things, and that increased more in my IIT journey. Q: What was your first job, and what did it teach you? Pokhriyal: I haven't had any job yet, I am still pursuing my CS degree from IIT. But I did one more product before PickMyVibe, that was StudySnapAI and it taught me a lot, about how to work in sprints, ideation, design decisions, pricing decisions and more. ### Part II: The road to PickMyVibe Q: What made you start PickMyVibe? Pokhriyal: What made me start PickMyVibe was since I've been doing Full Stack Development for the last 2 years in my college degree, I noticed a lot of time we get stuck on decisions like what color would match the vibe or finding a right color palette, cause a right and wrong color palette makes a lot of difference, that led me to PickMyVibe, since I wanted to simplify the things and better UI. Q: What frustrated you about choosing colours and fonts for your own projects that convinced you to build PickMyVibe? Pokhriyal: The frustrating part in choosing colors and fonts for projects was, often times most of the fonts and colors were generic and were not specific to the application and we had to spend hours finding the right fit that match the vibe we want. Q: Why did you decide to start from a mood or vibe instead of from design tools? Pokhriyal: I started from a mood or vibe instead of design tools is because as humans we tend to feel mood or vibe of something, so since other applications that I or any other user of PickMyVibe will be creating will be used by users, so it should convey the meaning they want to give and it sits right with color theory as well that different color gives different vibes. Q: What did you learn from the first developers and creators who used PickMyVibe? Pokhriyal: As the developer of PickMyVibe, what I learned during my journey of making PickMyVibe, was iterative design and rapid prototyping since I believe that is a very useful skill in today's era. Another thing I learned was to ship an SLC, while an MVP is good, SLC is what makes user want to come back to your website. ### Part III: The product Q: Why the name PickMyVibe? Pokhriyal: The name PickMyVibe represents, how I want the product to function, where the user can specify the vibe they want or the thing they want to convey and then the product generates the color palette and fonts which help them convey the vibe. Q: How does PickMyVibe turn a phrase like “rainy neon Tokyo night” into palettes, typography and gradients? Pokhriyal: You type a mood, or pick one like Neon Noir or Cottagecore. That phrase goes to a design-system prompt that asks the model for one structured result: a title and mood words, five colours with fixed jobs (Primary, Accent, Surface, Text, Highlight), two Google Font pairings with a short reason they belong together, and three CSS gradients. The model is told to treat the phrase as one atmosphere, so the browns of a café or the cyan and magenta of a rainy Tokyo night show up as a set, not as five unrelated swatches. The app then draws that JSON as palette cards, type pairings, and gradient strips you can copy. Q: How do you make sure the colours and fonts it suggests work well together and stay readable? Pokhriyal: Colour roles do the structural work. Text, Surface, and Primary are named jobs, so the palette is a small interface system rather than a bag of hex codes. After generation, the server ignores the model’s contrast numbers and recalculates three pairs with chroma.js: Text on Surface, Text on Primary, and Highlight on Surface. Each pair is labelled AAA (7:1 or higher), AA (4.5:1 or higher), AA Large (3:1 or higher), or Fail, and that score is shown on the result. Fonts are chosen as Google Font display and body pairs, each with a written rationale for why they suit the mood. The score is a check you can see before you ship. A failing pair is reported, and the palette is not silently rewritten. Q: How do people take a PickMyVibe result into their own code or design tools? Pokhriyal: Everything useful is a copy. A single hex, the whole palette, or a gradient’s linear-gradient(...). Each type pairing copies a Google Fonts @import plus font-family rules for headings and body. The export panel turns the same result into CSS custom properties (--primary, --surface, --gradient-1), a Tailwind theme.extend.colors snippet, or a plain role-and-hex list. You paste that into a stylesheet or Tailwind config and keep building. There is no Figma plugin or file download. The handoff is clipboard to code. Q: Who gets the most value from PickMyVibe today: indie hackers, developers or content creators? Pokhriyal: Indie hackers and developers. It was built for people who already know the feeling they want (“calm”, “cyberpunk”, “late-night café”) and get stuck turning that into colours, type, and gradients, especially without a designer on the project. The output is aimed at that moment: hex values, a closest Tailwind name, CSS variables, and a theme snippet you can drop into an app the same day. Content creators can use the palette and mood words as a direction, but the result is a theme for code, not a post template or a brand deck. Q: What do you deliberately leave out of PickMyVibe? Pokhriyal: A full design system. There is no spacing scale, radius, shadow, or component library, and no generated screens, logos, or illustrations. Five colours, two font pairs, and three gradients are a visual direction, not a UI kit. It also does not log you in, save a library of vibes, or sync to Figma. Readability is measured and shown. PickMyVibe does not auto-fix a palette that fails contrast, because the point is a fast, honest starting point you edit in your own project, not a finished brand. ### Part IV: Positioning Q: Who is PickMyVibe for, and what job does it do for them? Describe the person who gets the most value. Pokhriyal: PickMyVibe is for an indie hacker or developer who can ship the product and already knows the feeling they want, but gets stuck turning “rainy neon Tokyo night” or “calm Scandinavian app” into colours, type, and gradients. Their job is the first visual direction: five colours with roles, two font pairs, three CSS gradients, a real contrast score, and something they can paste into CSS or Tailwind the same day. The person who gets the most value is building alone, or without a designer, and needs the interface to feel intentional before they write the rest of the UI. Q: How do you view your positioning relative to Coolors, Khroma, Adobe Color, Fontjoy and AI design tools like Uizard and v0? Is there another competitor you think of first? Pokhriyal: Coolors, Khroma, and Adobe Color are colour tools. Fontjoy is a font-pairing tool. Uizard and v0 jump ahead and generate screens or components. PickMyVibe sits in the gap: one mood becomes a small theme, with colour roles, type, gradients, a measured contrast score, and a copy-paste handoff into code. It does not replace a colour wheel, and it does not lay out your app. The competitor I think of first is that gap itself. People either collect swatches, or they ask an AI to draw the whole interface. PickMyVibe is the step where the mood becomes a theme you can actually build from. Q: What is your core moat in AI-assisted design direction? Pokhriyal: The moat is the shape of the result, not a private model. The prompt locks the output to one atmosphere: five named colour jobs, two type pairs with a reason, and three gradients. The server then throws away the model’s contrast numbers and recomputes Text on Surface, Text on Primary, and Highlight on Surface with chroma.js, so the readability score is measured. The export is already code: CSS variables, a Tailwind colour extension, hex, and a Google Fonts import. Colour tools stop at swatches. UI generators skip this layer and draw screens. PickMyVibe owns the translation from a sentence to a theme you can ship. Q: What are some of the use cases for PickMyVibe that are not obvious and can add a lot of value to users? List as many as possible. Pokhriyal: Spinning up a Tailwind theme for a hackathon or weekend app before any mock exists. Putting two moods in front of a co-founder (“warm café” versus “brutalist”) so the argument is about a palette, not an adjective. Grabbing a hero gradient and a surface colour for a landing page the day you name the product. Checking that a dark, moody palette can still hold body text, because Fail stays Fail. Pairing fonts for a blog, docs site, or portfolio and copying the Google Fonts import straight in. Giving a designer or a developer the same five hex values so the first commit is not default slate. Re-skinning a side project when you are tired of the current look, without redrawing every screen. Setting one direction for a Product Hunt page, a pitch deck, and the app, from the same phrase. Teaching yourself WCAG by reading AAA, AA, and Fail on a palette you just described. Theming a game jam HUD, an event page, or a music project from an atmosphere instead of a logo. Writing the intended mood into a pull request or a design note, with colours attached. Trying “trustworthy fintech” and “playful fintech” as two generations, then keeping the one whose text-on-surface score you can live with. Matching a newsletter, slide header, or social banner to the same Primary, Surface, and Accent as the product. Starting a client conversation from their words (“late-night studio”, “desert minimal”) and showing a direction in the meeting. Building the personal site you have been postponing because you could not choose a type pair. Q: What is the one detail in PickMyVibe you are proudest of? Pokhriyal: The interface itself. The line “Turn vibes into beautiful design systems.” focuses one word at a time: that word stays sharp, the others blur, and a frame moves onto the active word. A soft blob trails the cursor. The menu opens as staggered colour layers, not a plain list. An aurora sits behind the whole page. I wanted the product to feel like a vibe before anyone typed one. ### Part V: Launch Q: What did you learn from your launch? Pokhriyal: Shipping the listing was the easy part. Product Hunt did not feature PickMyVibe, so the page never became a real conversation: a handful of upvotes, and the only comment was my own. I learned that a launch is a distribution problem. The maker note explained why I built it, and that still left people with nothing to react to. The useful responses came from sharing it myself. People answered the feeling of the product and the problem, which told me the next launch needs a sharper ask, a live example of a phrase turning into a theme, and an audience I bring, rather than a hope that the homepage will do it. Q: Which feedback from the launch surprised you? Pokhriyal: I asked for brutal honesty and barely got any. I expected people to pick apart the colours, the fonts, or the things the product does not do yet. The comments that did show up were short. On LinkedIn, one person said it looks lovely, and another said this is something designing needs. Nobody sent a feature list. The surprise was how little criticism there was, and that the first reaction was the look of the product and the problem it is trying to solve. ### Part VI: The hard part Q: How many people use PickMyVibe today? Pokhriyal: I do not have a public user count. PickMyVibe is a free web app, and the Product Hunt launch was small, with the conversation happening more on LinkedIn than on the listing. Usage is tracked in PostHog per generation, copy, and export, so I can see activity, but I am not going to put a made-up monthly number on this form. Q: What is a challenging time you had building PickMyVibe? Pokhriyal: The hard part was trust. The model will happily return a contrast ratio, and that number can be wrong while looking precise. I stopped using it. After the palette comes back, the server throws those ratios away and measures three pairs with chroma.js: Text on Surface, Text on Primary, and Highlight on Surface. Building the page had its own snag. The aurora background is WebGL, so it cannot render on the server. It has to load only in the browser, or the app breaks on first paint. Q: Who believed in you before there was anything to show? What did they say or do? Pokhriyal: My family believed in me, they supported me that in case I need any help they are ready to back me up. My dad funded the GPT tokens and also checked the design of my website once I completed it. Q: How do you keep AI-generated design systems original rather than generic? Pokhriyal: A result stays tied to the phrase because the prompt has to return one set: five colour roles, two font pairs with a reason, and three gradients. "Rainy neon Tokyo night" and "warm indie café" share that shape. The colours, type, and gradients are what change. I do not yet reject a palette that comes back generic. Font names are not checked against Google Fonts, and the Tailwind name is the model's guess. That is the part I am going to improve. Next I want a check after generation: measure the real contrast, confirm the fonts exist, and regenerate when the palette ignores the phrase and falls back to a default look. ### Part VII: What it is all for Q: What would make you say PickMyVibe has succeeded? Pokhriyal: A developer describes a feeling, pastes the result into a real project, and ships with it. The colours, type, and gradients still match the phrase a week later, and the text stays readable because the contrast score was measured. They come back for the next product instead of opening a blank Tailwind config and guessing. Q: What will design look like when anyone can describe a feeling and get a finished visual identity? Pokhriyal: Getting a first look will be easy. You describe a feeling and get the colours, fonts, and gradients right away. The real work will be choosing what fits the product and fixing what does not. A lot of apps will start from a similar draft, so the ones that feel different will be the ones a person actually edited. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Pokhriyal: Over the next 12 months, about 1,000 people who have generated a vibe, and about 100 who copied a theme into a real project. Revenue stays small on purpose. PickMyVibe is free today. Within those 12 months I want a simple paid plan for saving vibes and cleaner exports, with a target of about $500 a month. Q: And for the next 24 months? Pokhriyal: About 5,000 people who have generated a vibe, and about 200 paying users. Revenue target is about $2,000 a month. The product should remember a project’s vibe, check the fonts, and regenerate a palette that ignores the phrase. Q: Where do you want PickMyVibe to be in five years? Pokhriyal: The first place a builder goes when they know the feeling and need a theme they can ship. They type a sentence, get colours, type, and gradients they trust, and paste them into their own product. PickMyVibe holds the starting direction. The person still decides what stays. Q: What are your hopes and fears for a post AGI world? Pokhriyal: One fear that I have for post-AGI would be that creativity would decrease and people would start using generic templates, but I still have hope that people would always want their product to give a certain kind of vibe so PickMyVibe would still thrive in that. About PickMyVibe: PickMyVibe is an aesthetic copilot built by Dhruv Pokhriyal in Delhi. It turns a mood, vibe or phrase into a small design system: five colours with named roles, two Google Font pairings and three CSS gradients, with contrast measured by the app and exports as CSS variables, a Tailwind theme snippet or a Google Fonts import. It is free to use. ## PackProof (USA) URL: https://startup.talk/usa/packproof/ Founders: Collin Neri Collin Neri grew up in Columbus, Ohio, always tinkering and building things: Legos first, then software once a first laptop arrived. A first job at the swimming pool taught the satisfaction of earning money and putting it into something worth caring about. Before founding PackProof, Neri worked across analytics and communications and spent time in collectibles and auction coordination. PackProof grew out of collecting. Buying and selling collectibles made the need for a clear record of every transaction very real, and collector groups were full of sellers posting about the trust gap every single day. The product turns packing recordings, order details and shipping activity into a Proof: one neutral, source-linked record for a shipment. We spoke about why a Proof should never take a side, what a recording cannot prove, and the move from collectors to warehouse and logistics operations. Key takeaways: - A Proof is an organised record that ties photos or video, shipment identifiers, order details and shipping activity to one shipment, viewable in any browser. - PackProof stays neutral by design: it does not authenticate items, assign grades, decide disputes or replace a WMS, TMS or ERP. - Biometric attestation links each recording to the owner of the device that captured it, rather than a simple honesty checkbox. - A few thousand people use the seller workflow today, and Neri says PackProof is in discussions with close to a dozen logistics companies about warehouse and distribution pilots. The Principal / Founder: - Collin Neri (Founder) Grew up: Columbus, Ohio, USA Before: Analytics and communications; collectibles and auction coordination Website: thepackproof.com ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Neri: I grew up in Columbus, Ohio. I was always someone fascinated with technology and building things or tinkering with stuff. Q: What first got you into building things? Neri: I would say that my first laptop is what got me interested in building software, but in general it was definitely Legos that sparked my creativity. Q: What was your first job, and what did it teach you? Neri: My first job ever was at the swimming pool and it taught me about how great it feels to be able to make some money and then use that money to invest into something you enjoy. ### Part II: The road to PackProof Q: What made you start PackProof? Neri: I was a collector before I was a founder. Buying and selling collectibles made the need for a clear record of a transaction very real to me. PackProof started with packing and shipping evidence, but the bigger opportunity became clearer as I built it: businesses already have a lot of useful information, and it is scattered across different systems. I want to make that information easier to connect, understand and use. Q: What happened to you or to sellers you know that showed how often packing evidence gets lost in a camera roll or an inbox? Neri: Every single day in my social media feeds I see collecting groups where people are posting on a daily basis how they have hesitance and trust gaps with selling their collectible items and it became really clear that not only is this a real problem but it's one people deal with very regularly. Q: Why did you decide that PackProof should stay neutral instead of taking the seller's side in a dispute? Neri: Because the record needs to be useful regardless of which side it supports. If PackProof starts with a preferred outcome, it undermines the reason to trust it. Our job is to preserve and organize evidence, make its source and limits clear, and let the people responsible for the decision assess it. Q: What has it been like to build PackProof in its early days? Neri: It has involved a lot of learning and iteration. Building something that functions is one challenge; making it understandable to someone who has never used it is another. I have spent a lot of time simplifying the experience without removing the capabilities that make it useful. More recently, my focus has shifted toward finding pilot customers and learning where the product can solve a real operational problem. ### Part III: The product Q: Why the name PackProof? Neri: It actually came from a conversation I had with one of the people who has been working on it with me. They said: "Okay, yeah, a buyer can 'say that', but where's the proof?" The rest is history. Q: What does a Proof contain, and how does a seller create one while packing? Neri: A Proof is an organized record that connects evidence to a particular shipment or transaction. Depending on the workflow and the information available, that can include photos or video, shipment identifiers, order details and shipping activity. The packing workflow is about capturing the relevant evidence and associating it with the shipment so it can be found later. For warehouse operations, the direction is to use existing records wherever possible rather than ask staff to record everything again. The contents depend on the sources actually available; a Proof should not imply that missing information exists. We are incredibly proud of what we built especially with the latter logistical implementation of our systems. We effectively have created a platform that can be installed by someone who isn't even technically skilled, and have the entire creation of a proof process be hands off from there. Q: How do grading submissions and returns use PackProof differently from disputes? Neri: The question being documented changes. For a grading submission, the useful record concerns the item and its condition before it leaves the sender. For a return, it concerns what comes back and how that compares with the available outbound evidence. A dispute is only one reason someone might need those records. PackProof does not authenticate an item, assign a grade or decide whether a refund should be issued, but we can give collectors a robust single evidence packet documenting their item's condition so as to protect its integrity at the time of shipping in case there's any issue afterwards. Q: Who reviews a Proof: buyers, marketplaces, carriers or payment providers? Neri: The intended reviewer is whoever is authorized and needs to understand that shipment: that could be a seller, an operations manager or someone investigating an exception. Buyers, carriers, marketplaces and payment providers may have their own evidence requirements and submission channels. I do not claim that all of them accept a PackProof record or that sharing one guarantees a particular outcome. However, a Proof is viewable by all parties should it be chosen to have those permissions upon creation. Proofs are ubiquitously accessible in any browser too, so no matter who is viewing it, their existing tools and systems can open it. Q: What do you deliberately not claim a recording can prove? Neri: A recording shows what was captured, within the limits of its viewpoint and context. It cannot establish everything that happened outside the frame or after the recording stopped. A file hash can help check whether a file has changed; it does not prove that the original scene was truthful or complete. We should be explicit about gaps rather than turn a partial record into certainty. However, our biometric attestation adds an additional layer to this dynamic too. Rather than just having users check a box that asserts they are honestly and truthfully stating the item in the video or photo is the item they are shipping, we are having the biometrics validation check tie that to whoever owns the device where the video / photo is being taken from too. We have done an excellent job, I think, when it comes to being able to provide so much information to, let's say, a claims specialist that logic and reason will be easy to follow when doing their investigation with a Proof as their evidence submitted. Q: What do you deliberately leave out of PackProof? Neri: I deliberately keep adjudication out of the core product. PackProof should not decide who is at fault, guarantee a claim or quietly turn evidence into a refund decision. In the warehouse direction, I also do not want to replace a customer's WMS, TMS or ERP. The aim is to make the information those systems already hold more useful together. We understand that the world we live in is one where the very concept of what is true and what is not increasingly relies upon more robust evidence to bridge the trust gap that exists between people and the information they receive. We take that information and make it easy to read, and more useful than it would be in a typically fragmented state. ### Part IV: Positioning Q: Who is PackProof for, and what job does it do for them? Describe the person who gets the most value. Neri: The original audience was people shipping valuable items, including collectors and small sellers. The area I am now putting the most attention into is warehouse and logistics operations. The person I want to help is the one who has to reconstruct what happened to an order by checking several systems, documents and recordings. The job is to give that person a clear, source-linked record so an exception can be reviewed with less searching. We are pursuing pilots to test that value rather than assuming every operation has the same problem. In other words, we create evidence upstream of the problem. We do not want to wait until something happens in order to be compelled to go back and assemble evidence. That evidence should already be compiled and easy to navigate to and find what is required for any sort of issue ahead of time and that's what we set out to do and have achieved with our systems. Q: How do you view your positioning relative to Shipsurance, Route, Seel, marketplace seller protection programmes and sellers simply filming on their phones? Is there another competitor you think of first? Neri: We position ourselves as neutral evidence infrastructure because that truly is something we believe in at the core of what we do. Many of these companies are ones which simply adjust the liability terms of a shipment, but what we are doing is enhancing what they already do by creating an evidentiary packet in the form of a Proof that makes the information more usable and readable. Q: What is your core moat in shipment evidence? Neri: The differentiation I want to build is in reliable evidence linkage: preserving the original material, connecting it to the right transaction, and presenting it in a way people can actually navigate. Integration and ease of use matter as much as the record itself. I would rather demonstrate that value in real workflows than claim that we already have an unassailable moat. Q: What are some of the use cases for PackProof that are not obvious and can add a lot of value to users? List as many as possible. Neri: Beyond disputes, potential uses include routine shipment audits, reconstructing an order across warehouse preparation and carrier handoff, reviewing a packing or dispatch exception, comparing outbound and returned-item condition, and documenting a grading submission. The common thread is being able to review what happened without hunting through every source separately. These are use cases to validate against the available data and supported workflow, not a claim that every one is fully automated today. Q: What is the one detail in PackProof you are proudest of? Neri: I am most proud of how we have been able to create solutions that are as hands off as they can be for the user. Evidence exists in the world already but there is a lot of work that goes into creating something that is able to coherently assemble all of that into a way that's extremely easy to understand and is contextually relevant for the person using it without them having to know the specifics of how it works and that's what I'm proudest of for sure. ### Part V: Launch Q: What did you learn from your launch? Neri: It's a very competitive world out there! You really have to be on your game to stand out. Q: Which feedback from the launch surprised you? Neri: The biggest takeaway for me has been how much the experience matters alongside the idea. Something can make perfect sense to me as the person building it and still feel complicated to someone seeing it for the first time. That has pushed me to make PackProof do more of the work for the user. ### Part VI: The hard part Q: How many people use PackProof today? Neri: We are in discussions with several major logistics companies across a number of industries to integrate our systems into their warehouse and distribution workflows as we speak. Close to a dozen, with several being major players internationally. We have a few thousand active users at the time of writing this who use PackProof's seller workflow as well. Q: What is a challenging time you had building PackProof? Neri: One recurring challenge has been closing the gap between technical capability and a setup experience that an ordinary person can use. It is easy to expose every option and end up with an interface that makes sense to the builder but overwhelms everyone else. I have repeatedly pushed for simpler screens, clearer next steps and keeping technical settings out of the main flow. The goal is to reduce the burden on the person using it, not just get the software installed. Q: Who believed in you before there was anything to show? What did they say or do? Neri: For me, belief is most meaningful when someone takes the idea seriously enough to challenge it. Honest feedback and a willingness to help work through a problem matter more than simply being told something sounds great. That is the kind of support I value as I build PackProof. Q: How do you get marketplaces and payment providers to accept a Proof as evidence? Neri: That acceptance has to be earned and verified with each organization. The starting point is understanding what evidence it asks for, which formats and channels it accepts, and what context its reviewers need. PackProof can organize a record, but it cannot require a marketplace or payment provider to accept it or decide a case a certain way. Those relationships and workflows need practical testing rather than blanket claims. ### Part VII: What it is all for Q: What would make you say PackProof has succeeded? Neri: Success would mean people rely on PackProof because it makes a real part of their work easier. For the logistics direction, I want to see that it helps an operator assemble and review shipment information with less effort, using the systems they already have. A successful pilot should give us evidence of that value, and a sustainable business should follow from delivering it repeatedly. Q: How will AI change the way disputes between buyers and sellers are resolved? Neri: I think AI will make it much easier to organize evidence, spot inconsistencies and help reviewers understand what happened. But a confident answer is only useful if it is grounded in reliable information. The original evidence, its source and any gaps still need to be visible, with people accountable for consequential decisions. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Neri: Over the next 12 months, my priority is turning pilot conversations into working deployments and then paying customers. I want to demonstrate that PackProof saves operators time and is straightforward to adopt. Meaningful repeat usage and recurring revenue matter more to me than chasing a headline user count. Q: And for the next 24 months? Neri: By 24 months, I want a repeatable business: customers renewing because PackProof is useful, integrations that are easier to roll out, and recurring revenue that supports sustainable growth. The goal is to expand from successful pilots into broader use without losing the simplicity that makes the product worth adopting. Q: Where do you want PackProof to be in five years? Neri: I want PackProof to become a trusted layer for connected evidence across commercial workflows, with shipping as the starting point. The ambition is for it to fit into existing systems and make their information more useful without adding unnecessary work. Getting there means proving the value in specific use cases first, rather than trying to claim the whole market at once. Q: What are your hopes and fears for a post AGI world? Neri: My hope is that increasingly capable AI gives more people the ability to build useful things and solve problems that used to be out of reach. My concern is how much harder it could become to distinguish reliable information from convincing fabrication. We will need stronger ways to establish where information came from and keep people accountable for how it is used. About PackProof: PackProof is neutral evidence infrastructure for e-commerce and logistics, founded by Collin Neri. It connects packing recordings, order details and shipping activity into a Proof, a clear, source-linked record for each shipment, for sellers, collectors and warehouse operations, without replacing the systems that created the data. thepackproof.com ## Jotted (Philippines) URL: https://startup.talk/philippines/jotted/ Founders: Christine Zafra Christine Zafra was born in Parañaque City in Metro Manila and grew up in Davao City, an only child who made flashlights out of loose bulbs and batteries at eight and wired a phone extension into the family landline at nine. Zafra's career began in 2008 as an academic writer working to three-hour deadlines, then moved into virtual assistant work, where most of the day is spent in client meetings. Jotted grew out of those meetings. Recorders quit mid-call, notetaker bots were refused entry, and action items slipped while Zafra tried to listen and take notes at the same time. So Zafra built a Windows recorder that captures the call from your own screen with no bot, keeps the full video, and gives the free plan no limits. We spoke about a 2e ADHD diagnosis at 38, a launch day that looked like a flop, hours spent counting to 30 into different microphones, and why people should own their meetings. Key takeaways: - Jotted is a Windows meeting recorder that captures Zoom, Microsoft Teams and Google Meet calls from your own screen, with no bot in the call, and keeps the full video alongside the transcript. - The free plan records locally with no account and no limit on length or number of meetings. Pro, at $14 a month, adds AI notes, action items and Quill, which searches past meetings and jumps to the exact moment in the video. - Zafra built it alone, without telling anyone until it was finished. It launched mid-year, has 18 subscribers, and most updates since launch came from what users asked for. - The targets: 50 paying customers and a Mac version within 12 months, and 200 paying customers within 24. The Principal / Founder: - Christine Zafra (Founder) Based: Philippines Grew up: Davao City (born in Parañaque City, Metro Manila) Studied: De La Salle University; St. Scholastica's College, Manila Before: Academic writer (from 2008); virtual assistant Pricing: Free plan with no limits; Pro $14 a month Built: Solo Website: getjotted.com LinkedIn: Christine Zafra (https://www.linkedin.com/in/christine-zafra/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Zafra: I was born in Parañaque City in Metro Manila, but I grew up in Davao City, one of the major cities in the south of the Philippines. It’s a big city, but life there was pretty laid back. My playmates were my neighbors, and I only had two or three real friends out of them. We played outside until 6 PM because my grandmother was strict about it, and once it hit 6 I had to be inside the house. I’m an only child, so I played on my own a lot. I was also the youngest of all my cousins, and the next one up from me was about ten years older. So at family reunions or when we all went out together, I sometimes felt isolated because there was nobody my age to talk to. I got used to doing things by myself pretty early. Q: What first got you into building things? Zafra: I’ve been curious about how things work for as long as I can remember. When I was about 8, I made flashlights out of bulbs I found lying around and strapped batteries to them, and that’s how I figured out something about voltage, because one battery wouldn’t light the bulb but two would. At 9, I installed a phone extension for our landline. A phone extension was an added subscription, and my mom didn’t want to pay for one, so I found an old phone that still worked and wired it into the same line myself. At 10, I put together my first computer. Nobody showed me how, and I got it from the shapes, because every part only goes in one way and if something won’t fit a slot then it wasn’t made for that slot. I remember my mom let me stay home from school that day so I could keep playing with it. Q: What was your first job, and what did it teach you? Zafra: My first job was as an academic writer, and that’s really how my career started. This was around 2008, long before AI, and I was writing for students in the US, the Middle East, and Europe. The projects came with a three-hour deadline, and that covered everything: reading the sources, making sure I understood them, writing the whole paper from scratch, and adding the citations as I went. It taught me how to work under intense pressure. ### Part II: The road to Jotted Q: What made you start Jotted? Zafra: I went from being a writer to being a virtual assistant, and I know how it is at the entry level in the Philippines. The hourly rates start really low and you can be strapped for cash, which is why I built Jotted with a very generous free plan. I wanted VAs who are just starting out to be able to use it without paying. As a VA you do everything, and a big part of that is sitting in meetings all day. So I built Jotted to fix my own problem first. I was using a different tool for each part of a meeting. There are plenty of recorders out there, but some don’t keep the video, some only keep the transcript, and some only give you the audio. The tools that do all of it put a bot in the meeting, and that gets awkward. A lot of the time my bot wasn’t allowed in, which meant I didn’t get the whole recording. Some of the action items and to-dos got missed too, because I was in the meeting and trying to jot down notes at the same time. Q: What happened on client calls that convinced you the conversation itself was being lost? Zafra: I’m not going to name any tools, but I’ve lost a lot of meetings to recorders that quit on me. One AI transcription tool just stopped recording in the middle of a call. Another one froze mid-meeting, so I never got the rest of the conversation. My AirPods died on me during a call, and my laptop switched to a different audio input the way it’s supposed to, but the recorder never followed it to the new default Windows input. It stopped recording my voice, and I only found out afterward. I’ve lost a lot of meetings because of things like this. Q: How does being 2e ADHD shape the way you find problems and build products? Zafra: I’m a late diagnosis. I’ve always known my IQ, but the ADHD part was new to me. In high school and college that was really perplexing, because I knew I was smart but I couldn’t seem to follow through, and my grades weren’t good. Once I got into the workforce, I started noticing things I’d done in the past that leaned toward ADHD, but I didn’t have a name for it yet. I got my diagnosis at 38, and that’s when it finally made sense. I tunnel into things. When I’m really interested in something, I can learn in three hours what takes other people three weeks. I also don’t take no for an answer, so I go all the way to the bottom of a problem before I start building. And I have a lot of ideas, so I tend to start multiple projects at the same time. That kind of creativity is the good side of being 2e ADHD. Q: What is it like to build Jotted solo from the Philippines? Zafra: Building Jotted solo from the Philippines has been a good experience. There aren’t many Filipino creators out there, so being one of them feels like an honor. ### Part III: The product Q: Why the name Jotted? Zafra: I named it Jotted because it jots down the notes and action items for you, so the name says exactly what it does. Q: Why record without a bot, and how do you explain that choice to people worried about consent? Zafra: A bot in the call changes the room. An extra tile shows up with a notetaker’s name on it and sits there while everyone talks, and that gets awkward fast. If I’m not the host, the bot sometimes doesn’t even get let in, so the first thing I’d say on the call is a request to admit my notetaker. Jotted records from my side of the screen, so nothing joins the call at all. When it comes to consent, I always tell people out loud that I’m recording. It’s one of my must-dos. I tell them what the recording is for, and if they’d rather I didn’t record, I don’t. Q: Why keep the full video when most bot-free tools keep only a transcript? Zafra: A transcript can’t read emotion. You can go through every word of a call and still have no idea how the person reacted when something came up, like the price. The video is where you see the micro-expressions and the non-verbal cues, and I wasn’t willing to leave those out, so Jotted keeps the entire video. Q: Why start with Windows when most bot-free recorders are Mac-first? Zafra: Most bot-free recorders are Mac-first, which leaves a gap for people on Windows, so that’s where I started. It’s also what I use myself. Mac is next. Q: How does Quill change the way people go back to a past meeting? Zafra: Quill lets you keep all your meetings in one tool, Jotted, no matter how many clients you have, and ask questions across every recording. You don’t have to remember when something came up. You ask Quill about it, and if it finds the moment, it gives you the timestamp from the video. Click it and you jump to that exact spot in the recording, so you can watch it again for yourself. ### Part IV: Positioning Q: Who is Jotted for, and what job does it do for them? Describe the person who gets the most value. Zafra: Jotted is for people who spend their day in client meetings and are also the one who has to remember everything that was said. I started with virtual assistants because that’s the job I came from. As a VA you do a bit of everything, you’re on calls all day, and you still have to turn those calls into to-dos afterward. The person who gets the most value is someone like that, usually early in their career or freelancing, who doesn’t have much room in the budget and doesn’t want a bot sitting in their client’s call. Jotted records the video and audio from your side of the screen and gives you the transcript, and on Pro it also pulls out the notes and action items. That means you can be in the conversation instead of typing through it, and you can go back and check anything later. Q: How do you view your positioning relative to Otter.ai, Fireflies, Fathom, Granola, tl;dv and Zoom's own AI Companion? Is there another competitor you think of first? Zafra: Most of the tools on that list, like Otter, Fireflies, Fathom, and tl;dv, started out by sending a bot into your call. I built Jotted because that got awkward for me, and sometimes the bot didn’t even get let in. The one I think of first is Granola, because it doesn’t use a bot either. The difference is that Granola keeps the text and not the recording, and I wanted the full video so you can see how the person reacted. Zoom’s built-in AI is tied to Zoom, while Jotted is its own app that records from your side of the screen. Q: What is your core moat in AI meeting recording? Zafra: I’d say my moat is that the recording stays on your computer. Most of these tools run everything on their own servers, so they have costs on every recording, and that’s usually why free plans get limits. Jotted doesn’t work that way, so I can keep the free plan open with no limits. That’s hard for a company built around the cloud to copy without changing how it works. The other part is speed. I’m one person, so when I get an idea I can build it and ship it the same day. Q: What are some of the use cases for Jotted that are not obvious and can add a lot of value to users? List as many as possible. Zafra: A lot of people use Jotted just to have the notes, but there’s a lot more you can do with the recordings. If a client says that’s not what we agreed on, you can go back and watch what was said instead of arguing about it. You can also see how someone reacted when the price came up, which a transcript can’t show you. Quill searches across all your meetings, so you can find something from months ago without remembering which call it was in, and it takes you right to that spot in the video. It helps with people too. A new VA or teammate can watch how a client talks and what they care about, as long as the client knows it was recorded, and when you hand a client over to someone, the recordings give them the whole history. You can watch your own calls back to see where you talked too much or missed a question. If you bill by the hour, the recording shows when a call started and how long it ran. I have ADHD, so for me it’s also a way to catch up when my attention drifts. I can read the transcript for the minute I missed instead of asking someone to repeat it. It helps when English isn’t everyone’s first language, too, because the transcript and the video together make it easier to check what someone meant. And I’d use it for interviews, whether that’s hiring calls or interviews for an article, where you want the video and the words in one place. Q: What is the one detail in Jotted you are proudest of? Zafra: The detail I’m proudest of is that the free plan has no limit on how long you can record or how many meetings you record. That works because the recording is a file on your own computer, so I’m not paying to store your video on a server, and I don’t have to cut people off to cover the cost. A lot of the people I know are freelancers and virtual assistants who watch every dollar, and I think many of them would skip a tool like this if the free version stopped at 40 minutes. I wanted the free plan to be good enough for real client work, so more people can use it without paying for anything. I’m a solo founder and I still get a little kick out of that. ### Part V: Launch Q: What did you learn from your launch? Zafra: The launch itself didn’t go well. I didn’t get the numbers I was hoping for on the day, and I’ll say that plainly. What I didn’t expect is what happened after. In the days that followed I got emails and inquiries from people asking about Jotted, and some people sent me LinkedIn connection requests too. So the launch day looked like a flop, but the people who found it later were the ones who actually reached out. I’d also say a launch is just the first time people see it, and what happens over the next few weeks tells you more than the first day does. Q: Which feedback after the launch surprised you? Zafra: I expected mostly bug reports after launching, and what came in was mostly small ideas. We went live on a Sunday at 0.1.2 and we’re on 0.1.7 already, and most of those updates came from people telling me what they wished it did, like a setting they wanted somewhere else or a step they wanted to skip. I thought I’d spend that week fixing things, and it turned out I spent a lot of it adding things people asked for. ### Part VI: The hard part Q: How many people use Jotted today? Zafra: We launched mid-year, and so far we have 18 subscribers, and that number keeps growing. It’s a small number, but each of them came to us on their own, and a few of them have emailed me directly with ideas. Q: What is a challenging time you had building Jotted? Zafra: The hardest part was the audio switching. The recording was locked to whatever device was active when I pressed start, so if Windows changed devices halfway through a call (Bluetooth dropping is the usual one), the recording kept listening to a device that was no longer there. The fix is that Jotted now follows whatever Windows says is the active device and keeps going in the same file, with no gap. Testing it meant sitting in my room talking to myself for hours and counting to 30 into different microphones while I unplugged things, which looked as silly as it sounds. It took a lot of bug fixes before it held up, because every new device combination found something I hadn’t thought of. The payment providers falling through was the other one. The first provider had fees that didn’t work out at $14 a month, and the second wanted a live, downloadable product before they would approve me, which I couldn’t give them because I needed payments working to launch. So I was stuck between two providers and a launch date, and I ended up on Dodo Payments. Q: Who believed in you before there was anything to show? What did they say or do? Zafra: I kept Jotted to myself the whole time I was building it. I didn’t tell anyone until it was a finished product, partly because I wanted to know it worked before I had to explain it to somebody. When I finally showed the people close to me, they were really proud of me, and it meant a lot because they hadn’t known I was working on it. Q: How do you keep a free plan with no account sustainable? Zafra: The biggest reason is that the recording stays on your own computer. I’m not paying to store your video on a server, so a long recording costs me nothing, and that’s why the free plan has no limit on how long you record or how many meetings you record. The part that does cost me money is the AI notes and action items, because each one is a request I pay for. That’s what’s in Pro, at $14 a month, and there’s a seven day trial so people can try it before they pay. So the people who use the most of what costs me money are the ones paying for it, and everyone else can record for free as long as they like. ### Part VII: What it is all for Q: What would make you say Jotted has succeeded? Zafra: I’d say Jotted succeeded when it’s the thing people open before every client call without thinking about it, like they would with their mic or their calendar. I’d also want enough people paying for Pro that I can keep the free plan open for everyone else, because that’s the part I care about most. And it would mean a lot to hear from a freelancer or a VA like me who said a recording saved them from an argument with a client or from a task they would have forgotten. Q: What will meetings look like when AI agents attend, take notes and act on what was agreed? Zafra: I think agents will be in a lot of meetings, and some of them will take notes and do the follow up work too, which would save people a lot of time. What I worry about is what it does to the room. I’ve seen how a bot in the participant list changes how people talk, so if every meeting has several agents in it, I think people will get more careful about what they say, and the meetings will get worse for it. The other part is that someone still has to be accountable. If an agent promises something to a client on my behalf, I’d want to go back and see exactly what was said and how it was said, and I think a recording with the video does that better than a summary. So I think the people in the meeting will still matter, and the agents will mostly work around them. Whatever happens, everyone in the meeting should know who and what is listening, and agree to it first. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Zafra: In the next 12 months I want to go from 18 subscribers to 50 paying customers. I also want a lot more people using the free plan, because those are the people who tell others about it. I’d like to have the Mac version out by then too, since a lot of people have asked and I can’t help them until it exists. Q: And for the next 24 months? Zafra: In 24 months I want Jotted to be on both Windows and Mac, with 200 paying customers. I’d also like to have built the things people ask me for most, because most of the updates so far have come from people telling me what they wish it did. Q: Where do you want Jotted to be in five years? Zafra: In five years I hope Jotted is something freelancers and small teams all over the world use for their client calls, on both Windows and Mac, and that the free plan is still as open as it is today. I’d like the recording to still be a file that sits on your own computer, because I think people should own their meetings and not just have a login to them. I’m not sure yet if it will still be only me by then or a small team, and I’m okay with either one as long as I can keep shipping at the pace I do now. I’d also like the people who were using it in the first year to still be using it, because that would tell me we got the basics right. Q: What are your hopes and fears for a post AGI world? Zafra: My biggest hope is that more people can build the things they need. I built Jotted by myself, from the Philippines, as one person, and a few years ago that would have taken a team. If AI keeps getting better at that, I think a lot more people with ideas will be able to make them real and give them away or sell them cheaply, and not just the people who can afford a team. My biggest fear is about consent and ownership. If agents are listening in on every call and every message, I want people to know who is listening and where the recording goes, and I want them to own it. I also worry about people like me. I work as a virtual assistant, and a lot of that work is the kind that AI is getting good at, so I don’t know what it means for me yet. I’m not sure how it will go, and I think anyone who says they are sure is guessing. About Jotted: Jotted is a Windows meeting recorder built by Christine Zafra in the Philippines. It records Zoom, Microsoft Teams and Google Meet calls from your own screen, with no bot joining the call, and keeps the full video, audio and transcript on your computer. The free plan needs no account and has no limit on recording length; Jotted Pro ($14 a month) adds AI summaries, action items and Quill, which answers questions about past meetings. A Mac version is planned. getjotted.com ## IntelliBook AI (India) URL: https://startup.talk/india/intellibook_ai/ Founders: Sampath Kumar Sampath Kumar grew up in Davangere, Karnataka, in a family with no business background, and found in programming something worth exploring. That curiosity became a career as a software engineer, including work on large-scale healthcare applications, where Kumar learned that reliability, security and the details matter when software runs real-world workflows. Over time, building assigned features was not enough: Kumar wanted to build products that solve problems people actually have. IntelliBook AI is an AI receptionist for businesses that live on WhatsApp and Instagram. It answers enquiries, qualifies leads, books appointments, sends quotations, collects advance payments and follows up, using official Meta integrations and letting a business keep its existing WhatsApp number. Kumar builds it alongside a full-time engineering job. We spoke about what small businesses lose to a slow reply, why a deposit inside the chat changes a booking, when a human should take over, and the first client: an immigration consultancy in Canada. Key takeaways: - IntelliBook AI handles customer enquiries on WhatsApp and Instagram from start to finish: answering questions, qualifying leads, booking appointments, sending quotations, collecting advance payments and following up, with an integrated CRM. - It runs on Meta's official WhatsApp Business Platform, and businesses can keep using their existing WhatsApp Business number. - Staff can step in from a shared inbox at any time, and the AI stops replying once a human takes over. - It is early stage and self-funded, with one active client, an immigration consultancy in Canada. Kumar is building it alongside a full-time job as a software engineer. The Principal / Founder: - Sampath Kumar (Founder) Grew up: Davangere, Karnataka, India Day job: Software engineer Before: Large-scale healthcare applications Funding: Self-funded Channels: WhatsApp and Instagram, via official Meta integrations Website: intellibookai.com ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Kumar: I grew up in Davangere, Karnataka, India. I come from a regular family, and my journey has been fairly simple. I didn't grow up with a business background or a startup environment around me. Like many people, I was figuring out what I wanted to do with my life. Over time, technology became something I genuinely enjoyed exploring. Looking back, I think my curiosity and willingness to learn have played a big role in getting me where I am today. Q: What first got you into building things? Kumar: My interest started with programming. Initially, it was about understanding how websites and applications worked. But the more I learned, the more interesting it became to actually build something myself. I enjoyed the process of taking an idea, breaking it into smaller problems, writing code and eventually seeing something useful come out of it. Over time, I realised I didn't just want to work on assigned development tasks. I wanted to build products that solve problems people genuinely experience. That mindset eventually pushed me towards entrepreneurship. Q: What was your first job, and what did it teach you? Kumar: Working as a software engineer taught me that building software is about much more than writing code. You need to understand the actual problem, work with different people, handle unexpected situations and take responsibility for what you deliver. My experience working on large-scale healthcare applications has also taught me the importance of reliability, security and getting the details right. When software is used in real-world workflows, even small mistakes can have consequences. Those lessons influence how I approach IntelliBook AI today. I want to build something that businesses can genuinely depend on, not just something that looks impressive in a demo. ### Part II: The road to Intellibook AI Q: What made you start IntelliBook AI? Kumar: The idea came from a problem that seemed simple but had a real business impact. Businesses receive customer enquiries through WhatsApp and Instagram every day. But responding to every message, answering repetitive questions, managing bookings and following up with potential customers takes a lot of time. A customer might enquire at night, during a busy working hour or when nobody is available to respond. By the time the business replies, that customer may already have contacted a competitor. I started thinking: why should businesses lose potential customers just because they couldn't respond at the right time? That's what led me to build IntelliBook AI. I wanted to give businesses an AI-powered assistant that could handle customer interactions, bookings and follow-ups while helping the business owner stay in control. Q: What did you see salons, clinics and service businesses losing to slow WhatsApp replies that convinced you to build IntelliBook AI? Kumar: One thing I noticed is that many small businesses depend heavily on WhatsApp, but their customer management process is still mostly manual. A customer asks about pricing, another wants an appointment, someone else needs to reschedule, and another person is waiting for a follow-up. All these conversations end up mixed together. The problem isn't necessarily a lack of customer enquiries. Sometimes it's the inability to manage those enquiries properly. A missed reply can mean a missed sale. A forgotten follow-up can mean a lost customer. And a booking without an advance payment can turn into a no-show. I wanted to solve the entire workflow rather than build another chatbot that simply replies to messages. Q: Why did you build on the official WhatsApp Business Platform and let businesses keep their existing number? Kumar: Because businesses have already built relationships with their customers on WhatsApp. Asking them to change their number or move everyone to a new application creates unnecessary friction. We wanted IntelliBook AI to fit into the way businesses already operate. Using Meta's official WhatsApp Business Platform gives us a proper foundation for business messaging, automation and integrations. Our approach also supports WhatsApp coexistence, allowing eligible businesses to continue using their existing WhatsApp Business app alongside the platform. The idea is simple: businesses shouldn't have to change how their customers communicate just to start using AI. Q: What is it like to build IntelliBook AI as a founder selling to small businesses? Kumar: It's a completely different experience from building software as an engineer. As a developer, you often know what needs to be built. As a founder, you first have to understand whether the problem you're solving is important enough for someone to pay for. Selling to small businesses has taught me that having more features doesn't automatically make a product valuable. Business owners want to know whether it will save them time, help them convert more enquiries or reduce the work they deal with every day. I'm also building IntelliBook AI while working full-time as a software engineer. That means balancing product development, customer conversations, marketing and everything else that comes with running an early-stage startup. It's challenging, but talking directly to business owners and seeing how they use what we're building makes the effort worthwhile. ### Part III: The product Q: Why the name IntelliBook AI? Kumar: The name combines two ideas: intelligence and booking. We initially started with the idea of making appointments and bookings easier through AI. But as the product evolved, it became much bigger than scheduling. Today, IntelliBook AI helps businesses manage customer conversations, enquiries, leads, bookings, payments and follow-ups. The name still represents our original idea, while the product has grown into a broader business management platform. Q: How do deposits collected inside the chat change no-show rates for your customers? Kumar: A booking doesn't always mean a confirmed customer. Many businesses experience customers booking an appointment and then not showing up. This creates empty slots, wasted staff time and lost revenue. With IntelliBook AI, businesses can request an advance payment or deposit directly through the conversation using UPI or payment links. Once the payment is confirmed, the booking can be treated as a more committed reservation. Combined with automated reminders and follow-ups, this helps businesses reduce avoidable no-shows. The bigger idea is to connect conversations, bookings and payments instead of treating them as three separate activities. We don't believe deposits alone eliminate no-shows, but they can help create stronger booking commitments. Q: How does IntelliBook AI learn a business from its website so quickly? Kumar: We want onboarding to be as simple as possible, especially for business owners who aren't technical. Instead of asking someone to manually write hundreds of instructions, the idea is to use information already available about their business, including their website, services, pricing and business details. IntelliBook AI uses that information to build a business knowledge base that the AI can refer to when answering customer questions. Business owners can also provide additional documents, policies and instructions so the assistant understands how their business actually works. The goal is to reduce the time between signing up and having something useful working for the business. Q: When should a human take over a conversation, and how do you make that one tap? Kumar: AI should handle repetitive and predictable work, but it shouldn't replace human judgement where it matters. For example, a customer might have a special request, negotiate a large order, raise a complaint or ask something outside the AI's knowledge. Those situations should be handled by a real person. That's why human control is an important part of IntelliBook AI. Through the shared inbox, staff can review conversations and step in whenever necessary. When a human takes over, the AI should stop responding so that customers aren't receiving conflicting messages. Our philosophy is that AI should make business owners more capable, not take control away from them. Q: Which industries get the most value from a WhatsApp receptionist? Kumar: Any business that receives enquiries through messaging can benefit, but the value is especially high for businesses where a conversation can turn into a booking, quotation or sale. Some examples include: - Furniture and home decor showrooms managing product enquiries and showroom visits. - Interior designers and kitchen studios qualifying project requirements and arranging consultations. - Clinics, salons and wellness centres managing appointments and reminders. - Gyms and fitness studios handling membership enquiries and trial-session bookings. - Tiles and bathware stores answering product and pricing questions. - Automobile businesses managing test drives and service appointments. - Real estate businesses qualifying buyers and scheduling property visits. We are particularly interested in businesses that receive regular enquiries but struggle to respond and follow up consistently. Q: What do you deliberately leave out of IntelliBook AI? Kumar: We don't want to build a product that automates everything just for the sake of saying it's AI-powered. We deliberately don't want the AI making promises it cannot keep, inventing product information, confirming unavailable appointments or handling sensitive situations without appropriate human involvement. We also don't want to force customers to download another application or create another account just to interact with a business. The product should work within familiar communication channels, provide accurate business information and make it easy for a human to take over. For us, useful automation is more important than unnecessary automation. ### Part IV: Positioning Q: Who is IntelliBook AI for, and what job does it do for them? Describe the person who gets the most value. Kumar: IntelliBook AI is for business owners who receive customer enquiries through WhatsApp, Instagram and other digital channels but don't have the time or resources to manage every conversation manually. Think of a furniture showroom owner who receives 40 enquiries about different products, prices and availability. Or a clinic that spends a significant amount of time answering appointment questions and managing cancellations. These business owners don't necessarily need another employee just to answer repetitive questions. They need a better system. IntelliBook AI acts as an AI receptionist and customer management assistant. It helps capture enquiries, qualify leads, answer questions, manage bookings, collect payments and keep customer information organised through an integrated CRM. The people who benefit most are owners who want to focus on running their business instead of constantly checking their phones. Q: How do you view your positioning relative to WATI, Interakt, AiSensy, Calendly-style booking tools and the cheap AI bots flooding Instagram? Is there another competitor you think of first? Kumar: I think there are different categories of products solving different parts of the problem. Platforms like WATI, Interakt and AiSensy provide WhatsApp communication, marketing and automation capabilities. Calendly-style tools focus primarily on scheduling. Generic AI bots often focus on answering questions. We see an opportunity to bring these workflows closer together. IntelliBook AI isn't intended to be just another WhatsApp messaging tool or a booking calendar. We want it to act as an AI-powered front desk and customer management platform that can take an enquiry from the first conversation through qualification, booking, quotation, payment and follow-up. Our focus is on making the complete workflow useful for small and growing businesses, without requiring them to combine several disconnected tools. Ultimately, our biggest competition is often the existing manual process: an owner managing everything through their personal phone, spreadsheets and memory. Q: What is your core moat in AI booking on WhatsApp? Kumar: I don't think simply using an LLM or connecting to WhatsApp is a sustainable moat. Those capabilities will become increasingly accessible. Our focus is on understanding actual business workflows and building the operational layer around them. That includes industry-specific knowledge, customer history, lead qualification, staff availability, bookings, quotations, payment verification, reminders and follow-ups. The more a business uses IntelliBook AI, the more its customer interactions and workflows become organised in one place. We also care about the practical details: official integrations, existing-number support, human handover, transparent messaging costs and an experience that non-technical business owners can use. I believe the long-term advantage will come from execution, customer trust and how deeply we solve a business's day-to-day problems. Q: What are some of the use cases for IntelliBook AI that are not obvious and can add a lot of value to users? List as many as possible. Kumar: There are quite a few use cases beyond simply answering WhatsApp messages. Some examples: 1. Recovering lost enquiries by automatically following up with customers who didn't complete a purchase. 2. Qualifying leads based on budget, requirements, location and purchase timeline. 3. Generating itemised quotations for businesses like furniture stores and interior studios. 4. Scheduling showroom visits instead of just appointment bookings. 5. Collecting advance payments to confirm high-value appointments or orders. 6. Following up on quotations that customers haven't accepted. 7. Managing multiple staff members and their availability. 8. Sending reminders for appointments, consultations and service visits. 9. Collecting customer feedback and encouraging Google reviews after a service. 10. Responding to Instagram enquiries and converting them into qualified leads. 11. Understanding voice notes from customers who prefer speaking instead of typing. 12. Managing repeat customers using their previous interactions and purchase history. 13. Sending relevant offers to customer segments through approved WhatsApp broadcasts. 14. Helping business owners understand daily enquiries, bookings and business activity. 15. Managing customer enquiries from advertisements without requiring someone to manually respond to every lead. One use case I find particularly interesting is quotation recovery. A customer might enquire about a ₹2 lakh furniture order, receive a quotation and then disappear. A timely, relevant follow-up could make a real difference to that business. That's where we see AI creating measurable business value rather than simply generating replies. Q: What is the one detail in IntelliBook AI you are proudest of? Kumar: I'm proud that we're trying to solve the complete customer journey rather than stopping at the chatbot. A customer can discover a business, ask questions, get qualified, receive a quotation, book a visit or appointment, make a payment and receive follow-ups through a connected workflow. I also care about the fact that businesses can continue using their familiar WhatsApp number and that a human can step in whenever needed. As an engineer, I enjoy building the technology behind it. But as a founder, what makes me happiest is the possibility of giving a small business capabilities that previously required several people and different software tools. ### Part V: Launch Q: What did you learn from your launch? Kumar: Product Hunt taught me that building a product and getting people to discover it are two very different challenges. You can spend months building features, refining the experience and solving technical problems, but none of that automatically brings customers. The launch also pushed me to explain IntelliBook AI in a much simpler way. People should understand the problem we solve within a few seconds. I learned that founders need to spend as much time listening, communicating and talking to potential customers as they spend building the product. For an early-stage startup, a launch is not the finish line. It's another opportunity to learn, get feedback and improve. Q: Which feedback from the launch surprised you? Kumar: One thing that became clearer to me was that the value isn't just about having AI respond to messages. What matters more is what happens after the first reply. Can the business qualify the customer? Can it confirm a booking? Can it collect a deposit? Can it follow up if the customer disappears? That made me think even more about the complete customer journey. It reinforced our decision to build IntelliBook AI as a broader business workflow and CRM platform instead of limiting it to an AI chatbot. We're still learning from every conversation and using that feedback to shape the product. ### Part VI: The hard part Q: How many people use IntelliBook AI today? Kumar: Currently, we have one active client in Canada who is using IntelliBook AI for their immigration consultancy business. We're at an early stage, and having a real business use our product is an important milestone for us. We're learning from how they handle customer enquiries, qualify potential clients and manage follow-ups. Our focus right now is on working closely with early customers, understanding their needs and improving the product based on real-world usage. I'd rather build a product that genuinely helps a small number of businesses first and grow from there. Q: What is a challenging time you had building IntelliBook AI? Kumar: One of the biggest challenges has been building a startup while working full-time as a software engineer. During the day, I work on enterprise software, and outside working hours, I'm working on IntelliBook AI. That means product development, fixing issues, talking to potential customers, marketing and handling business operations all compete for the same limited time. Another challenge is that building something technically impressive doesn't guarantee that customers will understand its value or be willing to pay for it. There have been moments when I've had to rethink features, positioning and even the industries we should focus on. I've learned that entrepreneurship requires a lot of patience. Sometimes the hardest part isn't writing the code. It's figuring out what is worth building next. Q: Who believed in you before there was anything to show? What did they say or do? Kumar: I've had encouragement from people around me who were supportive of my decision to explore something beyond my regular software engineering career. Building a startup alongside a full-time job isn't always easy, and having people who understand that you're trying something different makes a difference. I've also learned to value the people who are willing to give honest feedback, test early versions and point out what isn't working. At this stage, I appreciate practical support more than big promises. Even someone taking the time to understand the product and give genuine feedback can make a difference. Q: How do you make an AI receptionist feel trustworthy to serious buyers instead of annoying them? Kumar: Trust is probably one of the most important things when you're introducing AI into customer communication. Customers don't want to feel like they're talking to a machine that keeps repeating scripted answers. They want accurate information, quick responses and the ability to speak to a real person when necessary. We approach this in a few ways. First, the AI should use verified business information instead of guessing about prices, services or availability. Second, it should understand the context of a conversation rather than simply match keywords. Third, businesses should always have control. Staff can review conversations and take over whenever required. We also build on official platform integrations and focus on responsible handling of customer information. Our goal isn't to make customers think they're talking to a human. It's to make the interaction genuinely helpful, transparent and reliable. ### Part VII: What it is all for Q: What would make you say IntelliBook AI has succeeded? Kumar: I'd consider IntelliBook AI successful when business owners can confidently say that it has made a real difference to how they run their business. If a salon owner is no longer spending hours answering repetitive questions, a furniture showroom is converting more enquiries into showroom visits, or a clinic is managing appointments and follow-ups more efficiently, that's success to me. I want IntelliBook AI to become something businesses genuinely depend on every day, not another subscription they forget about after a month. Of course, building a sustainable and profitable company matters too. But I want that growth to come from solving real problems for customers. Q: What will a small business front desk look like when AI answers every message instantly? Kumar: I think the traditional front desk will evolve into a combination of AI and human employees working together. AI will handle the repetitive activities: answering common questions, qualifying enquiries, checking availability, sending quotations, scheduling appointments, collecting payments and following up. Human employees will have more time to focus on customer relationships, complex requests, service quality and decisions that require personal judgement. A business owner should be able to see what's happening across the business without constantly checking multiple WhatsApp conversations. I don't see this as replacing the human element of a business. I see it as removing unnecessary administrative work so people can spend more time doing what they're actually good at. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Kumar: Over the next 12 months, my biggest priority is to establish strong product-market fit and build a sustainable customer base. I want to focus on a few things: - Onboard businesses that actively use IntelliBook AI in their daily operations. - Build deeper workflows for our initial target industries, particularly retail showrooms and service businesses. - Improve lead qualification, CRM, quotations, payment collection and automated follow-ups. - Expand our channel integrations and voice capabilities. - Build a predictable customer acquisition and retention process. - Reach consistent recurring revenue rather than depending on one-time sales. I'd rather grow with paying customers who genuinely depend on the product than set an aggressive user target without a strong foundation. Revenue growth should follow from solving a problem businesses are willing to pay to solve. Q: And for the next 24 months? Kumar: In 24 months, I want IntelliBook AI to move beyond being an early-stage AI receptionist and become a recognised business automation platform. I'd like to expand into more industries and markets, strengthen our CRM and business intelligence capabilities, and build deeper integrations with the tools businesses already use. I also want to build a stronger team so that product development, customer success and business growth don't depend on a single founder. International expansion is another area I'd like to explore, particularly markets where WhatsApp is an important business communication channel. The bigger objective is to build a sustainable SaaS business with recurring revenue and customers who stay because the product genuinely helps them operate better. Q: Where do you want IntelliBook AI to be in five years? Kumar: In five years, I'd love IntelliBook AI to become the AI operating layer for small and medium-sized businesses. Today, many businesses use separate tools for customer communication, bookings, payments, marketing and customer management. I think there's an opportunity to bring these activities together through intelligent automation. I want a business owner to be able to understand what's happening in their business, manage customer relationships, identify opportunities and automate routine operations from one place. I'd also like IntelliBook AI to become a product that works across countries and languages, especially for businesses that don't have the resources to build their own technology infrastructure. Ultimately, I want to build a company that makes advanced AI capabilities accessible to ordinary business owners, not just large enterprises. Q: What are your hopes and fears for a post AGI world? Kumar: I think AGI could fundamentally change how we build software, run businesses and approach everyday problems. My biggest hope is that advanced AI makes powerful technology accessible to everyone. A small business owner should eventually have access to capabilities that once required an entire team of specialists. I also think it could create opportunities for people who are willing to learn, experiment and build new things. My concern is how quickly the world might change. Job displacement, overdependence on AI, privacy, security and the concentration of technological power are all important issues. As someone building an AI product, I don't believe the goal should be to automate everything without thinking about the consequences. I hope we move towards a future where AI gives people more opportunities, helps smaller businesses compete and takes away repetitive work, while humans remain responsible for important decisions. I'm optimistic about AI, but I think building it responsibly matters just as much as building it quickly. About IntelliBook AI: IntelliBook AI is an AI receptionist and business management platform built by Sampath Kumar in India. It helps businesses manage customer enquiries on WhatsApp and Instagram, qualify leads, book appointments, share product catalogues, generate quotations, collect advance payments and automate follow-ups, with an integrated CRM. It uses official Meta integrations, and businesses can keep their existing WhatsApp Business number. intellibookai.com ## Harmonian (Malaysia) URL: https://startup.talk/malaysia/harmonian/ Founders: Bryan Lum Bryan Lum grew up in Kuala Lumpur, went to Chinese-medium schools from SJKC Taman Connaught to Tsun Jin High School, and fixed his first computer as a boy because he was afraid of being scolded for breaking his sister's. He studied IT at Asia Pacific University, co-founded PicaPica, an Instagram photo printing kiosk business that reached Korea and Thailand, and still helps run Wilda Machinery, the power tool distributor his family started in 1996. Harmonian is his fourth idea. It began as a free AI wellness companion that speaks the way Malaysians do, switching between English, Bahasa Malaysia, Mandarin and Cantonese, and today most of its work is GydeX, an AI workplace coach built with PsyHome Asia that helps employees prepare for the conversations they keep avoiding. He builds every product himself, with AI as his team. We spoke about five products and zero paying customers, the chance meeting with a fellow Tsun Jin alumnus that changed the company's direction, and why psychologists are his partners, not his competitors. Key takeaways: - Harmonian builds AI wellness tools for Malaysia: the free Harmonian AI companion (also on WhatsApp), GydeX for the workplace, the IPC Super App with International Psychology Centre, and the EmoCall peer voice app. - GydeX, built with PsyHome Asia, reads an employee's Mind Report against seven EQ frameworks and coaches them in English, Bahasa Malaysia or Chinese. It has about 500 users across some ten organisations, with teams at Sony, Daikin and Perodua. - The person who needs support never pays: psychology centres and employers do. Harmonian is self-funded and still pre-revenue, and Bryan runs it alone with no employees. - His targets: a first RM30,000 in sales before 1 April 2027, then six new companies a month, a 2027 forecast of about RM168,000 and Singapore next. The Principal / Founder: - Bryan Lum (Founder) Based: Petaling Jaya, Selangor, Malaysia Grew up: Kuala Lumpur Studied: IT (mobile technology), Asia Pacific University Before: Co-founder, PicaPica (2017–2019); Wilda Machinery, family business Products: Harmonian AI, GydeX, IPC Super App, EmoCall, WildaBot Funding: Self-funded Website: harmonian.online LinkedIn: Bryan Lum (https://www.linkedin.com/in/bryan-lum-malaysia) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Lum: I grew up in Kuala Lumpur and went to Chinese-medium schools all the way: SJKC Taman Connaught for primary, then Tsun Jin High School from Form 1 to Form 6, where I sat the UEC. My family started Wilda Machinery, a power tool distributor, in 1996, when I was about five, and I still help run it today. As a kid, I loved playing on my sister's computer, until one day I broke it. I was so afraid of getting scolded that I tried to fix it myself, and it worked. That's probably where my love for fixing and building things started. In high school, I was in charge of the computer society and ran computer courses for both junior and senior students. Growing up with Mandarin, Cantonese, English and Malay around me is also a big reason I care so much about people being able to talk in their own language. Q: What first got you into building things? Lum: Computers. I studied IT at Asia Pacific University, specialising in mobile technology, and spent those years writing Java, C# and Android apps and fixing PCs. But building something real came with PicaPica: we built Instagram photo printing kiosks from scratch, both the hardware and the software. That's when I got hooked on building things people actually use, and I've been building ever since. Q: What was your first job, and what did it teach you? Lum: After university I co-founded PicaPica, an Instagram photo printing kiosk business, in 2017. I handled the hardware and software, procurement, customer support and events, and managed a team of technicians covering all of Malaysia. We took it to Korea and Thailand before I exited and left my partner running it. It taught me that a good product is only half the job; operations, people and support are the other half. Later, at our family business Wilda Machinery, I did everything from IT and e-commerce to walk-in retail sales. That's where I saw the people problems SMEs deal with every day. Q: How does your family's power tool business shape the way you build? Lum: Wilda Machinery is my real-world test lab. My family started it in 1996, and working there, from IT and e-commerce to serving walk-in customers, showed me the everyday people and operations problems SMEs face. It's also where I built WildaBot, a WhatsApp AI sales assistant first used at Wilda as "Xiao Wei" (小威), which answers enquiries using live product data. Building for our own business first keeps me honest: if a tool doesn't help a real shop on a busy day, it doesn't matter how clever it is. ### Part II: The road to Harmonian Q: What made you start Harmonian? Lum: I came into mental wellness from tech, not counselling. In 2023 I started building small wellness apps and reading a lot of psychology research to bridge the two worlds. What kept bothering me is that struggles, especially at work, are often invisible until it's too late. People don't speak up, and by the time anyone notices, the damage is done. Harmonian is actually my fourth idea; I pivoted three times before this one. The long-term dream is simple: phones gained cameras, watches gained blood-oxygen sensors, and I believe everyday devices will one day notice when someone is struggling, so help comes before a crisis, not after. We're building the brain first, then the body. Q: Why build mental health support in English, Bahasa Malaysia and Chinese? Lum: Because that's how Malaysians actually talk. Most of us switch between English, Bahasa Malaysia, Mandarin, Cantonese and Manglish in one conversation. When you're upset, you think in your own language, and being forced into a "proper" one adds friction exactly when you need it least. I speak English, Mandarin, Cantonese and Manglish myself, so this was never a feature to add later. GydeX's AI coach works in English, Bahasa Malaysia and Chinese. Q: How did Harmonian find its first users in Malaysia? Lum: Through partners, not ads. GydeX reached its first users through PsyHome Asia, a corporate psychology and training company. Employees meet GydeX in PsyHome's workshops and then use it to follow up afterwards. That's how we reached around 500 users across about ten organisations, including teams at companies like Sony, Daikin and Perodua. For the consumer app it has been slower and more personal: a free WhatsApp version people share with friends and family, booths at events like Cradle LIVE, and talks I give at high schools. Q: What is it like to build a free mental health platform as a young company? Lum: Honestly, it's a balancing act. I run Harmonian on my own, with no employees. I use AI for much of the work I'd otherwise hire for, and I only need people for face-to-face meetings with partner centres and for events. The hard part is that cloud and AI bills arrive every month, before revenue does. Free for users only works if someone else pays, so our model is built around partners: psychology centres and employers pay, not the person who needs support. Q: What is it like building as a solo founder, with AI as your team? Lum: I've run Harmonian on my own for about three years, with no employees. I write the code for every product myself, and I use AI the way other founders use a team. I work in a very structured way, writing a spec and a task list before I build, so the AI and I stay on track. The only things AI can't do for me are face-to-face meetings with partner centres and showing up at events. It keeps costs low, but it also means everything depends on one person's time and focus, so choosing what not to do matters a lot. ### Part III: The product Q: Why the name Harmonian? Lum: It comes from the word "harmony". I wanted everyone who uses it to become a Harmonian, the same way people from a country carry its name: Malaysia gives you Malaysians. Think of Harmony as the place, and everyone in it as a citizen, a Harmonian. It's meant to feel like belonging somewhere, not just using another app. Q: How does the AI companion support people day to day? Lum: Harmonian AI is a wellness companion you can message any time, in the app or on WhatsApp, when you've had a rough day and need to think out loud. It's there to listen, help you sort out your thoughts, and point you to real people when you need more. At work, GydeX goes further. After an employee gets their Mind Report from PsyHome, GydeX reads it against seven research-backed EQ frameworks, follows up, and helps them practise the conversation they've been avoiding, like telling the boss about a problem or giving honest feedback. It's support and practice, not therapy or diagnosis. Q: What is EmoCall and why make it anonymous? Lum: EmoCall is a voice app where you can talk through what you're feeling with another person. Anonymity matters because many people here still worry about being judged for struggling, and it's easier to be honest when nobody knows who you are. To be honest, EmoCall is launched but hasn't found its audience yet, so it's not where my focus is right now. Q: How does Harmonian work with psychologists and counselling centres? Lum: We don't hire therapists ourselves, on purpose. Harmonian is the technology; licensed psychologists and counselling centres are the clinical side. With International Psychology Centre we built the IPC Super App, which matches people to IPC's services and supports them between sessions. With PsyHome Asia, GydeX picks up after their assessment and keeps the coaching going. The centres do what they're best at, and we handle what happens after the session or the report, at scale. Q: How do you keep users safe when someone is in crisis? Lum: Our first rule: Harmonian is a wellness tool, not a medical one. It doesn't diagnose, and it's not a replacement for a psychologist. When someone is in real distress, the right move is to get them to a human: a professional through our partner centres, or crisis lines such as Talian HEAL 15555 or Befrienders KL. That's also why we partner with clinical teams instead of pretending AI can do everything. Q: What do you deliberately leave out of Harmonian? Lum: Diagnosis and assessment: our psychology partners do that, not us. Medical claims: we're wellness, and pretending otherwise is dangerous. An in-house team of therapists: we partner with centres instead. Private conversations for employers: GydeX is designed so HR sees team-level patterns, not what any one person said. And we don't charge the person who needs support; centres and employers pay. ### Part IV: Positioning Q: Who is Harmonian for, and what job does it do for them? Describe the person who gets the most value. Lum: Right now, mostly the people GydeX serves at work: young employees, Gen Z first-jobbers and new managers in high-pressure teams. The person who gets the most value is someone with a hard conversation they keep avoiding: a problem they should raise with their boss, feedback they need to give, or tension with a colleague. GydeX helps them understand themselves, practise what to say and walk in prepared. For HR, the job is seeing the bigger picture across teams early, instead of finding out in an exit interview. Q: Why is your focus now on helping employees speak up at work? Lum: Because that's where I see the biggest gap. In most companies, the information is there, the ideas are there, the warnings are there, but nobody speaks up. Someone knows a customer is unhappy, or spots a mistake that's wasting money, and stays quiet because they're afraid of being blamed. GydeX, which we built with PsyHome Asia, helps employees prepare for those conversations: they reflect, get a Mind Report, practise what to say, and walk in ready. HR sees team-level patterns, not private chats. Mental wellbeing is still part of it, but helping people speak up is a business problem every company understands. Q: How do you view your positioning relative to Wysa, Calm, Headspace, BetterHelp, and local counselling services? Is there another competitor you think of first? Lum: Calm and Headspace are great for meditation and sleep content. Wysa is an AI wellbeing chatbot, and BetterHelp is online therapy. Local counselling services aren't our competitors; they're our partners, and our model is built to give them more reach. Our angle is different: built for Malaysia's languages, sold through psychology centres, and focused on what happens after an assessment, at work and in daily life. The competitor I think about first is actually ChatGPT. It's free and can help anyone practise a conversation, so we have to prove a focused, locally built experience is worth paying for. In the workplace space, Naluri and Intellect are the names I watch. Q: What is your core moat in mental health support? Lum: Partnerships more than technology. Psychology and counselling centres sell Harmonian inside their own services, so they become our channel, and they bring the clinical credibility we deliberately don't build ourselves. Add local languages, and the learning that only comes from real use in Malaysian workplaces. I'll be honest: a moat is something you earn over time, and we're still building ours. Q: What are some of the use cases for Harmonian that are not obvious and can add a lot of value to users? List as many as possible. Lum: - Rehearsing how to tell your boss about a problem before you walk into the room - Preparing to ask for a raise, a transfer or more support - New managers practising their first difficult feedback conversation - Gen Z first-jobbers making sense of unwritten workplace rules - Repairing things after an argument with a colleague - Getting ready for a performance review, from either side of the table - Keeping a training workshop alive after the day ends, instead of forgetting it by Monday - Wording a message to a colleague in the language they prefer - Coping with a restructure, or a role that's changing because of AI - Following up between sessions with a partner counselling centre - Salespeople deciding how to flag an unhappy customer early instead of hiding it - HR spotting team-wide patterns early to plan the right training Q: What is the one detail in Harmonian you are proudest of? Lum: How GydeX reads a Mind Report. An employee uploads the report from PsyHome, and GydeX pulls out the EQ data, interprets it against seven research-backed EQ frameworks, and turns it into coaching in English, Bahasa Malaysia or Chinese. That's the moment a one-off assessment becomes something people actually use. ### Part V: Launch Q: What did you learn from your first launch? Lum: Honestly, that a launch is not the same as distribution. I put Harmonian on Product Hunt as a solo founder with no audience behind me. It got a couple of upvotes and never got featured. It was humbling, but useful. Product Hunt rewards founders who already have a community ready to show up on launch day, and I didn't have one yet. It also showed me that a global tech crowd isn't really my user. The people who need Harmonian are Malaysian employees and families, and they found us through partners like PsyHome Asia, not through a launch page. Now I build the channel first, then launch. Q: What did early users tell you that surprised you? Lum: I asked people what would make them trust an AI with their feelings, and the answers surprised me. One person said the AI wellness tools they'd tried felt generic and forgot everything from one chat to the next. Another suggested a summary you could bring to your therapist. And both wanted the AI to recognise when it's time to say, "You should talk to a real person." That surprised me most: people didn't want AI to replace humans. They wanted it to remember them, know its limits and get them to the right person. That's very close to how Harmonian works today, with psychologists and counselling centres as our partners, not our competitors. ### Part VI: The hard part Q: How many people use Harmonian and GydeX today? Lum: GydeX has around 500 users across about ten organisations, reached through PsyHome Asia. Our last working count was about 490 users, with roughly 5,000 coaching messages and 555 Mind Reports analysed. These are usage numbers, not paying customers; we're still pre-revenue. The consumer app and its WhatsApp version are still small and early. Q: What is a challenging time you had building Harmonian? Lum: Realising I had five products and zero paying customers. I'd built Harmonian AI, EmoCall, GydeX, WildaBot and the IPC Super App, each one real and shipped, but spreading myself that thin meant none got the focus it needed. At the same time, monthly cloud and AI costs were straining cash, and I had to be open with my partners about it. The lesson was focus. Today most of my energy goes into GydeX and getting our first RM30,000 in real sales. Q: Who believed in you before there was anything to show? What did they say or do? Lum: In 2025 I got talking to a stranger I ran into by chance. It turned out we'd both studied at Tsun Jin High School, though we never knew each other there. He owns PsyHome Asia, a counselling and training company. He gave GydeX a real home: his workshops, his corporate clients, and later the door to working with other counselling firms. That one chance meeting changed the direction of Harmonian. Q: How do you keep support free for users while building a sustainable company? Lum: The person who needs support doesn't pay; partners do. For GydeX, an employer pays PsyHome for training with GydeX included, and PsyHome shares revenue with us under our joint venture. For the IPC Super App, we share in what International Psychology Centre bills its clients. Next, new partner centres will pay an onboarding and licence fee to join the platform. Because the centres sell Harmonian inside their own services, I don't need a sales team, and using AI instead of hiring keeps costs low. ### Part VII: What it is all for Q: What would make you say Harmonian has succeeded? Lum: Short term, three things: our first RM30,000 in real sales, companies renewing because they see the value, and employees telling us they feel more ready to speak up. Long term, success is when struggling at work is no longer invisible: when help reaches people before a crisis, and everyday devices can notice someone needs support and point them to the right human. Q: What would Malaysia look like if everyone had someone to talk to, any time, in their own language? Lum: Fewer problems would stay hidden until it's too late. An employee would raise a concern while it's still small, a manager would hear about it before it costs a customer or a resignation, and nobody would have to translate their feelings into a language they don't think in. Counselling centres would reach far more people than their rooms can hold. And talking about how you feel would just be normal, not something you hide. Q: What would you tell a young Malaysian who wants to build something? Lum: Start by fixing what's in front of you. I started by repairing my sister's computer because I was scared of getting scolded. Then build something real and learn from it: my first company, PicaPica, taught me that operations matter as much as the product. Expect to pivot; Harmonian is my fourth idea. Learn to focus earlier than I did; I once had five products and zero paying customers. And find your channel before your launch day. I share this whenever I give talks at high schools. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Lum: Revenue first: our first RM30,000 in real sales before 1 April 2027, then a steady six new companies a month on GydeX's Team package. That's roughly 120 new employees joining each month. If we hit that, our forecast for 2027 is about RM168,000 in sales. That's a forecast, not a result. Alongside that, we want paid pilots where we measure whether employees feel more ready to speak up, and companies that renew. Q: And for the next 24 months? Lum: Scaling to about 25 new companies a month, which puts our 2028 sales forecast at about RM699,300, with around 1,500 employees using GydeX at any one time. That's still a tiny slice of Malaysia's 12.65 million employees. In the same window we want proven outcomes and renewals, more sales capacity beyond PsyHome, and to be ready to enter Singapore. We plan to raise a seed round once our first milestones are proven. Q: Where do you want Harmonian to be in five years? Lum: From Malaysia to Asia: Singapore first, then the wider region. And closer to the big vision: technology in everyday devices, phones first, that can notice when someone is struggling, so prevention happens before a crisis. Harmonian won't be just another wellness app. It will be the layer that connects people to the right support, with psychology partners in every market. Q: What are your hopes and fears for a post AGI world? Lum: My hope is that AI takes over the busywork and gives people more time for each other, and that everyone has someone to talk to, any time, in their own language. My fear is the opposite: that we start talking to machines instead of each other, and that jobs change faster than people can adjust, with no one helping them through it. That's why I believe the more work AI does, the more the human side of work matters. At Harmonian, AI's job is to get people back to people: a better conversation with their boss, or a real psychologist when they need one. About Harmonian: Harmonian builds AI wellness and workplace coaching tools from Petaling Jaya, Malaysia, in English, Bahasa Malaysia and Chinese. Its products include the free Harmonian AI companion (also on WhatsApp); GydeX, an AI workplace EQ coach built with and sold through PsyHome Asia; the IPC Super App with International Psychology Centre; and EmoCall, a peer support voice app. Harmonian is a wellness tool, not a medical one, and works with licensed psychologists and counselling centres. It is built by Bryan Lum. harmonian.online ## Clienter (India) URL: https://startup.talk/india/clienter/ Founders: Talagana Rajesh Talagana Rajesh grew up in Odisha, India, wanting to do something different from what everyone else was doing. Talagana Rajesh built tools in college that students still use, then turned down a job offer at PwC to start Webcros, a web development agency, at 20. Clienter came out of running that agency. Leads, projects, invoices and follow-ups were scattered across WhatsApp, spreadsheets and notes, and the tools built for Western markets did not fit the way Indian freelancers and agencies work. So Talagana Rajesh built one workspace: a lead CRM with a Kanban pipeline, project management, GST-ready invoicing with Razorpay payments, and a white-label client portal that clients open with a magic link. We spoke about 1,000 sign-ups and only 14 paying customers, a father who simply said go for it, and why every feature in Clienter exists because Talagana Rajesh hit a wall running an agency. Key takeaways: - Clienter is one workspace for Indian freelancers and small agencies: a lead CRM with a Kanban pipeline, project management, GST-ready invoicing with Razorpay payments, proposals, and a white-label client portal. - Every feature came from running Webcros, Talagana Rajesh's own web development agency, which still runs on Clienter every day. - About 1,000 people signed up, mostly from an Instagram audience of about 69,000 followers, but only around 14 pay. Their retention has been strong, and the lesson was about reaching the right people, not more people. - The targets: 100 paying users in 12 months, 500 plus in 24 months, then service businesses across South Asia, built without venture capital. The Principal / Founder: - Talagana Rajesh (Founder) Grew up: Odisha, India Also runs: Webcros, a web development agency (started December 2025) Before: Turned down a job offer at PwC to start Webcros Users: About 1,000 sign-ups, around 14 paying Website: clienter.co.in LinkedIn: Webcros (https://www.linkedin.com/company/webcros) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Talagana Rajesh: I grew up in Odisha, India. I was always curious about new things and good at studies, but what really defined me was wanting to do something unique — something different from what everyone else was doing. That drive eventually led me to turn down a job offer at PwC right out of college and start my own web development agency, Webcros, at 20. Q: What first got you into building things? Talagana Rajesh: I love building products that are genuinely useful to people. In college, I built several tools that students still use today. That feeling of solving a real problem with software and watching people actually rely on it — that is what got me hooked. I have always seen software as the most direct way to fix things that are broken or missing. Q: What was your first job, and what did it teach you? Talagana Rajesh: I never took a traditional first job. I had an offer from PwC but turned it down to start Webcros, my web development agency, in December 2025. That decision taught me everything traditional employment never could — how to find clients, price my work, deliver under pressure, and take full ownership of outcomes. When there is no salary coming in and it is all on you, you learn fast. ### Part II: The road to Clienter Q: What made you start Clienter? Talagana Rajesh: I was running Webcros and managing multiple clients at once — leads, projects, invoices, follow-ups — and everything was scattered across WhatsApp, spreadsheets, and random notes. I looked for a tool that could handle all of it in one place, especially one built for the way Indian freelancers and agencies actually work. Nothing fit. So I built it myself. Clienter started because I needed it for my own agency. Q: What did you see agencies and freelancers in India struggling with that convinced you to build one workspace? Talagana Rajesh: Most freelancers and small agencies in India manage clients through WhatsApp chats and Excel sheets. There is no proper system for tracking leads, sending professional invoices with GST, sharing project updates with clients, or even getting paid on time. The tools that exist — HoneyBook, Dubsado, Bonsai — are built for Western markets with pricing and features that do not fit Indian workflows. I saw that gap clearly because I was living it every day at Webcros. Q: How did Clienter grow from client profiles into CRM, projects, invoicing and a client portal? Talagana Rajesh: It grew out of my own needs at Webcros. First I needed a way to store client details and track leads, so I built a CRM with a Kanban pipeline. Then I kept losing track of project timelines, so I added project management. Clients would ask for invoices and I was making them manually in Canva, so I built GST-ready invoicing with Razorpay payments. Then clients wanted a place to see their project status without messaging me on WhatsApp, so I built a white-label client portal with magic-link access. Every feature exists because I hit a wall running my own agency. Q: What is it like to build Clienter as a developer and founder? Talagana Rajesh: It is the best and hardest thing at the same time. As a developer-founder, I can ship features fast without depending on anyone — I see a problem and I can build the fix that same day. But it also means I am doing everything: writing code, handling Webcros clients, managing support, thinking about marketing, and trying to grow. There is no team behind Clienter yet — it is just me building the product and my teammate Ehsan helping with development at Webcros. The upside is that every decision is fast and every feature comes from real experience. ### Part III: The product Q: Why the name Clienter? Talagana Rajesh: Simple — it is the tool that puts your clients at the center. Client + er. Everything in Clienter revolves around the client: their profile, their projects, their invoices, their portal. The name had to reflect that. Q: Why build GST-ready invoicing into the product? Talagana Rajesh: Because if you are a freelancer or agency in India, GST invoicing is not optional — it is a legal requirement. And yet most client management tools either ignore it or bolt it on as an afterthought. I was generating GST invoices manually for Webcros clients, calculating CGST and SGST by hand, and it was painful. Building it natively into Clienter with Razorpay integration means you create a proper GST invoice and collect payment in one flow. That is table stakes for any Indian service business. Q: What does the white-label client portal change for agencies? Talagana Rajesh: It changes how clients perceive you. Before the portal, my clients at Webcros would message me on WhatsApp at all hours asking for project updates. With the white-label portal, each client gets their own branded login with a magic link — no passwords — and they can see their project status, invoices, and files anytime. It makes a solo freelancer or small agency look and feel like a professional operation. The white-labeling means it shows your brand, not Clienter’s. Q: How do verified reviews work, and why add them? Talagana Rajesh: Verified reviews are tied to actual completed projects inside Clienter. A client can only leave a review for a project they were part of, so the review is genuine and verifiable. I added them because trust is the biggest barrier for freelancers trying to land new clients. Having verified reviews that are tied to real work — not fake testimonials on a website — gives freelancers and agencies credible social proof. Q: What role do the free tools play in growing Clienter? Talagana Rajesh: The free tools are the top of the funnel. They solve a specific small problem — like generating an invoice or creating a proposal template — and bring people to Clienter who are already in the mindset of managing clients professionally. Once they see the free tool works well, they naturally explore the full platform. It is product-led growth: let the product sell itself by being genuinely useful before asking anyone to pay. Q: What do you deliberately leave out of Clienter? Talagana Rajesh: I deliberately keep Clienter focused on client management for service businesses. I do not try to be a project management tool like Asana or a full accounting suite like Zoho Books. No complex workflows, no enterprise features, no trying to serve everyone. If a freelancer or small agency manages clients and wants one clean workspace for leads, projects, invoices, and a client portal — that is Clienter. Everything else is noise I leave out on purpose. ### Part IV: Positioning Q: Who is Clienter for, and what job does it do for them? Describe the person who gets the most value. Talagana Rajesh: Clienter is for established freelancers with 5 or more active clients and small agency owners with 2 to 10 person teams. The person who gets the most value is someone who has outgrown WhatsApp and spreadsheets but does not need — or want to pay for — a bloated enterprise CRM. They are juggling multiple clients, losing track of follow-ups, sending invoices manually, and spending hours on admin instead of billable work. Clienter gives them one place to manage all of it so they can focus on delivery and growth. Q: How do you view your positioning relative to HoneyBook, Bonsai, Zoho, Dubsado and Notion? Is there another competitor you think of first? Talagana Rajesh: HoneyBook, Bonsai, and Dubsado are built for the US market. Their pricing, payment gateways, and invoicing do not work well for Indian freelancers and agencies. Zoho is powerful but it is an enterprise suite with a steep learning curve. Notion is flexible but it is a blank canvas, not a client management tool. Clienter is purpose-built for Indian service businesses. GST invoicing, Razorpay payments, a white-label client portal, and a simple interface that does not need a tutorial. I am not trying to compete with all of them. I am building the tool that none of them built for this market. Q: What is your core moat in client management for agencies and freelancers? Talagana Rajesh: I use Clienter every day to run Webcros. That is the moat. I am not building from a boardroom or reading market research. I am a practitioner who feels the pain firsthand and ships fixes the same week. The product is built by someone who lives the problem, and that shows in every feature. Competitors can copy features, but they cannot copy the context of actually running an agency in India with this tool as your daily operating system. Q: What are some of the use cases for Clienter that are not obvious and can add a lot of value to users? List as many as possible. Talagana Rajesh: A few that people do not expect: Using the lead CRM Kanban to track referral sources and see which channel brings the best clients. Using the client portal as a lightweight project dashboard that clients can share with their own stakeholders. Using proposals with e-signatures to close deals faster without back-and-forth PDFs. Using Clienter as a single source of truth for all payment tracking instead of maintaining separate spreadsheets. Freelancers also use the intake forms to qualify leads before getting on a call, which saves hours every week. Q: What is the one detail in Clienter you are proudest of? Talagana Rajesh: The white-label client portal with passwordless magic-link authentication. Each client gets a unique JWT session completely isolated from Supabase Auth. No passwords to remember, no shared login pages. Your client clicks a link, lands on a portal branded with your agency name, and sees everything about their project. I spent a lot of time getting the security right with AES-256-GCM encryption, and I am proud that it works seamlessly while being genuinely secure under the hood. ### Part V: Launch Q: What did you learn from your launch? Talagana Rajesh: The biggest lesson was that launching is not the finish line, it is the starting line. Product Hunt gave Clienter visibility and brought in a wave of signups, but most of those users were not the target audience. They were curious builders and Product Hunt regulars, not freelancers drowning in client chaos. I learned that the real work starts after launch: finding the right people, not just more people. Q: Which feedback from the launch surprised you? Talagana Rajesh: People loved the client portal concept. I did not expect that to be the standout feature. I thought the CRM and invoicing would get the most attention, but users kept saying the white-label portal was what made Clienter different from everything else they had tried. It validated that the portal was not just a nice-to-have, it was the core differentiator. ### Part VI: The hard part Q: How many people use Clienter today? Talagana Rajesh: About 1,000 signups total, but I will be honest: only around 14 are paying customers. Most of the 1,000 came from my Instagram audience, which at the time was mostly students and beginners who were curious but never intended to use the tool seriously. The paying users are a different story. They are real freelancers and agency owners who rely on it daily, and their retention has been strong. The problem was never the product. It was reaching the right audience. Q: What is a challenging time you had building Clienter? Talagana Rajesh: Realizing that 1,000 users meant almost nothing because they were the wrong people. I had spent months building features, polishing the product, and promoting it to my Instagram audience. Then I looked at the numbers and saw that only 14 were paying. That was a hard moment. It forced me to rethink everything about who I was building for and how I was reaching them. The product was solid. My go-to-market was completely wrong. Q: Who believed in you before there was anything to show? What did they say or do? Talagana Rajesh: My father. He is not into tech at all. He does not fully understand what I build or how SaaS works. But when I told him I wanted to reject the PwC offer and start my own thing, he did not push back. He did not pressure me to take the safe route. He just said go for it. That kind of unconditional belief, from someone who does not even fully understand what you are doing, means more than any mentor's advice. I did not have a traditional mentor. I taught myself everything. But my father's trust gave me the freedom to try. Q: How do you reach freelancers and small agencies across India? Talagana Rajesh: Right now it is mostly through my personal Instagram, which has about 69,000 followers, and I am starting to push harder on LinkedIn where the actual agency owners and established freelancers hang out. I also use free tools on the Clienter website as a top-of-funnel strategy. The content pivot is a big part of this. I am shifting from beginner-focused tech tips to content that speaks directly to people running agencies and managing real clients. Word of mouth from the paying users has also been surprisingly effective. ### Part VII: What it is all for Q: What would make you say Clienter has succeeded? Talagana Rajesh: When freelancers and agency owners across India tell me they cannot imagine going back to managing clients without Clienter. Not vanity metrics like user count, but real dependency. When the product is so embedded in their daily workflow that it becomes invisible, like electricity. That, and building a sustainable business out of it. I want Clienter to be profitable and growing on its own, not dependent on my Instagram presence or any single channel. Q: How will AI change the way freelancers and agencies run their businesses? Talagana Rajesh: AI will automate the repetitive parts that eat up a freelancer's day. Writing proposals, following up with leads, generating invoice reminders, drafting project updates. The agencies that adopt AI will be able to serve more clients with smaller teams. But AI will not replace the human judgment that clients pay for. Strategy, creative direction, and relationship building will stay human. The freelancers who win will be the ones who use AI as a tool to work faster, not the ones who resist it. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Talagana Rajesh: In the next 12 months I want to get to 100 paying users and reach a point where Clienter revenue covers its own costs. The focus is entirely on finding and converting the right audience, not growing vanity numbers. I also want to ship team collaboration features and deeper Razorpay integration. On the content side, I am building a LinkedIn presence specifically targeted at agency owners and established freelancers. Q: And for the next 24 months? Talagana Rajesh: By 24 months, I want Clienter to be the default client management tool for Indian web development agencies and freelancers. That means 500 plus paying users, a small team to handle support and growth, and enough revenue to reinvest into the product full time without needing Webcros agency income to sustain it. I also want to have expanded into adjacent service verticals beyond web development. Q: Where do you want Clienter to be in five years? Talagana Rajesh: In five years I want Clienter to be the go-to platform for service businesses across South Asia, not just India. I want it to be the tool that every freelancer and agency owner recommends to their peers. A profitable, bootstrapped product with a small dedicated team, growing through word of mouth because the product genuinely makes people's work lives easier. I do not want to raise venture capital and chase unicorn metrics. I want to build a real, sustainable software business. Q: What are your hopes and fears for a post AGI world? Talagana Rajesh: My hope is that AGI makes it possible for solo founders and tiny teams to build products that used to require 50-person engineering teams. That is exciting because it levels the playing field for builders in places like India. My fear is that it moves so fast that people stop valuing craftsmanship and human judgment. The best software is not just functional, it reflects real understanding of the user's world. I hope we do not lose that in the rush to automate everything. About Clienter: Clienter is client management software for freelancers and small agencies in India, built by Talagana Rajesh, founder of the web development agency Webcros. It brings client profiles, a lead CRM with a Kanban pipeline, project management, GST-ready invoicing with Razorpay payments, proposals with e-signatures, intake forms, verified reviews and a white-label client portal with passwordless magic-link access into one workspace, with a mobile app and free tools. clienter.co.in ## ATS Engine (UK) URL: https://startup.talk/uk/atsengine/ Founders: Dennis Nowland Dennis Nowland grew up in Conowingo, Maryland, where hunting, fishing and camping filled the days outside school, and moved to Scotland at 17. The first jobs were physical: loading heavy goods vehicles in a warehouse, then building heavy-duty industrial machinery as a mechanical fitter. A motorcycle accident ended that work, so Nowland moved into computing and software engineering, won Student of the Year at Fife College, and went on to spend more than 20 years as a software engineer and delivery lead. ATS Engine began after a redundancy. A head injury makes things hard to remember, the job-search spreadsheets were getting confusing, and Nowland discovered how applicant tracking systems quietly filter out CVs before a person ever sees them. The tool built to cope with that turned into a platform for both sides of hiring: job seekers see how well they match a role before applying, and recruiters get one desk for candidates, vacancies, compliance and what needs attention today. We spoke about a test batch of 20 CVs, why the final decision must stay human, and the competitor Nowland worries about most: the spreadsheet. Key takeaways: - ATS Engine serves both sides of hiring: job seekers see how well their CV matches a role before applying and can tailor a CV and cover letter in about a minute, while recruiters get candidates, vacancies, compliance and today's priorities in one place. - It started as a tool for Nowland's own job search. A test batch of 20 CVs brought 8 interviews and 1 offer. - The platform deliberately leaves the hiring decision with people and does not reduce a candidate to a single score. - ATS Engine is early stage and Nowland will not quote user numbers yet. The next 12 months are about proving three things: people want it, people pay for it, and people stay. The Principal / Founder: - Dennis Nowland (Founder) Based: Fife, Scotland Grew up: Conowingo, Maryland, USA Studied: Computing and software engineering, Fife College Before: Mechanical fitter; 20+ years as a software engineer and delivery lead Company: DennisNowland Ltd Website: www.ats-engine.com (https://www.ats-engine.com/) LinkedIn: Dennis Nowland (https://www.linkedin.com/in/dennis-nowland/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Nowland: I grew up in a town called Conowingo, Maryland, until I was 17, when I moved to Scotland for family reasons. To me, I was born and raised in the country, so hunting, fishing and camping were my life outside of school and TBH, sometimes, when I should have been in school lol. I had a great family life around Chesapeake City and Conowingo. Q: What first got you into building things? Nowland: I have always had an engineering mindset. I started out as a mechanical fitter building heavy-duty industrial machinery. After my motorcycle accident, I was no longer able to do the physical side, so I moved into computing and then software engineering, where I won the Student of the Year (valedictorian) award at Fife College. Since my accident, it just seems to come naturally to me. Q: What was your first job, and what did it teach you? Nowland: My very first job after leaving school was in a warehouse loading heavy goods vehicles, first by hand, then I got my forklift licence. Then later I became the shunter with my own unit to traffic the trailers in and out of loading. It taught me to be proud of what I had, to still live life, to enjoy the special times and not to take things for granted. ### Part II: The road to ATS Engine Q: What made you start ATS Engine? Nowland: I started developing ATS Engine when I was made redundant from my previous employment. With my head injury, I find it hard to remember things, and spreadsheets can get too confusing. For the job seeker side of ATS Engine: It all started when I was using ChatGPT to keep a running list of my submissions. Once I started, I found out that if you submit to the same role more than a couple of times, you can be excluded from the job with a standard email so you don't know why. After finding this out, I went investigating and found out about ATS and how it filters out CVs, so a lot don't even get seen by a human. This sent me investigating again into what was required to pass ATS checks, and I found there were a lot of gaps in the job search process: tasks that either never got done or, if they did, took hours, for example tailoring CVs and cover letters to the given role. Using ATS Engine, that job takes about one minute. As for the recruiter side: While searching I would get calls from recruiters but then I started wondering how, who and when and kept finding gaps between a job seeker and the recruiter. So this sent me on another investigation, and I found that there were so many different areas that could make my life easier, so I could be prepared for when recruiters did ask for anything like paperwork. So basically, it started out as a tool to assist me with my disabilities, but then I realised the potential to help others to land a new role using what I had learned and in the end the formula I used in the app works, as I sent 20 CVs away as a test set and received 8 interviews and 1 offer, which I accepted, since it cost money to continue to try and get ATS Engine known, so people can see for themselves the benefits it brings. Q: What did 20 years as a software engineer and delivery lead teach you about how hiring goes wrong? Nowland: After more than 20 years as a software engineer and delivery lead, I learned that hiring often goes wrong before a human ever gets the chance to make a decision. From the employer side, I saw how difficult it can be to identify the right people when there are large numbers of applications, inconsistent information and processes that rely heavily on systems and keywords. From the candidate side, I experienced the other half of the problem: people can be highly capable and have the right experience, but still struggle to get noticed because their CV doesn't communicate that experience in the way the recruitment process expects. I also learned that the hiring process is often fragmented. Job seekers are trying to understand adverts, tailor CVs, track applications and prepare for interviews, while recruiters are trying to manage candidates, compliance, documentation and communication. The technology exists to make much of this easier, but too often the technology creates another layer of complexity rather than removing it. That was one of the biggest lessons behind ATS Engine: Hiring shouldn't be about who is best at navigating the process. It should be about making it easier for the right person and the right opportunity to find each other. My experience in software engineering and delivery taught me to look at the entire process, identify where things break down, remove unnecessary work and use technology to support people rather than make their lives more complicated. That is the approach I have taken with ATS Engine. Q: Why build one system for both recruiters and job seekers? Nowland: To bridge the gaps that I found in my search and how I could make my day as stress-free as possible and as easy as possible. With all that I learned from my sleepless nights and worrying, I wanted to try to remove as much of that as possible, so I investigated the natural gaps that were there between the two and thought about how easy it could be if I could bring it all into one place, and that is how ATS Engine started and now covers every major gap recruiters have, with the help of clients of ATS Engine and their feedback. Q: How do you build and run ATS Engine alongside your role as a Scrum Master? Nowland: With a lot of hard work and dedication. I truly believe that ATS Engine can help many people out of any rut they are in, either by switching roles or by finding a new role if they are out of work. I also believe that if recruiters have the right tools, then they can help many more people out of their horrible position. This is one of the reasons the Jobster, the AI assistant, who also answers to Jobbie, as she removes a lot of the daily crap from your day. ### Part III: The product Q: Why the name ATS Engine? Nowland: I picked the name ATS Engine mainly because it is a more modern version of the original ATS: ATS Engine allows multiple CVs per user and uses all of them to generate the CVs and cover letters, so the word Engine was added because it ploughs through like a steam engine. Q: How does the intelligent candidate matching work? Nowland: Our intelligent candidate matching is designed to look beyond simple keyword matching. It considers the relationship between the candidate's experience, skills, evidence and the requirements of the role, helping identify how well the overall profile aligns with the opportunity. Rather than simply asking, “Does this CV contain the right words?”, ATS Engine looks at the wider picture and highlights where there is a strong match, where evidence is weaker, and where attention may be needed. The exact matching methodology is part of the ATS Engine technology, so we don't publish the underlying formula or weighting. The important thing for users is the outcome: a clearer, more meaningful view of candidate-to-role fit without requiring recruiters or job seekers to manually work through everything themselves. Q: How does ATS Engine tell a recruiter what needs them today? Nowland: ATS Engine is designed to help recruiters see what needs their attention today, rather than making them work through the entire recruitment process to find it. It brings together the things that require action and highlights the areas that may need attention — whether that's a candidate, an application, compliance information, an advert or something else within the recruitment workflow. Instead of simply showing recruiters more information, ATS Engine helps turn that information into clear priorities and actionable next steps. The aim is simple: Open ATS Engine and know what needs your attention, why it matters, and what you can do about it. Exactly how ATS Engine determines and prioritises those actions is part of the intelligence behind the platform, so we don't give away the underlying methodology. Q: What do job seekers learn from seeing their match before they apply? Nowland: Seeing their match before applying gives job seekers something they often don't have: a clearer picture of how well their CV aligns with the role before they invest their time in the application. It can help them understand where their experience and skills line up with what the employer is looking for, and where there may be areas they need to consider before applying. That means they can make a more informed decision about whether a role is worth pursuing, rather than simply applying and hoping for the best. And where there is an opportunity to improve their application, ATS Engine can help them understand what needs attention before they press submit. The aim isn't to tell someone “don't apply.” It's to help them apply with a better understanding of their position and a stronger application. Q: How do you handle compliance for recruitment agencies? Nowland: ATS Engine helps recruitment agencies keep compliance information organised, visible and easier to manage. Instead of relying on spreadsheets, folders and reminders to keep track of candidate documentation, ATS Engine brings the relevant information together so recruiters can see what is in place, what may be missing and what needs attention. The platform also provides visibility into areas such as what lapses next and overall compliance coverage, helping recruiters identify potential issues before they become a problem. The aim isn't to replace an agency's compliance process or make legal decisions for them. It is to give recruiters better visibility, organisation and control over the information they need to manage. And, as with the rest of ATS Engine, the detailed intelligence and methodology behind how those priorities are identified remains part of the platform's technology. Q: What do you deliberately leave out of ATS Engine? Nowland: There are some things we deliberately don't want ATS Engine to do. We don't want it to make the final hiring decision for a recruiter, or tell someone who they should hire. The platform is there to support better decisions, not replace human judgement. We don't want to make candidates into a score and reduce a person to a number. Matching is there to provide insight and help people understand fit, not to decide someone's value. We also don't deliberately hide information behind complicated systems or make users work through unnecessary steps simply because the technology allows us to. And we don't pretend to know something when the information isn't available. Where information is missing, the platform should make that clear rather than inventing an answer. Most importantly, we deliberately leave the human decision with the human. ATS Engine is designed to reduce the administration, highlight what matters and provide better information — so recruiters and job seekers can spend more time making good decisions and less time fighting the process. The technology should support people, not replace them. ### Part IV: Positioning Q: Who is ATS Engine for, and what job does it do for them? Describe the person who gets the most value. Nowland: ATS Engine is built for both sides of the hiring process — job seekers and recruiters — but the people who get the most value are those who want to spend less time fighting the process and more time making better decisions. For Job Seekers The person who gets the most value is someone who is serious about finding their next role and wants more control over their job search. They may be applying for multiple positions, struggling to understand why applications aren't progressing, or simply spending too much time tailoring CVs, writing cover letters and keeping track of everything. ATS Engine helps them understand which opportunities are worth pursuing, how well their CV matches a role, what may need attention before applying, and what they should focus on next. It turns a job search from a collection of applications into something they can actually manage. For Recruiters The greatest value comes to recruiters and recruitment agencies dealing with multiple candidates, vacancies, documents and compliance requirements. They need to know what needs their attention, which candidates are relevant, what information is missing, what is approaching expiry and where potential issues need addressing — without having to search through multiple systems and spreadsheets. ATS Engine brings that information together and helps turn it into clear actions. The Person Who Gets the Most Value Ultimately, ATS Engine is for someone who thinks: "There has to be an easier way of doing this." Whether they're a job seeker trying to get their CV seen or a recruiter trying to manage candidates and compliance, ATS Engine is designed to reduce the unnecessary work, highlight what matters and give them a clearer view of what to do next. It isn't about replacing people. It's about giving people better information, better organisation and better tools to make the decisions that still belong to them. Q: How do you view your positioning relative to Bullhorn, Vincere, Recruit CRM, Greenhouse and Teamtailor? Is there another competitor you think of first? Nowland: ATS Engine isn't trying to compete with Bullhorn, Vincere, Recruit CRM, Greenhouse or Teamtailor by simply becoming another large Applicant Tracking System. Those platforms are established recruitment systems, and they do many things extremely well. Our positioning is different. ATS Engine is focused on making the hiring process more intelligent and actionable for both sides of the process. For the job seeker, ATS Engine helps answer questions that traditional recruitment systems generally aren't designed to answer: - Is this role actually a good match for me? - How does my CV compare with what the employer is looking for? - What should I improve before I apply? - Which applications need my attention? - What should I do next? - How can I manage my entire job search more effectively? For the recruiter, ATS Engine focuses on questions such as: - What needs my attention today? - Which candidates are the strongest matches? - What information is missing? - What compliance is approaching expiry? - Which adverts may need attention? - Where are potential problems developing? So rather than positioning ATS Engine as "another ATS", I see it as a layer of intelligence and action across the recruitment journey. The established competitors are primarily systems for managing recruitment. ATS Engine is designed to help people understand recruitment and act on it. That distinction is important. Where I See the Competitive Opportunity Bullhorn, Vincere and Recruit CRM are particularly strong in recruitment agency and staffing workflows. Greenhouse and Teamtailor are strong in employer hiring and applicant management. ATS Engine can learn from all of those categories without trying to replicate everything they do. Our opportunity is to focus on the gaps between information and action — helping people understand what the information means and what deserves attention next. And importantly, ATS Engine is designed around both sides of the hiring relationship. That gives us a different perspective from a system designed primarily around the employer. The Competitor I Think About First If I had to name one competitor category rather than one individual company, it would actually be: the spreadsheet, inbox and disconnected collection of tools that people are currently using to manage recruitment and job searching. That's the competition I care about most. A recruiter might have an ATS, spreadsheets, email, document folders, calendars and reminders. A job seeker might have spreadsheets, Word documents, email, job boards, ChatGPT and folders full of different CV versions. ATS Engine is trying to reduce that fragmentation. So I don't want ATS Engine to win because it has more buttons than another ATS. I want it to win because someone opens it and thinks: "Finally — I can see what matters, I understand what I need to do, and I don't have to work everything out myself." That is the position I believe gives ATS Engine its own space in the market. Q: What is your core moat in recruitment software? Nowland: I think our core moat is not one individual feature. It is the combination of the intelligence, the workflow and the perspective we've built around the entire hiring journey. Most recruitment software is built primarily around the recruiter or employer workflow. ATS Engine was built from both directions. On the job seeker side, it understands the journey from finding an opportunity, assessing the match, preparing the application, applying, tracking the application and knowing what to do next. On the recruiter side, it looks at candidates, matching, compliance, adverts, priorities and what needs attention. The moat comes from bringing those perspectives together and building intelligence around the gaps between them. There is also a significant amount of underlying work that users don't necessarily see — how information is assessed, how priorities are identified, how matching is interpreted and how the platform turns large amounts of recruitment information into something actionable. I deliberately don't want to describe the underlying formulas or methodology publicly. That's part of the intellectual property we are continuing to develop. But I also think there is another important part of the moat: the data and learning generated by the way people actually use the platform. As ATS Engine grows, we can learn which problems job seekers encounter, which areas recruiters struggle with, what information matters, what causes applications to succeed or fail, and where people need the most help. That creates an opportunity for the platform to become increasingly useful over time. So I wouldn't describe our moat as: "We have a feature nobody else has." I'd describe it as: "We are building an intelligence layer around the hiring journey that understands both sides of the process and turns recruitment information into actionable decisions." The technology can be copied. Individual features can be copied. But building the product, intelligence, workflows, data, customer understanding and two-sided perspective together is considerably harder. And that is where I believe ATS Engine can build a lasting advantage. Q: What are some of the use cases for ATS Engine that are not obvious and can add a lot of value to users? List as many as possible. Nowland: One of the interesting things about ATS Engine is that its value isn't limited to the obvious "apply for jobs" or "manage candidates" use cases. Because it brings together matching, applications, documents, intelligence, compliance and actions, there are a number of less obvious ways it can help. Job Seeker Use Cases 1. Know whether a job is worth applying for Instead of applying to everything, a job seeker can assess the opportunity before investing time in an application. 2. Identify why a strong candidate may be getting overlooked A candidate may have the experience but fail to communicate it effectively in their CV. ATS Engine can highlight areas that need attention. 3. Decide between two similar vacancies A job seeker can compare opportunities based on their fit rather than simply choosing whichever advert looks most attractive. 4. Stop wasting time on poor-fit applications A poor match can be identified before the candidate spends hours tailoring documents. 5. Prioritise the job search When someone has several applications underway, ATS Engine can help identify which opportunities deserve attention first. 6. Rescue an application that has gone quiet Instead of immediately assuming rejection, the candidate can identify whether an application needs a follow-up or another action. 7. Know when to stop chasing an application Not every application deserves endless attention. Helping someone recognise when to move on is valuable too. 8. Maintain an application memory Candidates don't have to remember every application, CV version, conversation or next action themselves. 9. Remember what CV was used This becomes particularly useful when someone has multiple versions of their CV. 10. Understand their own job-search performance Over time, the candidate can see what types of roles are generating interest and which aren't. 11. Identify patterns in unsuccessful applications Instead of treating every rejection as an isolated event, recurring issues can become visible. 12. Build evidence from previous work Achievements and examples from previous roles can be organised so the candidate doesn't have to reinvent them for every application. 13. Prepare for interviews from their own evidence Previous achievements can become useful preparation material rather than forgotten information buried in an old CV. 14. Quickly create role-specific application material Tailoring can become a repeatable process rather than a major task for every application. 15. Keep application documents consistent The CV, cover letter and supporting evidence can be aligned around the same role requirements. Q: What is the one detail in ATS Engine you are proudest of? Nowland: The detail I'm probably proudest of is the way ATS Engine turns a complicated recruitment process into something that tells you what matters and what to do next. It sounds simple, but getting there isn't. When I started building ATS Engine, I wasn't trying to create another system full of screens, forms and information. I wanted to solve a problem I was experiencing myself: there was too much to remember, too many things to track, and too many parts of the job search that were easy to miss. That thinking carried through into the platform. For a job seeker, it isn't just about saying "your CV matches this job." It's about helping them understand the opportunity, identify what may need attention, prepare a better application and keep track of what happens afterwards. For a recruiter, it isn't just about storing candidate information. It's about helping them see what needs their attention today, what may be approaching a problem and where they should focus. So the detail I'm proudest of isn't necessarily something you can point to on one screen. It's the underlying philosophy: Don't just give people more information. Help them understand what that information means and what they can do with it. That came directly from building ATS Engine to solve problems I was experiencing myself — and I think that personal origin is still visible throughout the product. ### Part V: Launch Q: What did you learn from your launch? Nowland: Launching ATS Engine on Product Hunt taught me that getting a product in front of people is very different from getting people to understand why they need it. The launch gave me a much clearer view of the difference between building something useful and communicating its value quickly. You can have a lot of functionality, but people need to understand the problem you solve almost immediately. It also reinforced that the launch itself is only the beginning. Visibility is useful, but what matters afterwards is what you learn from the people who look at the product, click through, ask questions, sign up, or decide not to continue. One of the biggest lessons was that the messaging needs to be focused on the user's problem rather than the technology behind the product. People don't necessarily care how intelligent the system is — they care whether it can save them time, remove frustration, help them make better decisions, or improve their chances of getting the outcome they want. It also showed me the importance of having the wider journey ready before launching: the landing page, product explanation, walkthroughs, trial, pricing, tracking and follow-up all need to work together. Most importantly, Product Hunt reinforced something I have learned throughout building ATS Engine: listen to what people actually struggle with and keep improving the product around those problems. The launch wasn't just about gaining exposure. It was another stage of validating the idea, learning how people respond to it, and understanding what needs to improve next. ### Part VI: The hard part Q: How many people use ATS Engine today? Nowland: ATS Engine is still in its early stages, so I don't want to give a number that makes the product look further along than it is. We currently have an early user and subscriber base, with both job seekers and recruiters starting to use and explore the platform. We're deliberately focusing at this stage on learning from those users, improving the product and making sure the experience genuinely solves the problems we've built it to address. For me, the important number isn't simply how many people have registered. It's how many people are finding ATS Engine useful enough to come back, use it as part of their recruitment or job search process, and ultimately recommend it to someone else. We're building that user base steadily rather than chasing numbers for the sake of numbers. Q: What is a challenging time you had building ATS Engine? Nowland: One of the most challenging periods building ATS Engine was probably the point where it stopped being a personal tool and started becoming a real product. The original idea came from solving problems I was experiencing myself. I needed a better way to keep track of applications, understand where I stood with each opportunity, tailor CVs and documents, and remember what needed to happen next. Once I realised that these were problems other people were experiencing too, the scale of what I was trying to build changed completely. Suddenly it wasn't enough for something to work for me. It needed to work reliably for other job seekers and recruiters, cope with different situations, present information clearly and be simple enough that people didn't need to understand how everything worked underneath. There have been plenty of technical challenges along the way — databases, integrations, queues, matching, document generation and getting different parts of the platform to work together. But the harder challenge has often been deciding what not to build and keeping the product focused on solving real problems rather than simply adding more features. There were certainly moments when it would have been easier to stop or simplify the ambition. But each time I came back to the reason I started building it in the first place. ATS Engine began because I needed something that made the process easier for me. The challenge has been turning that personal solution into something that can genuinely make the process easier for other people too. That is still the challenge — and, ultimately, what keeps me building it. Q: Who believed in you before there was anything to show? What did they say or do? Nowland: Before there was really anything to show, the person who believed in me most was someone who could see that I was trying to solve a real problem rather than simply trying to build a business. At that stage, ATS Engine was mostly an idea, a lot of notes, experiments and many hours of work. There wasn't a polished product that I could demonstrate or a customer base that proved the idea would work. What mattered was having someone who didn't dismiss the idea as just another project. They encouraged me to keep working on it and, importantly, gave me the space to see where it could go. I think that's what believing in someone really means at that stage. It's not necessarily having all the answers or knowing that they will succeed. It's saying, in effect, “Keep going. You might be onto something.” That encouragement mattered because there were plenty of points where I could have decided that building ATS Engine was too much work or that the idea wasn't going anywhere. Today there is something tangible to show for it. But I haven't forgotten that, at the beginning, there wasn't. Q: How do you win recruitment agencies that are locked into older systems? Nowland: I use marketing via email, LinkedIn, Facebook and other ad sites to get the word out, but that is expensive. I am handling it all myself via the CRM app I built that works with ATS Engine. ### Part VII: What it is all for Q: What would make you say ATS Engine has succeeded? Nowland: I would say ATS Engine has succeeded when it is genuinely making the hiring process better for the people using it. For a job seeker, that might mean helping someone understand whether a role is worth applying for, giving them a stronger application, helping them stay organised and ultimately helping them get the right job. For a recruiter, it might mean making it easier to identify the right candidates, know what needs attention, keep on top of compliance and spend less time dealing with information and administration. Of course, I want ATS Engine to become a successful business. It needs customers, sustainable revenue and the ability to keep developing. But those are measures of the business rather than the reason I started it. The real measure of success would be hearing someone say, “ATS Engine made this process easier for me, and I don't know how I would have managed it without it.” If enough job seekers and recruiters genuinely feel that way, and ATS Engine becomes something people rely on rather than simply something they try, I would consider that success. Ultimately, I want ATS Engine to help close the gap between the right person and the right opportunity — and make the journey between the two more intelligent, organised and human. Q: What will hiring look like when AI agents apply for jobs and screen candidates? Nowland: I think hiring is going to change dramatically when AI agents are routinely applying for jobs on behalf of candidates and screening candidates on behalf of employers. We could reach a point where the first interaction isn't necessarily between a person and a company. A candidate's AI agent could understand their experience, identify suitable opportunities, tailor their application and potentially manage much of the application process. On the other side, an employer's AI could analyse applications, identify potential matches, check evidence and prioritise candidates for human review. That creates an interesting problem. If both sides are using AI, recruitment could become an arms race where candidates optimise applications for employer systems while employers optimise systems to detect and filter AI-generated applications. That wouldn't necessarily make hiring better. I think the opportunity is to move away from simply asking “Who can produce the best application?” and towards “Who is genuinely the best fit, and what evidence supports that?” AI agents should be able to help both sides understand that more effectively. A candidate should be able to understand why they match a role and where their evidence is weak. A recruiter should be able to understand why a candidate has been identified as a potential match and what still needs human assessment. I also think the human element becomes more important, not less. AI can help discover, compare, organise and prioritise, but hiring decisions involve judgement, context, motivation, culture and conversations that shouldn't simply be reduced to a score. That is one reason I believe ATS Engine has an interesting role to play in the future. The opportunity isn't just to automate today's recruitment process. It is to help both sides navigate a recruitment environment where AI is increasingly involved, while keeping the human decision at the centre. The future of hiring shouldn't be humans versus AI. It should be humans using AI to make better hiring decisions. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Nowland: Over the next 12 months, I want to move ATS Engine from an early-stage product into a sustainable, growing business. In terms of users, my first milestone is to build a meaningful base of paying job seekers and recruiters rather than simply chasing registrations. I want to see consistent growth month after month, strong repeat usage and evidence that people are making ATS Engine part of their normal job search or recruitment process. On the revenue side, the goal is to establish predictable recurring revenue that demonstrates the business model works and gives us the ability to keep investing in the platform. I don't want to measure success purely by a headline user number. A thousand people who sign up and never return is less valuable than a few hundred people who actively use ATS Engine and are willing to pay for it. The next 12 months are therefore about proving three things: People want it. Job seekers and recruiters see enough value to use ATS Engine. People pay for it. The product solves problems important enough that users are willing to subscribe. People stay. The value is strong enough to create repeat usage, retention and recommendations. If we can demonstrate those three things while continuing to improve the product, I will consider the next 12 months a success. The numbers will then give us the foundation for the next stage of growth. Q: And for the next 24 months? Nowland: Over the next 24 months, the ambition is to move ATS Engine beyond proving that the idea works and establish it as a genuinely sustainable recruitment technology business. The first 12 months are about proving the fundamentals: growing users, converting them into paying customers, understanding retention and building predictable recurring revenue. By 24 months, I want ATS Engine to have a substantial base of active paying users across both job seekers and recruiters, with the recruiter side becoming an increasingly important part of the business. I also want to see the product become much more intelligent through everything we learn from real-world usage. The aim isn't simply to add more features. It is to understand the problems users repeatedly encounter and make ATS Engine better at identifying what matters, explaining why it matters and helping them decide what to do next. Commercially, I want recurring revenue to have reached a level where ATS Engine can fund continued development and growth rather than relying on the founder continually putting resources into it. I'd also like to have established stronger partnerships and routes to market, particularly within recruitment, staffing and organisations that could benefit from the platform. Most importantly, by the end of those 24 months I want to be able to look at ATS Engine and say that it has moved from an idea I built to solve my own problem into a product that other people genuinely depend on. The exact user and revenue targets will evolve as we learn from the first year. I would rather set ambitious targets based on real customer behaviour than pick impressive numbers today simply for the sake of having a big number. Q: Where do you want ATS Engine to be in five years? Nowland: In five years, I want ATS Engine to be recognised as a serious recruitment technology platform that genuinely improves the relationship between job seekers, recruiters and employers. I don't want it to simply become another ATS with more features. I want ATS Engine to sit across the recruitment journey and help people understand what is happening, why it matters and what they should do next. For job seekers, I want it to be something they can rely on throughout their career — from finding opportunities and understanding their fit, through to applications, interviews, offers and future career decisions. For recruiters and organisations, I want ATS Engine to become a trusted intelligence layer that helps them identify the right candidates, understand what needs attention, manage compliance and make better decisions without adding unnecessary administration. By then, I expect AI agents to be a normal part of recruitment. I want ATS Engine to be at the forefront of that change, using AI to make recruitment more intelligent while keeping human judgement at the centre of important decisions. Commercially, I want a sustainable, profitable business with a strong customer base, recurring revenue, strategic partnerships and a team capable of taking the company well beyond what I could achieve on my own. But the biggest measure of success would be the impact. I started ATS Engine because I found the recruitment process unnecessarily complicated and difficult to manage. Five years from now, I want to have helped make that process fundamentally easier for thousands — and ultimately many more — people. If someone can say, “ATS Engine helped me find the right opportunity, helped me present myself properly, and helped me understand what to do next,” then I will know we built something that really matters. Q: What are your hopes and fears for a post AGI world? Nowland: My biggest hope for a post-AGI world is that AI gives people more freedom rather than simply making organisations more efficient. We have an opportunity to remove enormous amounts of repetitive work, help people learn faster, make information more accessible and give individuals capabilities that previously required large teams or years of experience. For me, that is one of the reasons I find AI so exciting. ATS Engine itself started from using technology to make a difficult process easier. A much more capable AI could take that idea much further. But my biggest fear is that we get the balance wrong. If AI becomes capable of doing more and more of the work, there is a risk that people become valued primarily for what an algorithm thinks they can contribute. We could end up with systems making decisions about people's careers, opportunities and lives without people really understanding how those decisions were reached. There is also a risk of creating a world where the people who control the most capable AI systems gain enormous economic and social power, while everyone else struggles to find their place in a rapidly changing economy. I don't think the answer is to resist AI. I think we need to be much more deliberate about how we use it. AI should augment human capability rather than remove human agency. It should help people understand, create, learn and make better decisions — while important decisions about people's lives should still have meaningful human oversight. That's particularly important to me in recruitment. A person's career can affect their finances, family, confidence and future. I don't want a post-AGI world where someone loses an opportunity because an invisible system decided they weren't worth considering. I'd like to see a future where AI handles much of the complexity behind the scenes, but people remain informed, empowered and ultimately in control. If we get that right, I think a post-AGI world could give people an extraordinary amount of opportunity. If we get it wrong, we could create incredibly powerful systems that make the world more efficient without necessarily making it better. About ATS Engine: ATS Engine is an AI recruitment platform for both sides of hiring, built and run by Dennis Nowland in Fife, Scotland (DennisNowland Ltd). Job seekers can see how well their CV matches a role before applying, tailor CVs and cover letters, and keep track of every application. Recruiters get one place for candidates, vacancies, pipeline, compliance and clients, with intelligent candidate matching and a view of what needs attention today. www.ats-engine.com ## AIStupidLevel (Romania) URL: https://startup.talk/romania/aistupidlevel/ Founders: Ionut Adrian Visan Ionut Adrian Visan grew up in Craiova, Romania, in the hopeful years after the fall of communism, and got a first computer in 1996: a 286 running MS-DOS and Norton Commander. Games like Prince of Persia came first, then the urge to find out how everything worked. A first job maintaining a neighbourhood company's computers, reinstalling Windows almost weekly, taught troubleshooting and patience. Visan has been building games for about 15 years and runs Slotty Labs alongside AIStupidLevel. AIStupidLevel started in 2024 as a tool for Visan's own work, after noticing that an AI model could be brilliant one moment and almost useless a few hours later. It now tests leading models continuously on real coding, reasoning and tool-use tasks, flags when a model quietly degrades, and feeds that data into a Smart Router that picks the right model for each request. It has reached more than 120,000 users without any paid marketing. We spoke about a name that started as a joke, retracting earlier alerts in public, and why choosing an AI model will one day feel like choosing a server. Key takeaways: - AIStupidLevel continuously benchmarks leading AI models on real tasks: fixing bugs in working repositories against hidden tests, multi-turn reasoning, and tool use in sandboxes. Coding tasks run seven times per model. - A cumulative-sum drift detector compares each model with its own history, so one bad answer does not trigger an alert, but a genuine decline does. - The Smart Router uses the same data to pick a model for each request by quality, cost, speed or consistency, using the developer's own API keys. - More than 120,000 users and two million visits with no paid marketing. The 12-month targets: 400,000 users and $5 million in annual recurring revenue. The Principal / Founder: - Ionut Adrian Visan (Founder and CEO) Based: Craiova, Romania Grew up: Craiova Also: Founder and CTO, Slotty Labs Background: Software architecture and full-stack engineering; building games for about 15 years Users: More than 120,000, all organic Website: aistupidlevel.info LinkedIn: Ionut Adrian Visan (https://www.linkedin.com/in/ionut-visan-205ab01a5/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Visan: I grew up in Craiova, Romania, in the first years after the fall of communism. There was a lot of optimism back then. Everything was changing, people were hopeful about the future, and with every passing year, the world seemed to become a little more colorful. My childhood was pretty typical for Eastern Europe. Hot summers, long winters, spending most of our time outside with friends. We didn't have much compared to what kids have today, but we had plenty of freedom and imagination. Looking back, it was a wonderful time to grow up. Q: What first got you into building things? Visan: In 1996, my dad bought me my first computer. It was a 286 PC running MS-DOS with Norton Commander. I immediately figured out how to play games on it. I still remember those big floppy disks and games like Prince of Persia, Volfied, and the old Accolade titles. I could spend entire days playing. But eventually, playing wasn't enough. I became curious about how everything worked. How the computer ran those games, what all those files did, and what happened if I changed something. That curiosity never really went away. Q: What was your first job, and what did it teach you? Visan: My first job was maintaining computers, both hardware and software, for a small neighborhood company. I'm talking about the days when reinstalling Windows was practically a weekly ritual. Something broke, you figured out what happened, fixed it, and a few days later something else broke. It taught me to troubleshoot and, more importantly, to be patient. You couldn't just ask an AI what was wrong. You had to understand the problem yourself, experiment, and find a solution. I think that shaped the way I approach building software even today. ### Part II: The road to AIStupidLevel Q: What made you start AIStupidLevel? Visan: Back in 2024, I started noticing something strange while working with AI models. A model could be absolutely brilliant one moment and almost useless a few hours later. Same model, same prompt, completely different results. And nobody seemed to be paying much attention to it. We had all these leaderboards telling us which model was the smartest, but very little information about whether that model was still performing the same way today as it did yesterday. I wanted something that would continuously test these models and tell me when something changed. So I started building it. Initially, it was just a tool for myself. I figured someone had to keep an eye on these things. Q: When did you first notice that the same AI model and prompt could give materially worse results from one day to the next? Visan: Actually, within the first few days of running AIStupidLevel. Once I started collecting results continuously, the differences became impossible to ignore. What was interesting wasn't just that models sometimes performed worse. It was how much their performance could fluctuate without any announcement or explanation. Of course, AI models are probabilistic, so some variation is normal. The real challenge was figuring out when we were seeing ordinary randomness and when something had genuinely changed. That became one of the most important problems we wanted to solve. Q: How did AIStupidLevel reach more than 120,000 active users without paid marketing? Visan: I think the simplest explanation is that it solves a real problem. I've never spent money on advertising AIStupidLevel. No paid campaigns, no growth hacks or marketing agency. Everything has been organic. Developers were experiencing the same frustrations I was, and when they found a tool that helped them understand what was happening with AI models, they started sharing it. A Reddit post went viral, publications picked up the story, and traffic started coming in from all over the world. I think when you build something genuinely useful, people eventually find it. It doesn't happen overnight, but word of mouth is incredibly powerful. Q: You also run Slotty Labs. How do you split your time between the two, and what is it like to build from Craiova? Visan: I've been building games for about 15 years, and I absolutely love it. Slotty Labs is what keeps that passion alive. I genuinely enjoy creating unusual game concepts, experimenting with mechanics, and watching an idea turn into something people can actually play. AIStupidLevel originally came from that same world. I was using AI heavily while developing software and games, and I needed a better way to know which models I could rely on. So in a way, the two businesses complement each other. Slotty Labs lets me keep creating, while AIStupidLevel helps me build more efficiently and has grown into something much bigger than the tool I originally imagined. As for Craiova, I love it here. The internet has made geography much less important for the kind of work we do. I can build products used by people across the world without having to move to San Francisco or London. You don't need to be in Silicon Valley to build something valuable. ### Part III: The product Q: Why the name AIStupidLevel? Visan: It started as a joke. I was frustrated with how inconsistent some AI models were, and I thought it would be funny to measure not just how intelligent they were, but how stupid they could become. I honestly never expected the project to get mainstream attention, so I didn't spend much time thinking about branding. By the time it started gaining traction, the name had stuck. And actually, I quite like it now. It's memorable, it doesn't take itself too seriously, and underneath the funny name there's some very serious engineering. Q: How do you design benchmarks that measure real coding, reasoning and tool-use quality rather than test-set memorisation? Visan: The most important thing is that we don't just ask models questions and check whether their answers look correct. We make them actually do things. For coding, we give models working repositories with real bugs to find and fix. Then we execute their code against tests, including hidden tests they haven't seen. For reasoning, we use longer, multi-turn tasks where models need to remember instructions, maintain a plan, deal with new information, and recover from mistakes. For tool use, we put them in sandboxed environments where they have to call real tools and complete actual tasks. We also repeat tests. For example, our coding tasks run seven times per model because a single successful answer doesn't tell you much about reliability. The benchmark tasks themselves are private to reduce contamination, while the scoring methodology is public. My philosophy is pretty simple: I care much more about whether a model can get the job done than whether it can produce an impressive-looking answer. Q: How does CUSUM drift detection decide that a model has actually degraded? Visan: The basic idea is that one bad result doesn't mean a model has become worse. AI models naturally fluctuate. Sometimes they make mistakes, sometimes they perform exceptionally well. If we sent an alert every time a score dropped, the system would be practically useless. We use a cumulative-sum approach, specifically a Page-Hinkley detector, that compares a model's recent performance against its own historical behavior. Rather than reacting to every fluctuation, it looks for evidence that the model has shifted away from its normal performance level. We also monitor coding, reasoning, and tool use separately because each behaves differently. And we require enough historical data before making judgments. The goal isn't to catch every bad answer. It's to identify meaningful degradation without constantly raising false alarms. Q: How does the Smart Router choose the right model for each request? Visan: It uses the data we're already collecting. Instead of hardcoding a model name into your application, you can tell our router what matters to you: coding performance, reasoning, tool use, overall quality, cost, speed, or consistency. The router looks at the latest benchmark results, considers your preferences and constraints, and selects the most suitable available model. It also accounts for provider health and degradation signals. If a provider is having problems, or a model is no longer performing as expected, the router can move to another suitable option. You connect your own provider API keys, so you're not forced into a particular ecosystem. I wanted developers to spend their time building things, not constantly checking leaderboards and changing model names in their code. Q: Who gets the most value from AIStupidLevel: individual developers, startups or enterprises running AI in production? Visan: All three, but in different ways. Individual developers use it to figure out which models are worth their time and money. Startups can use the Smart Router to reduce API costs without sacrificing too much quality, which can make a big difference when you're operating on a limited budget. For enterprises, reliability becomes extremely important. When AI is part of a production system, you need to know if a model starts behaving differently, if your provider is experiencing problems, or if a cheaper alternative could deliver similar results. The stakes are different, but the underlying problem is the same. Q: What do you deliberately leave out of AIStupidLevel? Visan: We're not trying to build another chatbot, and we're not trying to create our own foundation model. I also don't want to turn AIStupidLevel into a leaderboard where we add hundreds of meaningless measurements just to make it look impressive. I'd rather measure fewer things properly than publish numbers I can't defend. We deliberately avoid presenting statistical noise as meaningful differences, and we don't publish our hidden benchmark tasks because that would eventually make them useless. The focus is on measuring what matters, being transparent about limitations, and making the results useful. ### Part IV: Positioning Q: Who is AIStupidLevel for, and what job does it do for them? Describe the person who gets the most value. Visan: Imagine you're a developer or CTO building an application that depends heavily on AI. You're spending money on API calls every day. New models are being released constantly. Prices change, performance changes, and sometimes the model you've been relying on starts making mistakes it wasn't making last week. You don't have the time or resources to continuously evaluate every available model yourself. That's the person we're building AIStupidLevel for. We do the testing, monitor changes, and help them choose and use the right models. Essentially, we want to remove the guesswork from working with AI. Q: How do you view your positioning relative to LMArena, Artificial Analysis, OpenRouter and the labs' own published benchmarks? Is there another competitor you think of first? Visan: I think all of them are doing valuable work, and the market is big enough for different approaches. LMArena is particularly useful for understanding human preferences between model outputs. Artificial Analysis has built a very comprehensive platform for comparing capabilities, performance, and pricing. OpenRouter is an important player in model access and routing, and they've also developed their own intelligent routing capabilities. The labs' published benchmarks are useful too, although naturally I think independent verification is important. Our focus with AIStupidLevel is continuous, reproducible measurement tied directly to operational decisions. We want to know not only how a model compares with others, but whether it has changed, how reliable it is over time, and whether you should still be using it for your particular workload. Then we connect that information directly to the Smart Router. I don't really think of this as a winner-takes-all competition. Different products answer different questions. Q: What is your core moat in AI model benchmarking and routing? Visan: The historical data, the evaluation infrastructure, and the connection between measurement and routing. We've conducted more than 500,000 controlled evaluations, and every new run adds to our understanding of how models behave over time. A single benchmark result is relatively easy to reproduce. Building an infrastructure that continuously measures models, understands normal fluctuations, detects meaningful changes, and turns that information into routing decisions is considerably harder. But I don't think having a complicated algorithm is a moat by itself. The real advantage comes from accumulating reliable data, proving that your measurements mean something, and earning the trust of developers who depend on them. That takes time. Q: What are some of the use cases for AIStupidLevel that are not obvious and can add a lot of value to users? List as many as possible. Visan: People usually discover us because they want to compare AI models, but there's quite a bit more you can do. For example: -Detect when a model you've been using starts performing worse, even if the provider hasn't announced any changes. -Reduce API spending by identifying cheaper models that perform nearly as well on a specific workload. -Automatically switch models when a provider is unavailable or a model becomes unreliable. -Monitor the consistency of AI agents that depend on tool calling. -Compare how different providers serve models and how that affects reliability. -Track model performance before and after updates. -Build custom evaluation pipelines using our historical data. -Choose different models for coding, reasoning, and tool-heavy tasks instead of using one expensive model for everything. -Identify models that are significantly more expensive without delivering meaningfully better results. -Help engineering teams make evidence-based decisions about which AI providers to adopt. -Monitor the performance of models used inside automated business workflows. -Research how model reliability changes over weeks and months. -Use benchmark history to decide when migrating away from an older model makes sense. -Detect instability before it becomes a bigger problem for an application. We're also expanding the platform so model providers and hosting companies can bring their own models into our benchmarking and routing ecosystem. And for companies with specialized AI workloads, custom benchmarks and evaluations are an especially interesting direction. There's much more value in understanding how AI behaves in the real world than simply knowing which model happens to be number one today. Q: What is the one detail in AIStupidLevel you are proudest of? Visan: Probably the fact that we're willing to say when we don't know. That might sound strange, but AI benchmarking is full of uncertainty. Sometimes two models have slightly different scores, but statistically, there's no meaningful difference between them. We don't pretend otherwise. They can share the same rank. The same applies to drift detection. A model having a bad day doesn't automatically mean it has degraded. We've put a lot of work into making the system honest about what the data can and cannot tell us. I'm proud of that because it's very easy to publish impressive numbers. It's much harder to make sure those numbers are actually meaningful. ### Part V: Launch Q: What did you learn from your launch? Visan: That launching a product and building a useful product are two very different things. A launch can bring attention, but that attention doesn't mean much if people don't have a reason to come back. For me, the most important lesson was to keep focusing on the product itself. I would rather have a smaller group of developers who genuinely depend on what we're building than a massive spike in traffic that disappears the next day. The real work starts after the launch. Q: Which feedback from the launch surprised you? Visan: What I found most interesting was how the conversation could move beyond rankings. You might expect everyone to ask which AI model is the smartest. But the more useful question is often whether the model they're already using is still as good as it was before. That's a very different problem. And it helped reinforce the direction we're taking with drift monitoring and the Smart Router. It's not enough to tell people what's happening. You also want to help them do something about it. ### Part VI: The hard part Q: How many people use AIStupidLevel today? Visan: We've reached more than 120,000 users historically, entirely through organic growth. We've also passed two million total visits. I think it's important to distinguish between cumulative users, monthly visitors, and registered accounts, because they're not the same thing. What interests me most now is how many people move from checking the benchmarks to using our data and routing infrastructure as part of their everyday work. That's where we can create lasting value. Q: What is a challenging time you had building AIStupidLevel? Visan: One of the hardest parts has been making the measurements genuinely trustworthy. It's surprisingly easy to build a benchmark that produces numbers. It's much harder to build one that produces numbers you can confidently stand behind. At one point, we discovered that parts of our earlier alerting methodology were too sensitive and were identifying changes that weren't sufficiently supported by statistical evidence. We had to revise the methodology and retract earlier incidents. It wasn't pleasant, but it was absolutely the right thing to do. There's also the financial side. Running hundreds of thousands of real AI evaluations costs a significant amount of money, especially when you're bootstrapping. But I've always believed that if you're building a product around trust, you have to be prepared to admit mistakes and fix them publicly. Q: Who believed in you before there was anything to show? What did they say or do? Visan: In the beginning, I wasn't really looking for anyone's permission or approval. I was building something I needed myself, and that was enough of a reason to start. The first real validation came from strangers on the internet. People who had no reason to support me started using the platform, sharing it, and coming back. Some were developers working on their own projects, others were simply curious about how these models behaved. Seeing people find value in something I originally built for myself meant a lot. There's something special about realizing that a problem you've been struggling with is also a problem thousands of other people are facing. Q: How do you stay independent and trusted when the companies you benchmark are also potential partners or customers? Visan: By making sure that commercial relationships don't determine evaluation results. Every model is measured using the same methodology, the same tasks, and the same scoring rules. We publish how our measurements work, including their limitations. The actual benchmark tasks remain private so they can't simply be optimized against. As we expand to allow AI providers and hosting companies to submit their own models, the same principle will apply. A provider can participate in the ecosystem, but that doesn't mean it gets a better score. If we ever lose the trust of developers who rely on our data, the entire business loses its purpose. Independence isn't just a nice thing to have. It's fundamental to what we're building. ### Part VII: What it is all for Q: What would make you say AIStupidLevel has succeeded? Visan: When developers no longer have to worry about which AI model they should use. That might sound a little strange coming from someone who built an AI benchmarking platform, but that's genuinely the goal. I want people to focus on building their applications while the infrastructure handles model selection, reliability, performance monitoring, and cost optimization. If thousands of businesses are using our data and Smart Router every day, and they can trust that their AI infrastructure is making sensible decisions without constant manual intervention, I would consider that a huge success. Q: What will choosing an AI model look like when there are thousands of them and they change every week? Visan: I don't think people will choose individual models the way they do today. There will simply be too many of them. You won't want to spend your morning comparing 3,000 different models, reading release notes, and trying to figure out whether the newest one is actually better. Instead, you'll specify what you need. Maybe you want the highest coding accuracy within a certain budget. Maybe you need extremely reliable tool calling. Maybe latency matters more than anything else. The infrastructure will select the appropriate model and adjust as the available options change. I think choosing an AI model will eventually become a lot like choosing which server handles a request. Most people won't need to think about it. That's the future we're working toward. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Visan: The next 12 months are really about turning the organic interest we've built into sustained production adoption. We've already demonstrated that developers want independent information about AI models. Now we need to make AIStupidLevel something they actively integrate into their applications. My targets are to reach 400000 users and 5000000 USD ARR, with a growing number of companies using our Smart Router and Data API in production. We're also expanding support for model providers and hosting companies, as well as working toward more specialized enterprise evaluations. I care about user growth, of course, but I'd much rather have a thousand companies genuinely depending on the platform than a million people visiting once out of curiosity. Q: And for the next 24 months? Visan: In 24 months, I want AIStupidLevel to be a recognized part of the AI infrastructure ecosystem. By then, I want us to be supporting a much broader range of models and hosting providers, including self-hosted and open-weight deployments. The idea is to build something that scales internationally without losing the independence and transparency that made people interested in the first place. Q: Where do you want AIStupidLevel to be in five years? Visan: I want AIStupidLevel to become one of the standard intelligence layers behind AI applications. Not necessarily something every end user knows about, but something that developers and businesses depend on. When a company needs to understand how an AI model performs, whether it's reliable, how its behavior has changed, or which alternative would work better, I want AIStupidLevel to be one of the first places they turn to. And ideally, much of that process will happen automatically through our infrastructure. Five years is a very long time in AI. I have no idea what models will look like by then. But I'm fairly confident that as AI becomes more capable and more widely used, independent measurement and reliability will become more important, not less. Q: What are your hopes and fears for a post AGI world? Visan: My biggest hope is that intelligence becomes abundant and accessible to everyone. Imagine what that could mean for medicine, scientific research, education, or even something as simple as giving a child anywhere in the world access to the best teacher imaginable. I think the possibilities are extraordinary. My biggest fear is that we become so dependent on these systems that we stop questioning them. Or that an enormous amount of power becomes concentrated in the hands of a very small number of companies or governments. I don't think having machines that are more intelligent than us is necessarily the frightening part. What worries me more is whether we'll use that intelligence wisely and whether people will still have meaningful control over the decisions that shape their lives. I'm optimistic about the future, though. I wouldn't be building in this industry if I weren't. I just think we need to be a little more careful about what we're creating and a lot more serious about understanding it. About AIStupidLevel: AIStupidLevel is an independent AI benchmarking platform built by Ionut Adrian Visan in Craiova, Romania. It continuously tests leading AI models on real coding, reasoning and tool-use tasks, detects when a model's performance drifts, and powers a Smart Router that selects the best available model for each request using the developer's own API keys. aistupidlevel.info ## AFFLIGO (India) URL: https://startup.talk/india/affligo/ Founders: Karan Kumar Karan Kumar grew up in Gaya, Bihar, in a middle-class family without many resources, curious from early on about how computers and websites work. While studying Electrical Engineering at Government Polytechnic, Gaya, Kumar started learning web development independently, through tutorials, experiments and AI tools. There has been no formal job yet: the time has gone into learning and building. AFFLIGO started with a domain bought for affiliate marketing and no money to carry on with that plan. Kumar turned it into a home for free online tools instead, after growing tired of file tools with size limits, forced sign-ups and too many ads. It now has more than 65 PDF, image and file tools, many of which process files directly in the browser. We spoke about building on an old laptop with no working battery and a mobile hotspot, about 2,200 hours of work, a Reddit comment that kept the project going, and why a small city is no barrier to a global product. Key takeaways: - AFFLIGO offers more than 65 free PDF, image, file and calculator tools with no sign-up. Tools such as PDF Merge and Image Compress process files in the user's own browser. - Its Device Guard system checks a device's available memory and adjusts processing, so the tools also work on low-powered devices. - Kumar built it solo from Gaya, Bihar, using prompt engineering and AI tools, over about 2,200 hours on an old laptop and a mobile hotspot. - Google Analytics recorded 447 active users between 6 and 16 November. The plan is one simple banner ad per tool page and no pop-ups, to cover running costs. The Principal / Founder: - Karan Kumar (Founder) Based: Gaya, Bihar, India Studied: Electrical Engineering, Government Polytechnic, Gaya Built: Solo, with AI tools and prompt engineering Tools: 65+ free PDF, image, file and calculator tools Website: affligo.com ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Kumar: I grew up in Gaya, a city in Bihar, India. My childhood was simple. I come from a middle-class family, and we did not have many resources. From an early age, I was curious about technology and computers. I always liked learning new things and trying to understand how things work. I did not have access to many advanced facilities, but I never stopped learning. I believe my background taught me to work with what I have and find my own ways to learn and grow. Q: What first got you into building things? Kumar: I have always been curious about how websites and online tools work. I used to see different websites and wonder how people build them. This made me interested in learning web development. I started learning by watching tutorials, trying things myself, and using different AI tools to help me understand coding. Slowly, I started making small projects. I enjoyed the process of turning an idea into something that people could actually use. That feeling made me want to build more things. Q: What was your first job, and what did it teach you? Kumar: I haven't had a formal job yet. Most of my time has gone into learning new skills and working on my own projects. This taught me how to manage my time, solve problems on my own, and keep learning without anyone guiding me all the time. Working on my own projects also helped me understand that building something takes patience, practice, and a lot of trial and error. ### Part II: The road to AFFLIGO Q: What made you start AFFLIGO? Kumar: I started AFFLIGO because I wanted to build something useful that people could use for free. I noticed that many online PDF and image tools either had too many limits, required payment, or showed too many ads. I wanted to create a platform where people could easily edit, convert, and manage their files without paying for basic features. I started working on this idea with the skills I had and learned new things along the way. That's how AFFLIGO started. Q: How did you go from studying electrical engineering at Government Polytechnic, Gaya to teaching yourself web development? Kumar: I was studying Electrical Engineering at Government Polytechnic, Gaya. Along with my studies, I developed an interest in web development and started learning it on my own. I know the basics of coding, but I have focused more on prompt engineering and learning how to use AI tools effectively. I use AI to help me write code, solve problems, and turn my ideas into working websites and tools. I am still learning every day. I believe I don't need to know everything myself to build something useful. What matters is having an idea, learning the right skills, and knowing how to use available tools to make it happen. Q: What frustrated you about existing free file tools that convinced you to build your own? Kumar: When I used free online file tools, I noticed many problems. Some websites had file size limits, some asked users to create an account, and others showed too many ads. Many tools also had limited features in their free versions. I wanted to make something different. My goal was to build a platform where people could use useful PDF, image, and file tools without paying or creating an account. I also wanted to make the tools simple and easy to use for everyone. These problems gave me the idea to start AFFLIGO. Q: What is it like to build a global product from Gaya, Bihar? Kumar: Building a global product from Gaya, Bihar is a great experience for me. I may be building from a small city, but the internet gives me the chance to reach people from all over the world. I don't have a big team or a big office. I mostly work on my own, using my laptop, the internet, and AI tools. Sometimes I face challenges, especially when I don't know how to solve a technical problem, but I try to learn and find a solution. For me, AFFLIGO is proof that you don't need to be in a big city to build something that people around the world can use. ### Part III: The product Q: Why the name AFFLIGO? Kumar: Actually, I originally bought this domain for affiliate marketing. But I didn't have enough money to continue with that plan, so I decided to turn it into a website for free online tools. Interestingly, I later found out that AFFLIGO has a Latin meaning related to "crush". I liked this coincidence and decided to use the tagline "Crush Your File Problems". That's how the name and tagline became connected to the idea behind the platform. Q: How do you keep files private when people merge PDFs or compress images on AFFLIGO? Kumar: Privacy is very important to me. For tools like PDF Merge and Image Compress, file processing happens directly in the user's browser instead of uploading files to our servers. This means users can work with their files on their own devices, and we don't need to store their personal files on our servers. I also want to make file processing simple and safe without asking users to create an account. My goal is to give people useful tools while respecting their privacy. Q: How do you decide which tool to build next? Kumar: I usually look at what people need and what problems they face while working with files. I also research popular searches on Google to understand which tools people are looking for. Sometimes I get ideas from my own experience or from tools available on other websites. I try to focus on tools that are useful, easy to use, and can solve real problems. I also use AI tools to explore ideas and understand how I can build new features. My goal is to keep adding tools that make AFFLIGO more useful for everyone. Q: Which tools are used most, and by whom? Kumar: Based on our Google Analytics data, some of our most viewed tools are PDF Compressor, Password Protect PDF, Image Enhancer, JPG to PDF, and PDF Merger. These tools are useful for students, office workers, freelancers, and people who work with documents and images in their daily lives. For example, students may need to compress PDFs for assignments, while office workers may need to merge or protect important documents. Image tools are also useful for people who work with photos. We are still growing, but the data helps me understand what people are interested in and which tools I should improve next. Q: You are open that the site is AI-assisted. How do you use AI in building it? Kumar: AI plays a big role in building AFFLIGO. I know basic coding, but I use AI tools a lot to help me with development. I mainly use prompt engineering to explain my ideas and get help with writing code, fixing errors, improving designs, and adding new features. I also use AI to understand technical problems that I cannot solve on my own. I don't just copy everything AI gives me. I test things, make changes, and check whether they work properly. AI has helped me build AFFLIGO with limited technical knowledge and resources. For me, AI is like a helpful partner that makes learning and building things easier. Q: What do you deliberately leave out of AFFLIGO? Kumar: I deliberately try to keep AFFLIGO simple. I don't want users to deal with unnecessary steps, forced sign-ups, or confusing designs just to use a basic tool. I also don't want to put basic features behind a payment wall. My goal is to make useful tools freely available to everyone. I believe a tool website should solve problems, not create more problems. I want users to visit AFFLIGO, get their work done, and leave with a good experience. ### Part IV: Positioning Q: Who is AFFLIGO for, and what job does it do for them? Describe the person who gets the most value. Kumar: AFFLIGO is for anyone who needs to work with PDF files, images, or other documents. I think students, freelancers, office workers, and small business owners can get the most value from it. For example, a student may need to compress a PDF before submitting an assignment, while an office worker may need to merge or protect documents. Someone working with images may need to resize, convert, or improve them. AFFLIGO helps them complete these tasks quickly without spending money on expensive software. I want it to be a simple place where anyone can find the tool they need and get their work done without technical knowledge. Q: How do you view your positioning relative to iLovePDF, Smallpdf, TinyPNG, remove.bg and Canva's free tools? Is there another competitor you think of first? Kumar: I see these platforms as well-established products that have been helping users for years. I don't think AFFLIGO needs to copy them or compete with every feature they offer. My goal is to build a platform where people can find different PDF, image, and file tools in one place. I want to keep the tools simple, free, and easy to use without forcing users to create an account. iLovePDF and Smallpdf are the closest competitors for our PDF tools, while TinyPNG and remove.bg are more focused on specific image tasks. Canva also offers many useful features. If I had to name one competitor first, I would say iLovePDF because we both focus on online PDF tools. But AFFLIGO is still growing, and I want to build its own identity by offering more useful tools in one place. Q: What is your core moat in free online file tools? Kumar: I think our biggest strength is that we are trying to bring many useful tools together in one place. Users don't need to visit different websites for every small task. AFFLIGO focuses on PDF, image, and file tools that are free and easy to use. I also give importance to privacy, and many of our tools process files directly in the user's browser. Apart from the main tools platform, I have also built AFFLIGO Studio. My idea is to make AFFLIGO more than just a file tools website by adding more useful features and services over time. Another advantage is that I use AI tools and prompt engineering to develop new features and improve the platform, even with limited resources. I know there are already many established companies in this space, and AFFLIGO is still growing. I don't want to claim that we are better than them right now. My goal is to build trust by keeping the platform useful, simple, free, and continuously improving it. I believe that our combination of free tools, browser-based processing, AFFLIGO Studio, and continuous development can help us build our own place in this market. Q: What are some of the use cases for AFFLIGO that are not obvious and can add a lot of value to users? List as many as possible. Kumar: AFFLIGO can be useful in many situations beyond basic file conversion. Here are some use cases that I think can bring a lot of value to users: 1. Students: Compress large PDFs before submitting assignments, merge notes from different subjects, convert images into PDFs, and protect important documents. 2. Job Seekers: Resize profile photos, convert certificates into PDFs, merge resumes and documents, and reduce file sizes before uploading them to job portals. 3. Small Business Owners: Compress product images, create PDF invoices, merge business documents, and convert files into formats needed by clients. 4. Freelancers: Resize images for clients, convert documents, compress large files, and prepare professional-looking files for delivery. 5. Teachers: Combine study materials, create PDFs from scanned pages, compress learning materials, and share documents more easily. 6. Office Workers: Merge reports, split large PDFs, add page numbers, rotate pages, and protect important documents. 7. Online Applications: Prepare documents according to upload size limits on government, college, and job application websites. 8. Content Creators: Resize thumbnails, compress images for faster website loading, convert image formats, and prepare files for social media. 9. Website Owners: Compress images, convert JPG and PNG files to WebP, resize website graphics, and reduce page loading time. 10. People Using Smartphones: Convert photos into PDFs, reduce image sizes, and manage documents without installing extra apps. 11. Privacy-Conscious Users: Use browser-based tools to process files without uploading them to a server, wherever the tool supports local processing. 12. Small Shops and Local Businesses: Prepare customer bills, scan documents, resize product photos, and organize digital records. 13. Researchers: Combine research papers, split large documents into smaller sections, and organize scanned study materials. 14. Real Estate Agents: Compress property photos, prepare property documents, and combine multiple pages into a single PDF. 15. Travelers: Combine tickets, hotel bookings, and travel documents into one PDF for easy access. I believe these are just some examples. Different people have different file-related problems, and AFFLIGO aims to make these everyday tasks easier without requiring expensive software or advanced technical knowledge. Q: What is the one detail in AFFLIGO you are proudest of? Kumar: The thing I am proudest of is that I built AFFLIGO from scratch with limited resources and without a professional development background. One detail I am especially proud of is the Device Guard system. It checks the available device memory and adjusts file processing based on the device's capacity. It also shows users how much file size their device can handle and how many files can be added to the processing queue. I wanted to make AFFLIGO useful for people with different types of devices, not just those with powerful computers. I am also proud that I have been able to build more than 65 tools while learning and using AI along the way. For me, AFFLIGO is not just a website; it is the result of continuous learning, experiments, and hard work. ### Part V: Launch Q: What did you learn from your launch? Kumar: Launching on Product Hunt taught me that building a product is only one part of the journey. Making people aware of it is equally important. I learned that getting attention and reaching the right audience takes time and effort. I also understood the importance of listening to feedback and improving the product based on what users need. As a solo founder from a small city, launching on Product Hunt was a new experience for me. It gave me confidence that even a small project built with limited resources can reach a global audience. Most importantly, I learned that a launch is not the finish line. It is just the beginning of building and improving a product. Q: Which feedback from the launch surprised you? Kumar: During the launch, I also shared AFFLIGO on Reddit and asked people to give their feedback. Some people made fun of the website and gave negative comments, which was a little disappointing. But one comment really surprised me. Someone said that my website deserved millions of users. Honestly, that one comment made me very happy and gave me a lot of motivation. It reminded me that even if many people don't like what you build, one positive comment can make all the hard work feel worth it. It encouraged me to keep working on AFFLIGO and make it better. ### Part VI: The hard part Q: How many people use AFFLIGO today? Kumar: AFFLIGO is still in its early stage, and we are growing gradually. According to our Google Analytics data, we recorded 447 active users and 769 page views between November 6 and November 16. However, these numbers don't show the complete picture. We had some issues with Google Analytics tracking earlier, so data from previous visits was not properly recorded. Also, some visitors use ad blockers or don't give consent to analytics tracking. Because of this, I cannot give an exact number of total users. But I am happy to see people from different countries visiting and using AFFLIGO. My focus right now is on improving the platform and reaching more people. Q: What is a challenging time you had building AFFLIGO? Kumar: One of the most challenging parts of building AFFLIGO was working with very limited resources. I was using an old laptop that didn't even have a working battery. I also didn't have Wi-Fi at home, so I used my mobile hotspot to connect to the internet and continue my work. I used to spend around 8–9 hours every day working on AFFLIGO. So far, I have spent around 2,200 hours building and improving it. There were many times when I faced technical problems and didn't know how to fix them. Since I was working mostly on my own and had only basic coding knowledge, I had to depend on learning, experimenting, and AI tools to find solutions. It wasn't always easy, but I never wanted to stop just because I had limited resources. Looking at AFFLIGO today, I feel proud that I was able to build so much with what I had. Q: Who believed in you before there was anything to show? What did they say or do? Kumar: Honestly, the person who believed in me before AFFLIGO became something was me. In the beginning, I didn't have a team, investors, or anyone who could promise that this idea would work. I just had an idea and the desire to build something useful. There were times when things were difficult, and I didn't know whether AFFLIGO would succeed or not. But I kept working because I believed that I could make it happen. I didn't wait for someone else to believe in my idea. I decided to believe in myself and give it a real try. Today, seeing AFFLIGO grow makes me feel that taking that first step was worth it. Q: How do you keep a free tool site sustainable without bombarding people with ads? Kumar: I believe free tools should remain easy to use, and users should not have to deal with too many ads just to get their work done. My plan is to use only simple banner ads, with just one banner on a tool page. I don't want to use pop-ups or other ad formats that disturb users while they are working. We are currently testing banner ads on 10 tool pages for 7–10 days to understand how they perform. If the results are not good or the ads affect the user experience, we will remove them and wait until we have more users. I want AFFLIGO to earn enough to cover its running costs and support future development, but not at the cost of a bad user experience. My priority is to build trust first and grow the platform gradually. ### Part VII: What it is all for Q: What would make you say AFFLIGO has succeeded? Kumar: For me, AFFLIGO will be successful when people from different parts of the world use it regularly and trust it to solve their everyday file-related problems. I don't want AFFLIGO to become just another website with hundreds of tools. I want it to become a platform that people remember whenever they need to work with PDFs, images, or other files. I would also consider it a success if AFFLIGO could support its own running costs, grow without depending too much on outside support, and provide real value to users. My biggest satisfaction would be hearing someone say that AFFLIGO made their work easier and saved them time and money. I am still at the beginning of this journey, but I want to build something useful that can grow for many years. Q: Will people still need file tools when AI assistants can edit documents directly? Kumar: Yes, I believe people will still need file tools even when AI assistants become more advanced. AI can help people edit documents and perform many tasks, but people will still need simple tools for specific jobs like compressing PDFs, merging files, converting images, resizing photos, and protecting documents. Not everyone wants to use an AI assistant for every small task. Sometimes people just want to open a website, upload or select a file, click one button, and get the result. I also believe privacy, speed, and ease of use will remain important. That is why I want AFFLIGO to keep improving and adapting to new technology instead of ignoring it. I see AI as something that can help us build better tools, not just something to compete against. My goal is to make AFFLIGO useful even as technology changes. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Kumar: Over the next 12 months, my main goal is to grow AFFLIGO and reach more users around the world. I want to increase our monthly active users, improve our existing tools, add new useful features, and make AFFLIGO Studio more valuable. On the revenue side, I want to start with simple banner ads without disturbing the user experience. My first goal is to generate enough revenue to cover the website's running costs. After that, I want to slowly increase revenue as our user base grows. I don't want to focus only on big numbers. I want to build a platform that people trust and use regularly. If we can achieve steady growth in users and revenue over the next 12 months, I will consider that a big achievement. Q: And for the next 24 months? Kumar: Over the next 24 months, I want to make AFFLIGO a well-known platform for free online file tools. My goal is to reach more users globally, add more useful tools, improve the performance of existing tools, and continue developing AFFLIGO Studio. I also want to build a more sustainable business model so that AFFLIGO can support its own costs and generate a stable income. If the platform grows well, I would like to invest that revenue back into development and bring more ideas to life. I also hope to build a small team in the future when the platform is ready for it. I know it will take time, but I want AFFLIGO to become a trusted name that people can depend on for their everyday file-related needs. Q: Where do you want AFFLIGO to be in five years? Kumar: In the next five years, I want AFFLIGO to become a trusted global platform for online tools. I want people from different countries to know AFFLIGO and use it for their everyday work. I want to expand beyond PDF and image tools and offer more useful services through AFFLIGO Studio. I also want to keep improving our technology, making our tools faster, easier to use, and available to everyone. My dream is to build a proper company around AFFLIGO, create a small team, and make the platform financially strong enough to support its own growth. Most importantly, I want AFFLIGO to be a product that I can be proud of, something that started with one person and limited resources in Gaya, Bihar, but grew to help millions of people around the world. That would be a big achievement for me. Q: What are your hopes and fears for a post AGI world? Kumar: I feel excited about a future where AI can help people do things that were very difficult or expensive before. I hope it will give more opportunities to people who don't have money, resources, or access to good education. As someone who uses AI to build AFFLIGO, I have seen how much it can help one person turn an idea into something real. I hope that in the future, more people from small towns and simple backgrounds will be able to build their own products and businesses. But I also have some fears. I worry that many people may become too dependent on AI and stop learning things on their own. There is also the risk of job losses and people using AI in harmful ways. I believe AI should help people grow, not make them feel useless. My hope is that we build a future where technology creates more opportunities for everyone, not just a few people. About AFFLIGO: AFFLIGO is a collection of more than 65 free online PDF, image, file and calculator tools for students, professionals and creators, built by Karan Kumar in Gaya, Bihar, India. No sign-up is needed, and tools such as PDF Merge and Image Compress process files in the browser. Its tagline is “Crush Your File Problems”. affligo.com ## ScaScan (Nepal) URL: https://startup.talk/nepal/scascan/ Founders: Machhindra Neupane Machhindra Neupane was born in 1993 and grew up in Sitapaila, Kathmandu, where his parents had moved from a village in Nuwakot in search of a better life. The family was not wealthy, but his friends had computers, and he was playing Need for Speed by 1997. At school he doubted he would pass his School Leaving Certificate; he went on to earn a Bachelor of Information Management with first division at the National College of Computer Studies. Since 2017 he has built Android and Flutter apps at Himalayan Tech, Bidhee, Podamite Nepal and KrispCall, and today he works remotely as a Senior Android Engineer at DialBird. ScaScan began as an Android document scanner and became a brand: free browser tools that process files on your own device, ScaScan Restaurant, which turns a photo of a printed menu into a QR menu with ordering, an AI keyboard, a habit app and a media authenticity checker. He builds it alone around his day job, with AI coding tools, and he is unusually candid about where it stands. We spoke about the LinkedIn video that led to his billing partner, why a free menu builder nobody used became a restaurant platform, and why, in his media checks, human review is the last step, not the first. Key takeaways: - ScaScan is a privacy-first brand: 25+ free browser tools that process files locally where possible, plus ScaScan Restaurant, ScaScan AI Keyboard, UpHabit AI and the Media Signal Analyser. - Machhindra builds it alone, around a full-time remote job as a Senior Android Engineer at DialBird, using AI coding tools to move without a team. - Today: about 100 to 500 monthly users and no paying customer yet. Paddle payments are live, and the Google for Startups Cloud Program gave about $2,000 in credits. - His targets: 1,000 to 3,000 monthly users and $100 to $500 in monthly recurring revenue within 12 months, then 5,000 to 10,000 users and $1,000 to $3,000 within 24 months. The Principal / Founder: - Machhindra Neupane (Founder and Lead Developer) Based: Kathmandu, Nepal Grew up: Sitapaila, Kathmandu Studied: BIM, National College of Computer Studies Before: Android and Flutter developer since 2017: Himalayan Tech, Bidhee, KrispCall Day job: Senior Android Engineer, DialBird Funding: Bootstrapped Website: scascan.com LinkedIn: Machhindra Neupane (https://www.linkedin.com/in/machhindra-neupane-542019148/) ### Part I: Before any of this Q: Where did you grow up in Nepal, and what got you into programming? Neupane: I grew up in Sitapaila, Kathmandu. My parents originally came from a village in Nuwakot and moved to Kathmandu in search of a better life. I was born in 1993, and we were not a wealthy family. My parents worked hard, often working for wealthy families, but I grew up surrounded by people who had access to things we couldn't afford. Interestingly, even though my family was poor, I was exposed to technology at a very young age because of my friends. I remember playing Need for Speed around 1997 and FIFA in the early 2000s. I also got to experience things like Toyota cars through my friends. So, while I didn't personally own those things, I had the opportunity to experience them. Growing up in that environment gave me a different perspective. I could see what technology was capable of, even though my family couldn't afford it. That early exposure made me curious about technology. Over time, my curiosity developed into an interest in computers and programming. Eventually, I became a software developer, and today I'm building my own technology products through ScaScan. Looking back, I think growing up in a financially struggling family while being surrounded by technology played an important role in shaping who I am today. Q: You studied at the National College of Computer Studies in Kathmandu. What did those years teach you? Neupane: My journey at the National College of Computer Studies (NCCS) taught me something much more important than just academic knowledge. It taught me that hard work and determination can change your life. To understand that, I have to go back to my school days. Honestly, I was never a strong student. I struggled with my studies, and at one point, I wasn't even confident that I would pass my School Leaving Certificate (SLC) examination. But something unexpected happened during the final months before my SLC exams. My performance in mathematics suddenly started improving. Even today, I find it difficult to explain exactly how that happened. I started doing very well in mathematics and accounting, scoring around 80 to 90 or above in those subjects. Those marks helped improve my overall SLC results. After that, I joined NCCS for my +2 education in Management. English was still a challenge for me, and I wasn't naturally good at every subject. But I kept working hard and eventually passed with a first division, around 60%. For someone who once doubted whether he could even pass his school examinations, achieving first division was a dream come true. After completing my +2, I decided I didn't want to continue studying only management. I was interested in technology, so I chose Bachelor of Information Management (BIM) at the same college. It was a combination of information technology and management, which gave me the opportunity to explore both fields. My bachelor's journey wasn't easy either, but I continued putting in the effort and eventually graduated with first division again. Looking back, NCCS taught me two things. Academically, it introduced me to information technology and helped me build the foundation for my future career. Personally, it taught me that your past academic performance doesn't determine your future. I was once a student who wasn't sure he could pass his SLC. Years later, I completed my bachelor's degree with first division and went on to build a career in software development. For me, that transformation is the most valuable lesson I took from NCCS. Q: You started as a junior Android developer in 2017. What was your first app? Neupane: Actually, my first Android application wasn't a professional project. It was a simple expense-tracking app that I built myself while studying at the National College of Computer Studies (NCCS), probably during my fourth or fifth semester of my bachelor's degree. At that time, I was just beginning to explore Android development. I wanted to understand how a real application worked, so I started building an expense tracker. The app was quite simple. I used SQLite as a local database to store expenses offline. I learned how to create database tables, save and retrieve records, and display information in lists. What made that project special to me was that I was figuring things out on my own. I wasn't just studying programming concepts anymore. I was actually applying them to build something that worked. After that, I developed another Android application for credit management, which helped me gain more practical experience. My professional Android development journey began in 2017 at Himalayan Tech, where I worked on a school management system. Later, I had opportunities to work on other projects, including Mero Lagani and Fence AI. Looking back, that first expense tracker was a very small application compared to the projects I've worked on throughout my career. But it was important because it gave me my first practical experience of building an Android application from scratch. That simple project was where my Android development journey really began. ### Part II: Eight years of mobile apps Q: You built apps at Himalayan Tech, Bidhee and KrispCall. Which project taught you the most? Neupane: Honestly, I don't think I can choose one particular project or company that taught me the most. Every project was different, and each gave me an opportunity to apply my knowledge, explore new technologies, and solve different problems. My professional journey started at Himalayan Tech in 2017. Interestingly, although I wanted to become an Android developer, I couldn't immediately get an Android development role. Instead, I worked with Angular for around seven months before moving into Android development, where I worked on a school management system. Later, at Bidhee, I worked on a ride-sharing application. After that, I contributed to Mero Lagani, a stock-market application, while working at Podamite Nepal. Another important part of my journey was KrispCall, where I worked on a VoIP application. That project involved both Android and Flutter development. It was also where I transitioned into Flutter while continuing to work with native Android functionality and plugins. But there is something important about how I see my career. I have always believed in learning things myself. Even in the early stages of my career, I had the curiosity and confidence to explore technologies independently. I didn't necessarily need someone to teach me everything step by step. When I faced something unfamiliar, I tried to understand it and figure out how to make it work. I'm not saying I knew everything when I started, or that I haven't learned anything since then. Of course, my practical experience has grown significantly. But the fundamental mindset I had as a beginner is still the same today. Technology changes, frameworks change, and projects become more complex. But my approach has always been to explore, experiment, and learn independently. So, if you ask me which project taught me the most, I would say every project contributed to my experience, but self-learning has been the biggest influence on my development as an engineer. I believe that mindset is also what eventually gave me the confidence to start building my own products. Q: You now work as a senior Android engineer for DialBird, an international company. How did that change the way you work? Neupane: Honestly, joining DialBird didn't significantly change the way I work because I was already familiar with international development practices during my time at KrispCall. Although KrispCall is a Nepal-based company, its customers are international. We already followed professional software development practices, including Scrum, daily stand-up meetings, sprint planning, retrospectives, and team collaboration. During my final seven to eight months at KrispCall, I was already handling responsibilities similar to those of a senior developer. Although I wasn't formally given the senior developer title due to internal organizational circumstances, my team trusted me with senior-level responsibilities. When the opportunity at DialBird came, I went through their technical assessment, passed it, and was hired as a Senior Android Engineer. Interestingly, KrispCall and DialBird operate in the same industry and offer competing communication products. Because of that, the transition was quite natural for me. I was already familiar with the technology, the industry, and the development processes. So I wouldn't say joining DialBird completely changed my working style. The biggest difference was that my experience and responsibilities were now formally recognized through a senior engineering position. For me, career growth isn't just about changing companies or getting a new title. It's about developing your abilities, taking responsibility, and consistently proving what you can do. Q: You have advised Nepali developers to aim for a global stage. Why does that matter to you? Neupane: I believe Nepali developers should aim for the global market because talent should not be limited by geography. Let me give you my own example. My English was never very good. Even during my school days, I struggled with it. But just like my mathematics suddenly started improving before my SLC examinations, my English has also gradually improved over time. Even today, I don't speak English like a native speaker, and I don't think I need to. I can understand English, communicate my ideas, and work with international teams. That's what matters. I think many Nepali developers underestimate themselves because of language barriers or a lack of confidence. But you don't need perfect English or a foreign accent to work internationally. You need technical skills, good communication, and, most importantly, honesty. For me, honesty and trust are extremely important. There are unreliable people everywhere, whether in Nepal or abroad. If you are honest about your abilities, take responsibility for your work, and deliver what you promise, you can build trust with international clients. Another important factor is financial growth. Many Nepali developers already work with international clients through Nepal-based companies. We follow the same development practices, use the same technologies, attend Scrum meetings, and solve similar technical problems. So my question is simple: If we are already doing work that meets international standards, why should we limit ourselves to the local market and local salary expectations? I'm not saying Nepali companies are bad or that developers shouldn't work for them. They provide valuable opportunities. But if someone has the skills to compete globally, they should explore those opportunities instead of limiting themselves geographically. Today, with AI tools like ChatGPT and other technologies, learning and communication have become easier than ever. We have more resources to improve our skills and solve problems independently. I don't believe there is some fundamental difference between a Nepali developer and a developer working in Europe, Australia, or the United States. We all work with programming languages, frameworks, software architectures, and development standards. The real difference is often access to opportunities, confidence, and financial rewards. That's why I encourage Nepali developers to think globally. Improve your skills, develop your English communication, be honest about what you can do, and don't be afraid to approach international companies or clients. You don't need to be perfect to start working globally. You need to be capable, honest, responsible, and willing to improve. I believe Nepali developers have the potential to compete internationally, and they should have the confidence to explore that potential. ### Part III: From tools to ScaScan Q: Where did the idea for ScaScan come from? Neupane: ScaScan actually started with a very simple idea. Initially, I was developing an Android scanning application inspired by apps like CamScanner. At that time, I was thinking about building a scanning tool, not an entire platform. But as I continued developing it, my thinking started to change. I realized that I didn't want ScaScan to be limited to just one application or one specific purpose. I wanted to build a brand where I could create different kinds of useful tools and products. That's when I started exploring browser-based tools that people could use for free. At that time, AI was not as advanced or accessible as it is today. ChatGPT and DeepSeek were available, but AI coding agents were not as common or capable as they are now. I used my own programming knowledge, along with the available technologies, to start building free online tools. One thing I always wanted to focus on was privacy, security, and honesty. I believe people should be able to use useful technology without unnecessarily compromising their personal information. Even the name ScaScan reflects part of that thinking. It comes from the idea of scanning, examining, and understanding things. For me, it's also connected to truthfulness. I want ScaScan to represent tools that people can trust, rather than something misleading or fake. But my vision for ScaScan is much broader than scanning. I don't want to define ScaScan as only an image-processing platform, an AI platform, or a restaurant-management product. Today, we have free browser tools and products like ScaScan Restaurant, but tomorrow it could be something completely different. I want ScaScan to be a place where I can continue experimenting, developing ideas, and turning them into useful products. Right now, I know ScaScan is still small. We haven't built anything revolutionary yet, and I don't want to pretend otherwise. But who knows what the future holds? Maybe one day, one of the most useful or innovative tools will come from ScaScan. That is the real idea behind ScaScan. I don't want to limit what it can become before I've even had the opportunity to explore its full potential. Q: Why the name ScaScan? Neupane: The name ScaScan originally came from the idea of scanning. When I first started the project, I was developing an Android document-scanning application inspired by CamScanner, so I wanted a name connected to scanning. But as my vision grew, the meaning of ScaScan also became broader. For me, scanning is not just about scanning documents or images. It's also about examining things, understanding information, and finding the truth. I wanted ScaScan to represent something people could trust. That's why privacy, security, and honesty became important parts of the brand. I also didn't want a name that would limit us to one particular technology or industry. Today, ScaScan offers free browser-based tools, AI-powered features, and products like ScaScan Restaurant. In the future, it could include many completely different products. So, although the name started with a simple document scanner, today it represents a much bigger vision. For me, ScaScan is not just the name of an application. It's a brand where we can continue exploring ideas, building useful technology, and creating products that people can trust. Q: You started with free browser tools that never upload files. Why build them that way? Neupane: The main reason was privacy, security, and trust. These have been important parts of my vision for ScaScan from the beginning. When I started building free browser tools, I had a simple question in my mind: Why should someone have to upload their personal files to a server just to perform a simple task? For example, if someone wants to crop an image, compress a photo, or convert a file, many of these operations can be performed directly in their browser. There is no need to send their personal data somewhere else. As a developer, I knew that modern browsers were capable of handling many of these tasks locally. So I wanted to use that technology to create tools that were free, simple, and privacy-friendly. Another advantage was that browser-based processing could reduce server costs. Since I was building ScaScan with limited financial resources, this approach made sense both technically and financially. But I also want to clarify something. Not every ScaScan product works entirely inside the browser. Some advanced AI features and SaaS products require server-side processing. My philosophy is that if something can be processed locally, we should try to process it locally. If a server is necessary, we should be transparent about it. For me, privacy is not just a marketing feature. It's about respecting users and their data. I want people to trust ScaScan because of how our products actually work, not just because we tell them to trust us. That's the thinking behind our free browser tools, and it's a principle I want to carry forward as ScaScan grows. Q: How do you build ScaScan alongside a full-time engineering job? Neupane: Actually, the way I manage both is quite practical. I currently work remotely as a Senior Android Engineer at DialBird. Because of my experience and the nature of my work, I can usually complete my professional responsibilities in around four to five focused hours per day. Once I finish my DialBird responsibilities, I dedicate the remaining time to ScaScan. I also have two days off each week, and I use much of that time to develop and improve ScaScan. Another important factor is AI. Today, we have powerful tools like ChatGPT and other AI coding assistants that have completely changed how independent developers can build software. I don't necessarily have to write every line of code manually anymore. My main responsibility is to come up with ideas, understand the requirements, define how a feature should work, and provide the right technical context and instructions to AI tools. They help me turn those ideas into working implementations much faster. Of course, my years of programming experience are important here. I need to understand the architecture, evaluate the generated code, identify problems, and make sure everything works correctly. AI helps with implementation, but the technical decisions and responsibility remain mine. This approach allows me to make progress on ScaScan without needing a large development team. What previously might have taken me several days of manual coding can sometimes be accomplished much faster with the right instructions and AI assistance. So, for me, building ScaScan alongside a full-time job is possible because of three things: remote work flexibility, effective time management, and the ability to combine my programming experience with AI tools. I take my responsibilities at DialBird seriously, but I also use my available time to build something of my own. That's how I'm gradually developing ScaScan while continuing my professional career. Q: You work with Raynetic Tech Solutions on billing. How did that partnership start? Neupane: The partnership with Raynetic Tech Solutions actually started through LinkedIn, and there's an interesting story behind it. When I initially started ScaScan, my focus was on providing free online tools. I was building everything independently, and at that stage, ScaScan wasn't generating any revenue. Since I'm based in Nepal, I didn't want to spend money registering and maintaining a company before the business had even started earning. I wanted to focus my limited resources on building useful products first. However, as ScaScan started expanding, I realized that free tools alone wouldn't be enough to support my long-term vision. I wanted to introduce advanced AI features, including image generation for restaurant menus, background processing, and media authentication tools. In the future, I also want to explore other products, such as VoIP and ride-sharing applications. Developing and maintaining these services requires financial resources. I realized that ScaScan needed a sustainable business model and a way to accept international payments. Interestingly, I connected with the person behind Raynetic Tech Solutions through LinkedIn. He had seen one of the videos I created and was impressed by my work. That's how our communication started. We gradually got to know each other, discussed our ideas, and eventually began working together through Raynetic Tech Solutions. I explained my situation and what I wanted to achieve with ScaScan, and he agreed to support me. That partnership helped me move toward offering paid services and handling international billing. For me, the most interesting part is that the relationship didn't begin with a formal business proposal. It started because someone discovered my work, appreciated what I was doing, and we built a connection through LinkedIn. It also taught me that sharing your work publicly can create unexpected opportunities. Sometimes, a simple video can lead to a meaningful business relationship. Today, that collaboration is helping me develop ScaScan beyond its original collection of free tools while keeping my focus on building useful products for a global audience. ### Part IV: The product Q: Who is ScaScan for, and what job does it do for them? Describe the user who gets the most value. Neupane: I believe ScaScan is for everyone who can benefit from technology. My vision is not to limit it to one type of user, application, or industry. I want ScaScan to become a technology brand that develops practical solutions for different human needs. One of my main motivations is to make technology more accessible, privacy-friendly, and efficient. Today, AI is becoming part of almost everything. People use it for image generation, background removal, photo editing, and many other tasks. But I believe not every operation needs to depend on a powerful cloud-based AI model. Modern devices already have significant processing capabilities. If we can perform certain operations directly in the browser or on the user's device, why should we unnecessarily upload personal files or depend on large remote systems? That's one of the ideas behind ScaScan. I want to explore how we can use local processing and smaller AI models wherever practical. This can help protect privacy, reduce unnecessary server usage, and potentially reduce the computing resources and environmental impact associated with some tasks. Of course, some advanced features require cloud-based AI, and I want ScaScan to support those technologies too. The goal is to choose the right technology for the problem rather than depend on the biggest model for everything. Another important part of ScaScan is helping people and businesses solve real problems. For example, ScaScan Restaurant helps restaurant owners create digital menus, generate QR codes, and manage customer orders without needing to develop their own software. But restaurants are only one example. ScaScan is not intended to be just a restaurant-management company. I want to build different applications and services under the same brand, whether they involve AI, digital media, communication, business management, or entirely new ideas. The people who get the most value from ScaScan are those who need practical technology but don't want unnecessary complexity, expensive solutions, or compromises to their privacy. I also want to keep useful tools free wherever that's financially sustainable, while offering paid services for features that require ongoing infrastructure and AI processing. Ultimately, I don't want ScaScan to be defined by one product. I want it to be defined by the problems it solves and the value it creates for people. Today, ScaScan is still small, but my ambition is to keep exploring new ideas and building useful technology that can benefit people around the world. Q: How do you view your positioning relative to MenuTiger and Menubly for restaurant menus, and iLoveIMG and TinyPNG for image tools? Is there another competitor you think of first? Neupane: I respect platforms like MenuTiger, Menubly, iLoveIMG, and TinyPNG because they have already built successful products that solve real problems. But my original thinking behind ScaScan was a little different. Let me explain how ScaScan Restaurant actually started. Initially, I wasn't planning to build a restaurant ordering system or a paid SaaS product. My idea was much simpler. I wanted to create a completely free menu builder for small restaurants. A restaurant owner could design a menu using ScaScan, export it as a PDF, upload that PDF to Google Drive, and generate a QR code linking to it. Customers could then scan the QR code and view the menu. That was the entire concept. Restaurants wouldn't need to purchase hosting, maintain a complicated system, or pay a monthly subscription just to share their menus digitally. When I explored the market, I noticed that many restaurant menu solutions were built around paid services. I wanted to offer a simpler alternative, especially for small restaurant owners who might not need an advanced management system. But something interesting happened. People weren't using my free menu builder as much as I had expected. As time passed and AI technology became more accessible, I started thinking about expanding the product. Instead of offering only a basic menu designer, I explored how AI could help restaurant owners create menus from photos, generate food images, and manage their menus more easily. Gradually, that simple offline menu builder evolved into ScaScan Restaurant, which now includes hosted digital menus, QR-based ordering, and restaurant order management. So ScaScan Restaurant wasn't originally planned as a direct competitor to MenuTiger or Menubly. It evolved from a simple free tool into a more complete restaurant solution. I still want to preserve the original idea. Restaurant owners who only need to design and export a menu should have access to free tools without being forced into a paid subscription. More advanced services, such as hosting and online ordering, can be part of our paid offerings because maintaining those services requires infrastructure and ongoing costs. The same philosophy applies to our image-processing tools. When I look at platforms like iLoveIMG and TinyPNG, I appreciate what they have built. But with ScaScan, I'm particularly interested in privacy, local processing, and efficient use of technology. I believe that if a task can be performed directly on a user's device, we should explore that option rather than unnecessarily uploading files to servers or using large cloud-based AI models. That's why many ScaScan tools are designed to process files locally in the browser wherever practical. I don't have one company that I consider my biggest competitor because ScaScan is not limited to restaurant software or image tools. Today, we have free browser tools, AI features, and ScaScan Restaurant. In the future, I want to explore other products, including communication platforms and different business applications. I'm also realistic about our position. ScaScan is still small, and established companies have much larger user bases and resources. But my goal is not simply to copy what those companies have built. I want ScaScan to become a technology brand that starts with simple human problems, provides free solutions wherever possible, and gradually develops more advanced products when there is a genuine need. That's how ScaScan Restaurant evolved, and that's the philosophy I want to continue following. Q: What is your core moat that prevents MenuTiger, iLoveIMG or anyone else from replicating your feature set? Neupane: Honestly, I don't believe there is any technology that another company cannot replicate. If MenuTiger, iLoveIMG, or another established company wants to copy one of our features, they probably can. They have more resources, larger teams, and established customer bases. But I don't think building a successful technology company is just about having features that nobody else can copy. For me, ScaScan's strength comes from the way I approach problems. For example, when I initially developed our menu builder, I wasn't thinking about subscriptions or competing with restaurant SaaS companies. I simply wanted small restaurant owners to design a menu for free, export it as a PDF, upload it to Google Drive, and share it through a QR code. Later, that product evolved into ScaScan Restaurant, with AI features, hosted menus, and an ordering system. But the original free menu-building concept still exists. Similarly, with our image-processing tools, my first question is whether we can solve the problem directly on the user's device. If we can, why unnecessarily upload their files to a server or use expensive cloud AI? That's why privacy, efficient computing, and accessibility are important parts of ScaScan. Another advantage is my development experience. I've worked with Android, Flutter, and different software systems for years. Today, I combine that experience with AI coding tools to turn ideas into working applications much faster than I could through manual development alone. I don't need to build a large team every time I want to experiment with a new idea. And unlike a company focused on one specific product category, I don't want to limit ScaScan to restaurant management or image processing. I want to explore different problems and develop solutions wherever I see opportunities. However, I also understand that these things are not an unbeatable competitive moat. Other developers can use AI, build privacy-focused tools, and launch new products too. ScaScan is still at an early stage. We haven't yet built the large customer base, brand recognition, or ecosystem that would make it difficult for someone to compete with us. I believe those advantages have to be earned over time by building reliable products, understanding users, improving continuously, and gaining their trust. So, I wouldn't say my moat is a feature that competitors cannot copy. My goal is to build a company that people continue to choose because they trust its products, appreciate its approach, and find genuine value in what it creates. Features can be copied. Building and maintaining that trust is a much longer process, and that's what I want to focus on with ScaScan. Q: What are some of the use cases for ScaScan that are not obvious and can add a lot of value to users? List as many as possible. Neupane: I think the interesting thing about ScaScan is that even a simple tool can solve problems beyond its original purpose. For example, our background remover isn't useful only for photo editing. A small online seller can use it to prepare product images without paying a professional designer. Our image privacy tools can help people check whether a photo contains sensitive metadata, such as location information, before sharing it publicly. Our media analysis tools can help journalists, researchers, and ordinary internet users investigate potentially AI-generated or manipulated content. Of course, AI detection isn't always conclusive, but these tools can provide useful signals. Another interesting example is ScaScan Restaurant. Initially, I built a completely free menu builder. A small restaurant could design a menu, export it as a PDF, upload it to Google Drive, and generate a QR code. It didn't need to purchase an expensive restaurant management system just to offer a digital menu. Later, that evolved into a more advanced restaurant platform with AI-assisted menu creation, hosted menus, and customer ordering. But even the free menu builder can be useful beyond restaurants. Small cafes, home-based food businesses, catering services, and temporary food stalls could use similar digital menus. I also see opportunities in education, freelancing, and small businesses. Students can use image and document utilities, freelancers can prepare digital assets, and small businesses can create QR codes and other materials without purchasing complicated software. One area I'm particularly interested in is using the computing power people already have. Today, almost everyone talks about using larger AI models, but I believe some tasks can be completed directly on users' devices or with smaller models. That can help protect privacy, reduce operating costs, and avoid unnecessary cloud processing. These are only a few examples. My broader vision is that ScaScan shouldn't be limited to the obvious purpose of each tool. I want to keep exploring how existing technology can be used to solve different human problems. Sometimes, a simple free tool can provide more practical value to someone than an expensive, complicated application. That's the kind of opportunity I want ScaScan to explore. Q: The Menu Builder turns a photo of a printed menu into a live QR menu. What was hardest to get right? Neupane: Actually, converting a photo of a printed menu into a digital menu is not the most difficult part anymore, especially with how powerful AI models have become. I think the interesting part is how I designed the system from the beginning. When I initially developed ScaScan Menu Builder, it was completely offline. There was no server database or hosted ordering system. I simply wanted restaurant owners to create their menus, design them, export them as PDFs, and share them through QR codes. But because of my software engineering experience, I decided to use a structured JSON format to manage the menu data from the beginning. Everything, including categories, menu items, descriptions, prices, and image references, was organized in JSON. That architectural decision became very useful later. When I started introducing AI-powered menu scanning, I didn't need to completely redesign the application or change its underlying data structure. The process is quite straightforward. A restaurant owner uploads a photo of a printed menu. AI analyzes the image and extracts the information, including food names, categories, descriptions, and prices. That information is then converted into the same JSON structure the Menu Builder already understands. Once the JSON is generated correctly, the existing system can use that data to create the digital menu, apply different designs, and prepare it for publishing. That's the beauty of the architecture. The Menu Builder doesn't really care whether the JSON was created manually by the restaurant owner or generated automatically using AI. Of course, AI extraction isn't always perfect. Different menu layouts, image quality, and unclear text can affect the results, so users still need the ability to review and correct the information. But the most important thing is that I didn't have to rebuild the entire system when AI became more capable. If you ask me what was hardest to get right, I would say it was creating a flexible foundation that could evolve without breaking the original product. At the beginning, I was just building a free offline menu designer. I didn't know it would eventually become an AI-powered menu creation and restaurant ordering system. Looking back, choosing JSON as the common data structure was one of the most useful technical decisions I made. It allowed me to introduce new capabilities while keeping the core architecture largely unchanged. For me, that's an important lesson in software engineering: sometimes a simple architectural decision made early can make future development much easier. Q: You also built an AI keyboard and UpHabitAi. How do they fit with ScaScan? Neupane: Both ScaScan AI Keyboard and UpHabit AI fit naturally into my vision for ScaScan because, for me, ScaScan is about using technology to improve people's lives. I never wanted ScaScan to be limited to one application or one industry. My goal is to build different tools that help people solve problems, improve their abilities, and develop better habits. For example, ScaScan AI Keyboard focuses on improving communication. It helps people write better using AI-powered features such as grammar correction, translation, professional tone, and text improvement. I believe communication is an important part of personal and professional growth. Not everyone is confident in English or writing, but that shouldn't prevent them from expressing their ideas effectively. ScaScan AI Keyboard helps make that process easier. UpHabit AI takes a different approach. It focuses on understanding and improving people's habits. The idea behind UpHabit AI is to analyze people's habits and behavioral patterns, identify areas where they could improve, and provide useful AI-powered suggestions. Instead of simply tracking habits, I want it to help people understand their behavior and make positive changes in their daily lives. That's also where the name ScaScan connects with the product. Just as scanning helps us examine and understand something, UpHabit AI applies that idea to personal habits. It helps people examine their routines, recognize patterns, and discover opportunities for improvement. Although these two applications solve different problems, they share the same underlying philosophy. ScaScan AI Keyboard helps people improve how they communicate, while UpHabit AI helps people improve how they live and manage their habits. I don't believe AI should exist only to generate content or automate tasks. I believe it should also help people understand themselves, develop useful skills, and make better decisions. Ultimately, my vision for ScaScan is not just to build applications. It is to build technology that genuinely benefits people. That's why both ScaScan AI Keyboard and UpHabit AI belong within the ScaScan vision. They are different products, but they work toward the same goal: helping people improve through technology. Q: Your principle is that AI should strengthen human judgment, not replace it. Where does that show in your products? Neupane: I believe AI should help people understand things better and make more informed decisions, but it should not completely replace human judgment. One example of this philosophy is ScaScan Media Signal Analyser, which helps people investigate whether images and videos may be AI-generated or manipulated. Today, AI-generated content is becoming so realistic that even humans sometimes struggle to distinguish between real and artificial media. And interestingly, AI detection systems themselves are not always accurate. That's why I don't believe we should depend on a single AI model to decide whether something is real or fake. My approach is to build a verification process with multiple stages. First, our media analysis tool uses technical libraries and algorithms to examine different signals in an image or video. This initial analysis can run locally on the user's device, supporting our privacy-first approach. If the result is uncertain or suspicious, the next stage involves more advanced AI analysis through external AI services, such as Google's AI models. But even then, AI cannot guarantee that its conclusion is correct. That's where human judgment becomes important. My longer-term plan is to introduce a human verification stage for cases where automated analysis cannot reach a reliable conclusion. A human reviewer could examine the evidence, investigate the source, and consider the context before reaching a more informed assessment. The important thing is that machines and humans have different strengths. Computers can analyze large amounts of technical information quickly, while humans can consider context, intention, and circumstances that automated systems might miss. The browser-based analysis is already working, while the broader AI-assisted and human-review workflow is something I'm continuing to develop. And this philosophy is not limited to media analysis. I want to apply it across future ScaScan products. I don't want AI to become a system that simply tells people what to believe. I want it to become a tool that helps people investigate, understand, and decide for themselves. For me, that is the real meaning of using AI to strengthen human judgment rather than replace it. ### Part V: Launch Q: When did ScaScan go live, and what happened in the first weeks? Neupane: When I first launched ScaScan, honestly, almost nothing happened. I had built the website and made the tools available to the public, but there were hardly any visitors. Nobody really knew about ScaScan. The biggest challenge was marketing. I was building everything independently, and I didn't have a proper marketing budget. Even today, my marketing resources are very limited. I also realized that simply launching a website doesn't mean people will automatically discover it. Google indexing was slow, and only around 35 to 36 pages were getting indexed, despite having more content and tools available. So the first few weeks were very quiet. There was no sudden growth, no large number of users, and no exciting success story to share. But I kept trying. I continued developing tools, improving the website, working on SEO, and exploring different ways to reach people. Sometimes I would make changes and still see almost no results, but I didn't stop. Actually, I think this is something many people don't understand about building a startup. Developing a product and attracting people to use that product are two completely different challenges. As a software engineer, I know how to build applications. But getting people to discover those applications is another skill that I'm still learning. ScaScan is still at an early stage, and attracting users remains one of my biggest challenges. I'm not going to pretend that everything became successful immediately after launch. The reality is that ScaScan didn't grow because of a successful launch. It is still moving forward because I keep trying, experimenting, and improving it, even when the results are very small. For me, that's what the first few weeks taught me: building something is only the beginning. Getting people to know about it and trust it is the real challenge. Q: Milestone one was paid plans through Paddle. Who were your first paying users? Neupane: Actually, we haven't acquired our first paying customers yet, and I want to be honest about that. Integrating Paddle was an important milestone for ScaScan because it allowed us to move beyond offering only free tools and start building a sustainable business model. Initially, ScaScan was completely focused on free browser-based tools. But as I started introducing advanced AI features and products like ScaScan Restaurant, I realized that some services would require ongoing infrastructure, AI processing, and hosting costs. That's why I decided to introduce paid plans. However, developing a product and finding paying customers are two different challenges. As a software engineer, I can build the application, integrate payment systems, and implement technical features. But marketing and customer acquisition are areas I'm still learning. Since ScaScan is bootstrapped and my marketing budget is very limited, attracting our first customers has been difficult. Right now, my focus is on improving the products, understanding what users actually need, and finding ways to reach the right audience. I don't want to claim that ScaScan is already a successful business just because we've implemented a payment system. For me, Paddle integration was a technical milestone. Getting our first genuine paying customer will be the real business milestone. And that's something I'm still working toward. Q: ScaScan was accepted into the Google for Startups Cloud Program. What will you use it for? Neupane: Getting accepted into the Google for Startups Cloud Program was an important milestone for ScaScan, especially because I'm building everything with very limited financial resources. We received around $2,000 in cloud credits, which may not sound like a huge amount for an established company, but for a bootstrapped project like ScaScan, it is valuable support. My main goal is to use these resources to develop and experiment with more advanced AI-powered products. For example, one of the products I'm currently working on is ScaScan Media Signal Analyser, which helps people investigate whether images and videos may be AI-generated or manipulated. The initial analysis already works using browser-based technical libraries. But I want to introduce additional AI verification using more powerful models, including Google's AI technologies, when the initial analysis is uncertain. My longer-term vision is to combine technical analysis, AI verification, and human judgment to create a more reliable verification process. Another area is AI-powered image processing and content generation. We already have features such as AI-assisted menu creation and image generation, and I want to continue improving them. However, I don't want ScaScan to become completely dependent on expensive cloud AI. My philosophy is that if something can run locally on a user's device, we should try to process it there. But when we genuinely need powerful AI models or cloud infrastructure, that's where Google Cloud can help us experiment and scale. The credits also give me an opportunity to test new ideas without immediately worrying about every infrastructure expense. And I want to clarify something: these are cloud credits, not direct financial investment or revenue. We still have to build useful products, attract users, and eventually generate sustainable income. For me, the biggest benefit is the freedom to experiment. ScaScan is not limited to one application. I want to continue exploring different technologies and developing products that genuinely benefit people. I don't know which ScaScan product will become the most successful in the future. But support like this gives me more opportunities to explore ideas, test them, and hopefully turn some of them into something valuable. ### Part VI: The hard part Q: ScaScan is bootstrapped. How do you decide where to spend your own money and time? Neupane: Actually, because ScaScan is completely bootstrapped, I have to be very careful with both my money and my time. I don't have investors or a large financial budget behind ScaScan. I'm still working as a Senior Android Engineer at DialBird, and my income from that job helps me continue developing my own products. So I try to spend money only where it is necessary. For example, when I started ScaScan, I focused on free browser-based tools. One reason was privacy, but another advantage was cost. If a tool can process an image or perform a task directly on the user's device, I don't need to pay for expensive server infrastructure. That allows me to provide useful tools for free while keeping operating costs low. I also didn't want to spend money registering and maintaining a company before ScaScan started generating revenue. Later, when I needed international billing for paid services, I received support through my partnership with Raynetic Tech Solutions. Time is another important resource. Since I work remotely at DialBird, I can usually complete my professional responsibilities in around four to five focused hours per day. I dedicate much of my remaining time, including my two weekly days off, to ScaScan. Another major advantage is AI. I use ChatGPT and other AI coding assistants to help turn my ideas into working applications. Because I already have years of software engineering experience, I can define the architecture, explain the requirements, review the implementation, and solve technical problems without needing to hire a large development team. I'm not saying AI does everything perfectly. But it has significantly reduced the time and money I need to experiment with new ideas. Q: How many people use ScaScan today? Neupane: Right now, ScaScan is still at an early stage. I would estimate that around 100 to 500 people use or visit our free tools each month. Most of our traffic comes from people discovering the tools through search engines or other online sources. Since our marketing budget is very limited, we're still working on improving our visibility and reaching more users. At the moment, we haven't acquired any paying customers. Although we've introduced paid plans through Paddle and launched products like ScaScan Restaurant, converting visitors into paying customers is still a challenge. I don't want to exaggerate our numbers or pretend that we're growing faster than we actually are. For now, my focus is on improving the products, attracting more users, and understanding what people genuinely find useful. We have a small number of free users, no paying customers yet, and a lot of work ahead of us. But ScaScan is live, the products are working, and I'm continuing to build and improve them. Q: You are exploring restaurant technology and media authenticity at the same time. How do you choose between them? Neupane: Actually, I don't see restaurant technology and media authenticity as two competing ideas. They are different products, but both fit into my broader vision for ScaScan. As I mentioned earlier, I don't want ScaScan to be limited to one industry. My goal is to build technology that solves real problems and benefits people. For example, ScaScan Restaurant started as a simple free offline menu builder. I initially wanted to help small restaurants create digital menus without paying for expensive software. But as I continued developing it and AI technology improved, I expanded it into a restaurant ordering and management system. On the other hand, ScaScan Media Signal Analyser came from a completely different problem. Today, AI-generated images and videos are becoming increasingly realistic. Sometimes, even humans cannot easily identify whether content is authentic or AI-generated. I believe this will become an important challenge as AI continues developing. That's why I'm working on a system that combines technical analysis, AI verification, and eventually human judgment to help people investigate digital media more effectively. So how do I choose between them? I think the answer is practical value, available resources, and opportunities for growth. Both products are already working in their current forms, although some advanced features are still under development. Since I'm building ScaScan independently, I have to manage my time carefully. I don't want to spend all my resources developing one product while completely abandoning the others. At the same time, I understand that I cannot give every product equal attention. I want to observe how people use these products, understand their problems, and gradually focus more resources on the areas where I see genuine demand and opportunities. AI coding tools also make this approach more practical. Because of my software engineering background, I can use AI assistance to develop and improve different products without maintaining a large development team. But I also recognize that building products is not the same as building a successful business. Right now, attracting users and generating revenue are still major challenges for ScaScan. For me, it's not about choosing restaurant technology over media authenticity, or vice versa. It's about identifying which problems are worth solving and which solutions people genuinely find valuable. If one product starts attracting more users or shows stronger commercial potential, naturally I will give it more attention and resources. But I don't want to permanently limit ScaScan to whichever product becomes successful first. My long-term vision is to create a technology brand where different products can exist under the same philosophy: making technology useful, accessible, privacy-conscious, and beneficial to people. Q: What is a challenging time you had building this company? Neupane: Honestly, the biggest challenge I've faced while building ScaScan is money. As a software engineer, I have confidence in my technical abilities. I've spent years developing applications, working with different technologies, and solving complex problems. Today, with AI tools like ChatGPT and other coding assistants, I feel even more confident about turning my ideas into working products. I have so many ideas for ScaScan, and I believe I can build many of them. But building something and having the financial resources to grow it are two different things. For example, I can develop an application, deploy it, and make it available to the public. But after that, I need users. Reaching those users requires marketing, and marketing requires resources that I currently don't have. Even today, ScaScan has only around 100 to 500 estimated monthly free users, and we haven't acquired our first paying customer. That's not because the applications aren't working. The bigger problem is that many people simply don't know ScaScan exists. Another challenge is the cost of infrastructure, AI services, and maintaining products as they grow. Although I try to reduce these expenses through browser-based processing, efficient architecture, and AI tools, some costs are unavoidable. That's also why support like the Google for Startups Cloud credits is valuable to me. If I had sufficient financial resources, I believe I could develop and expand ScaScan much faster. I could invest in marketing, infrastructure, and the people needed to support different products. Of course, money doesn't guarantee success, and building a business involves challenges beyond programming. But right now, financial limitations are the biggest obstacle between my ideas and what I want ScaScan to become. I don't feel limited by the number of ideas I have or my ability to build software. I feel limited by the resources available to bring those ideas to people. That's the most challenging part of building ScaScan as a bootstrapped founder. ### Part VII: What it is all for Q: Who believed in ScaScan before there was anything to show? What did they say or do? Neupane: Honestly, no one supported ScaScan in the beginning. There wasn't anyone encouraging me, investing in my idea, or telling me that ScaScan would become successful. I started ScaScan myself because I felt that I wanted to contribute something meaningful to the world. I've been working as a software engineer for years, building applications and products for different companies. But at some point, I started thinking: Why shouldn't I use my knowledge and experience to create something of my own that could benefit people? That's where my motivation came from. I didn't start ScaScan because someone believed in my idea or promised to support me. I started because I believed I could build something useful. Even today, ScaScan is still small. We have limited users, no paying customers yet, and very little money for marketing. But I continue developing it because I believe the idea is worth pursuing. I don't know whether ScaScan will become a big company in the future. Maybe it will, maybe it won't. But I want to give my ideas a real opportunity rather than spend my life wondering what would have happened if I had tried. So, if you ask me who believed in ScaScan before there was anything to show, my answer is simple: I did. And that belief was enough for me to get started. Q: Why does media authenticity matter to you, and why keep a human reviewer in the loop? Neupane: Media authenticity matters to me because I believe people deserve to know whether the information they see online is trustworthy. Today, AI can generate incredibly realistic images and videos. Sometimes, even advanced AI models cannot confidently determine whether something is real or AI-generated. That creates a serious problem because people can be misled by manipulated content without even realizing it. That's why I'm developing ScaScan Media Signal Analyser. But my approach is a little different. I don't want to depend entirely on one AI model to decide whether something is real or fake. My idea is to create a step-by-step verification process. First, we assume that someone is using our tool because they genuinely don't know whether an image or video is authentic. The first stage uses browser-based libraries and technical analysis to examine metadata, file characteristics, and other signals. If the system finds strong evidence, it can explain those findings to the user. But what happens if the metadata has been removed, or the technical analysis cannot reach a reliable conclusion? That's where the second stage comes in. Users can choose to continue with a more advanced investigation using powerful cloud-based AI models. Because these models require additional computing resources and API costs, this stage can be offered as a paid service. However, even advanced AI models cannot guarantee accurate detection. An image might be AI-generated without containing obvious technical signals, or a genuine image might look artificial to an automated detector. So if the system still cannot provide a reliable assessment, the user can choose a third stage: human review. Human review is the final stage, not the first. A human reviewer can examine the available evidence, investigate the source, consider the context, and provide an assessment. Of course, even human experts cannot always determine authenticity with complete certainty, but their judgment can add another valuable perspective. That's the verification process I'm working toward. The initial browser-based analysis is already available, while the advanced AI and human-review workflow is part of the system I'm continuing to develop. But I also want to explain that my vision for ScaScan goes beyond detecting AI-generated media. I believe the internet is becoming increasingly difficult to navigate because people are constantly exposed to misleading, harmful, or unwanted content. In the future, I want to explore whether ScaScan can help people take greater control over the content they consume. For example, imagine an AI-powered system that understands a user's content preferences and helps filter unwanted material across different digital platforms. I know that building something like this would be technically challenging. Android, iOS, and other platforms place restrictions on how applications can access information from other apps. Privacy, security, and legal requirements would also need to be considered carefully. But I don't think those challenges mean we should stop exploring the idea. Maybe it would require a new type of application, deeper operating-system integration, or even developing our own platform someday. Of course, that would require significant financial resources, research, and development. I'm not saying ScaScan will build this next year or even within five years. It might take ten or twenty years, or the idea might evolve into something completely different. My point is that I don't want to limit ScaScan's future based on what is technically or financially possible for me today. For me, media authenticity is just one part of a much bigger vision: using technology to help people understand what they see, protect their privacy, and have more control over their digital lives. I believe AI should support human judgment, not replace it. Whether we're investigating a suspicious video or helping people manage their online experiences, the final goal should always be to give humans more understanding, more choice, and more control. Q: What would make you say ScaScan has succeeded? Neupane: For me, ScaScan's success is not just about making money or becoming a big company. Of course, financial success is important because without money, we cannot continue developing products, maintaining infrastructure, or investing in new technologies. But my original reason for creating ScaScan was different. I wanted to contribute something meaningful to the world through my programming knowledge. I have always believed that technology should benefit people. That's why I started with free browser-based tools, focusing on privacy, security, and accessibility. Today, ScaScan is still small. We have limited users, no paying customers yet, and very little money for marketing. But I have a much bigger vision for what it could become. Imagine someone using ScaScan AI Keyboard to communicate more confidently, a small business using our free tools to save money, or someone using our media analysis technology to investigate potentially misleading content. Even these small improvements in people's lives mean something to me. In the future, I want ScaScan to create many different products. Some might help businesses, some might improve people's daily habits, and others might address problems we haven't even identified yet. I also want ScaScan to become financially independent so that I can spend more time developing ideas, supporting research, and building technologies that aren't possible with my current resources. Maybe one day ScaScan will create something that makes a significant contribution to society. I don't know what that product will be or how long it will take. It might take five years, ten years, or even twenty years. But I don't want to measure success only by the number of applications we build or the revenue we generate. I would say ScaScan has truly succeeded when people around the world are benefiting from our technology, when they trust our products, and when the company can sustain itself while continuing to innovate. I started ScaScan because I wanted to contribute something to the world. If, one day, I can look back and see that ScaScan genuinely improved people's lives, I will consider that my biggest achievement. ### Part VIII: The road ahead Q: Over the next 6 to 12 months you plan restaurant pilots and media verification tests. What do you need to see to keep going? Neupane: Actually, I don't think I need a huge number of users or a lot of revenue within the next 6 to 12 months to continue building ScaScan. I started ScaScan with almost no financial support, and I'm still developing it because I believe I can contribute something useful to the world. But I do want to see whether the products I'm building are solving real problems for people. For ScaScan Restaurant, I want to see real restaurant owners using the system in their daily operations. I want to understand whether the AI menu creation actually saves them time, whether customers are comfortable with QR ordering, and whether the order management system makes running their restaurants easier. I already know that not every customer wants to order through a QR code. That's why I'm also thinking about solutions that can support traditional ordering methods alongside digital technology. For ScaScan Media Signal Analyser, my expectations are different. I want to test how accurately the system can identify suspicious media signals and, more importantly, how it handles uncertainty. My planned process involves browser-based analysis first, optional advanced AI verification when the initial result is inconclusive, and eventually human review when additional investigation is needed. I want to see whether that approach genuinely helps people understand the authenticity of digital content without giving them false confidence. Q: What are your milestones for the next 12 months, in users and revenue? Neupane: Actually, I want to keep my expectations realistic because ScaScan is still a very early-stage, bootstrapped project. Right now, we have approximately 100 to 500 monthly free users or visitors, and we haven't acquired our first paying customer yet. Over the next 12 months, my initial goal would be to grow ScaScan to around 1,000 to 3,000 monthly users and attract our first 10 to 30 paying customers. In terms of revenue, reaching around $100 to $500 in monthly recurring revenue would be an encouraging first milestone. I know those numbers might sound small compared to established startups, but for someone building independently with very limited marketing resources, they would represent meaningful progress. My biggest challenge right now is not developing software. It's getting people to discover ScaScan and convincing them that our products provide enough value to use regularly or pay for. I want to focus on improving our Google search visibility, understanding how people use our free tools, and introducing our paid products to the right audiences. For ScaScan Restaurant, I want to see real businesses adopting the system. For our media analysis tools, I want to see whether people find our verification approach useful and trustworthy. But I also don't want to measure everything only through revenue. If thousands of people are benefiting from our free tools, that is valuable too. It means ScaScan is contributing something useful, even if those users never pay. Q: And for the next 24 months? Neupane: Over the next 24 months, I want ScaScan to move beyond the experimental stage and become a sustainable technology business. My goal would be to reach around 5,000 to 10,000 monthly users across our products and attract approximately 50 to 150 paying customers. In terms of revenue, I would like to see ScaScan generating around $1,000 to $3,000 in monthly recurring revenue. For me, reaching that level would be an important achievement because ScaScan is currently bootstrapped, and I'm financing its development through my own engineering career. I also want to see ScaScan Restaurant being used by real restaurants, not just as a digital menu builder but as a practical ordering and management solution. For our media authenticity technology, I want to advance the verification process by combining browser-based analysis, more powerful AI models, and eventually human review. But my ambition is not limited to these products. I want to continue exploring new ideas under the ScaScan brand. The most important milestone would be achieving enough consistent revenue to reinvest in marketing, infrastructure, research, and future product development. I also understand that these are goals, not guaranteed results. If growth is slower than expected, I'll continue learning and improving. Within 24 months, I want ScaScan to demonstrate that it can attract users, generate sustainable revenue, and develop products that genuinely benefit people. My long-term vision is much bigger than two years, but reaching these milestones would give me the financial foundation to pursue it. Q: Where do you want ScaScan to be in five years? Neupane: In five years, I want ScaScan to become a recognized global technology brand, not just a website offering free tools or a company focused on one particular product. From the beginning, my vision for ScaScan has been much bigger than a single application. I want to create an ecosystem where we can continuously develop different products that solve real problems and improve people's lives. Today, we have free browser-based tools, ScaScan Restaurant, AI-powered media analysis, and other applications. But these are just the beginning. Over the next five years, I want to explore many more areas, including AI, communication technology, business management, and other services that can benefit people. I also want ScaScan to become financially independent. Right now, I'm supporting its development through my full-time engineering job. But in five years, I hope ScaScan will generate enough revenue to support its own infrastructure, research, marketing, and product development. With sufficient financial resources, I would also like to build a team of talented engineers and researchers who share my vision. Another important goal is to continue focusing on privacy, security, honesty, and efficient technology. I believe not every problem requires a huge AI model or expensive cloud infrastructure. Sometimes, a smaller model or an application running directly on someone's device can solve the problem more efficiently. I want ScaScan to continue exploring those possibilities. And I don't want to limit my ambition to what seems possible today. Maybe in five years, ScaScan will have dozens of useful applications. Maybe one of those products will solve a problem in a way that hasn't been done before. I can't predict exactly what will happen. But I would love to see people from different countries using ScaScan products in their daily lives. I also want to prove that someone from Nepal, without a wealthy background, a large investment, or an established business network, can build technology that reaches people around the world. My dream is not simply to build a company that makes money. I want to build a company that makes a meaningful contribution to the world. Q: What would you build through ScaScan if money were no longer a limitation? Neupane: If money were no longer a limitation, I would explore much bigger problems than the ones ScaScan is solving today. One idea I'm particularly interested in is giving people more control over the digital content they consume. Today, people are exposed to enormous amounts of content online, including misleading information, AI-generated media, and material they may not want to see. I imagine a future where AI could help people understand, verify, and filter content according to their own preferences, while respecting their privacy and freedom of choice. Building something like that might require deeper operating-system integration, new infrastructure, or even developing our own platform. I know it's not something I can easily build today. It might take ten or twenty years. But that's why I created ScaScan as a technology brand rather than limiting it to one application. I want the freedom to explore ambitious ideas, even when they seem impossible with my current resources. For me, money would not create the ideas. I already have ideas. It would give me more opportunities to turn them into reality. Q: What are your hopes and fears for a post AGI world? Neupane: I have both excitement and concerns about a post-AGI world, but personally, I am more excited about the possibilities. As a software engineer, I have already experienced how much AI can change the way we work. Today, with tools like ChatGPT and other AI coding assistants, I can turn my ideas into working applications much faster than before. I'm building ScaScan independently while continuing my full-time engineering job. Without these AI tools, developing so many different products would require significantly more time, money, and people. Now imagine what could become possible if we achieve AGI. My biggest hope is that intelligence and opportunity will become more accessible to everyone. For example, someone growing up in a financially struggling family in Nepal might have brilliant ideas but lack money, education, or connections. In a post-AGI world, I hope that person will have access to powerful technology that helps them learn, create, and compete globally. I also hope AGI will help humanity solve problems in healthcare, education, scientific research, environmental protection, and many other areas. But I also have concerns. One of my biggest fears is that people may become too dependent on AI and gradually stop using their own judgment. We are already seeing how difficult it can be to distinguish between real and AI-generated images or videos. If AGI makes synthetic content even more convincing, how will ordinary people know what to believe? This is one reason I'm developing ScaScan Media Signal Analyser. I believe we need technology that helps people investigate information rather than blindly trusting automated conclusions. Another concern is inequality. If only a few powerful companies or wealthy individuals control advanced AI, the benefits might not reach everyone. I would rather see a world where people from countries like Nepal have opportunities to use and develop advanced technology too. Privacy is another important issue. I don't want a future where every human activity, decision, or personal detail must be shared with a powerful AI system. And although AGI could automate many jobs, I hope society will find ways to help people adapt and benefit from the increased productivity rather than leaving them behind. But my biggest concern is losing human control over important decisions. Even if machines become more intelligent than humans in many areas, I don't believe that means we should automatically give them authority over everything. About ScaScan: ScaScan is a privacy-first technology brand from Kathmandu, Nepal. It offers 25+ free browser tools for images, QR codes, metadata and developers that process files on the user's device where possible, and ScaScan Restaurant, a menu builder that turns a photo of a printed menu into a hosted QR menu with ordering. Its other products include ScaScan AI Keyboard, UpHabit AI and the Media Signal Analyser. Free tools, with Pro and Ultra plans for AI features. ScaScan is built by Machhindra Neupane. scascan.com ## Multifactor (USA) URL: https://startup.talk/usa/multifactor/ Founders: Vivek Nair Vivek Nair was born in Phoenix and grew up in Austin, Madison and Singapore, with a computer as the one constant. An internship at a Singapore health-tech startup put him on the front line of SIM-swapping attacks, led to his first cybersecurity patent and pulled him into authentication for good. He earned a Ph.D. in computer science at UC Berkeley at 22, building the multi-factor key derivation function (MFKDF), and went on to technical lead roles in cyber units of the CIA and the Department of Defense. Multifactor, backed by Y Combinator and a $15 million seed round led by Nexus Venture Partners, lets you share any online account with a person or an AI agent by sending a link. The other side never sees the password, works only within the permissions you set, every action is logged, and access ends the moment you delete the link. We spoke about why AI agents need their own keys, how the company earns trust without asking for it, and what security looks like when agents log in on our behalf. Key takeaways: - Multifactor shares accounts by link, not by password: the recipient uses the site within set permissions and never holds the credential. - Passwords are encrypted on the device, so the company's servers only ever see ciphertext. - The product rests on a decade of peer-reviewed research and seven granted U.S. patents, including one on sharing online accounts. - AI agents are the pull: most early interest came from people who wanted to give an agent scoped, logged, revocable access. The Principal / Founder: - Vivek Nair (Founder) Based: San Francisco, California Grew up: Phoenix, Austin, Madison and Singapore Studied: Ph.D., Computer Science, UC Berkeley Before: Technical lead in cyber units of the CIA and the Department of Defense Funding: $15M seed led by Nexus Venture Partners; Y Combinator F25 Website: multifactor.com LinkedIn: Vivek Nair (https://www.linkedin.com/in/nair/) Company LinkedIn: multifactorcom (https://www.linkedin.com/company/multifactorcom) GitHub: github.com/multifactor X: @multifactorcom (https://x.com/multifactorcom) Instagram: @multifactorcom (https://www.instagram.com/multifactorcom) YouTube: @multifactorcom (https://www.youtube.com/@multifactorcom) Facebook: multifactorinc (https://www.facebook.com/multifactorinc) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Nair: I was born in Phoenix, Arizona, and grew up all over the world, including Austin, Madison and Singapore. Moving that often meant starting over a lot, and a computer was the one thing that came with me everywhere. Q: What first got you into building things? Nair: Coding as a kid. I taught myself to program early, and once I saw that I could make something and put it in front of other people, I didn't want to stop. Q: What was your first job, and what did it teach you? Nair: My first real job was as an intern at Holmusk, a healthcare tech startup in Singapore. It was small enough that we all had our hands in everything. It taught me that at a startup the problem in front of you is your job, whatever your title says. ### Part II: The road to Multifactor Q: How did you first get into computer security? Nair: At Holmusk I ended up helping fight SIM-swapping attacks. I patented my first cybersecurity technology out of that work, and I realized I was more interested in security than in healthcare tech. I have been working on authentication ever since. Q: What did your PhD research at Berkeley focus on? Nair: Two things. The first was applied cryptography for authentication: I built the Multi-Factor Key Derivation Function (MFKDF), which derives an encryption key from several authentication factors at once, so that a password alone is never enough. The second was privacy in virtual reality. We showed that head and hand motion alone can identify a person among more than 50,000 users, and then built tools to protect people from that. Q: What can you share about how your work at the CIA and the Defense Department shaped how you think about security? Nair: I can't discuss the work itself. What I can say is that it taught me to assume every system will be attacked by someone capable and patient, and to design so that a single stolen credential is never the end of the story. Q: Why leave government research to start a company? Nair: Two reasons. Government work protects a small number of systems, and I wanted the same ideas in everyone's hands. And the timing was right: AI agents had started needing access to our accounts, and nobody had a safe way to give it to them. ### Part III: The product Q: Why the name Multifactor? Nair: It comes from the research the product is built on, multi-factor key derivation. It also states the principle: security should never rest on a single factor. Q: Why does account access need to work for AI agents as well as people? Nair: Agents are starting to do real work inside our accounts, and today the only way to let one in is to hand over the password. That gives the agent everything, forever, with no record of what it did. An agent should get a link with exactly the permissions it needs, a log of every action, and access you can end in one click. Q: How does secure sharing work without exposing the password itself? Nair: You save the password once, and it is encrypted on your device before it reaches us. When you share an account, the other person gets a link, not the password. They use the site through that link, within the permissions you set, and they never see or hold the credential. Delete the link and their access ends at once, with no password reset. Q: How do you earn trust when you ask people to store their secrets with you? Nair: By not asking them to trust us. Our servers only ever see ciphertext, so even we can't read your passwords. The cryptography behind it is peer-reviewed and patented, and we are incorporated as a public benefit corporation, so the mission to make the internet safer is written into our charter. Q: Who are your first users: teams, developers or families? Nair: People who use AI agents, and the developers who build them. They feel the problem most sharply, because an agent can't do much until it can get into your accounts. Q: What do you deliberately leave out of Multifactor? Nair: Everything but sharing. We cut the product down to the one thing we care about most: sharing an online account by sending a link, with permissions you choose, a live log of every action, and instant revocation. ### Part IV: Positioning Q: Who is Multifactor for, and what job does it do for them? Describe the person who gets the most value. Nair: Anyone who needs to let someone else into an account without giving it away. The person who gets the most value is someone handing work to an AI agent, an assistant or a contractor: they need to give real access, limit what it can do, see what happened, and take it back. Q: How do you view your positioning relative to 1Password, Bitwarden, LastPass and Dashlane? Is there another competitor you think of first? Nair: Those products store passwords and, when you share, they hand the password to the other person. We start from the opposite end: the recipient never gets the password at all. You can share code with a link on GitHub, files with a link on Dropbox and videos with a link on YouTube. Why can't you share an online account with a link? That is the product. Q: What is your core moat that prevents 1Password or Bitwarden from replicating your feature set? Nair: A decade of research. The product is built on cryptography we published at top security conferences, and we hold seven granted U.S. patents, including one on sharing online accounts and one on multi-factor key derivation. Q: What are some of the use cases for Multifactor that are not obvious and can add a lot of value to users? List as many as possible. Nair: - Let an assistant read your email without being able to send one. - Let an accountant view bank transactions without being able to make a transfer. - Give an AI agent read-only access to an account so it can automate a task. - Share a streaming account with a friend without ending up signed out of it yourself. - End a contractor's access the moment the job is done, with no password change. - See exactly what was done in a shared account, including what was blocked. Q: What is the one detail in Multifactor you are proudest of? Nair: How easy and fast it is. You can sign up and share any online account with a person or an AI agent just by sending a link, and it takes about a minute. Making it that seamless took a lot of work. ### Part V: Launch Q: What did you learn from the launch? Nair: That AI agents are the pull. Most of the interest was in sharing accounts with agents, more than with other people. Q: Which feedback from the launch surprised you? Nair: People responded to the simplicity, not the cryptography. We spent years on the security underneath, and what people talked about was that it took a minute to share an account. ### Part VI: The hard part Q: How many people use Multifactor today? Nair: Thousands of people have already shared their accounts with people and AI agents using Multifactor, and usage is growing exponentially. Q: What is a challenging time you had building Multifactor? Nair: Deciding to do one thing, and then making that one thing feel effortless. Cutting a broad security platform down to account sharing was a hard call. Getting sign-up and sharing down to a minute, across real websites, then took far more work than we expected. Q: Who believed in you before there was anything to show? What did they say or do? Nair: My Ph.D. advisor, Dawn Song, and the Hertz Foundation. They backed the research years before it was a product, and both are still behind us: Dawn is on our advisory board, and Hertz supported Multifactor through its entrepreneurial initiative. Q: What did Y Combinator change about how you build? Nair: Focus, speed, and a habit of talking to users. We ship faster, we do fewer things, and we build what people ask for. ### Part VII: What it is all for Q: What would make you say Multifactor has succeeded? Nair: When Multifactor is the gold standard for securely sharing online accounts, so that sending a link is simply how it is done. Q: What does online security look like when AI agents log in on our behalf? Nair: Passwords stop being something you hand out. Every agent gets its own scoped, revocable, logged access, and the question shifts from "who knows the password" to "what is this agent allowed to do, and what did it do". ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Nair: We don't publish user or revenue figures. The goal for the next year is to keep usage growing exponentially and to make Multifactor the obvious way to give an AI agent access to an account. Q: And for the next 24 months? Nair: To be the default way people share accounts with each other as well as with agents. Q: Where do you want Multifactor to be in five years? Nair: The gold standard for securely sharing online accounts, used the way people use a share link for a file today. Q: What are your hopes and fears for a post AGI world? Nair: My hope is that agents take on the work people don't want to do. My fear is agents acting inside our accounts with no limits and no record. That fear is a lot smaller with security guardrails for agents in place, and those guardrails are what we are building. About Multifactor: Multifactor lets you share any online account with people and AI agents by sending a link, without sharing the password. Passwords are encrypted on your device, recipients act only within the permissions you choose, every action is logged, and access can be revoked instantly. Multifactor is a public benefit corporation based in San Francisco. multifactor.com ## LoupeKit (Slovakia) URL: https://startup.talk/slovakia/loupekit/ Founders: Ján Turský Ján Turský is the middle of three brothers in a family that works in IT. He fell for Android development because he could see what he had programmed working almost at once. He studied at Matej Bel University in Banská Bystrica, started out as an Android developer at a small company in the same city, and has spent more than 14 years building mobile and web apps. Today he runs Apptiary, his own studio in Bratislava, Slovakia. LoupeKit grew out of his search for tools to improve his own websites. It is a browser extension for Chrome, Edge and Firefox that names a page’s tech stack with evidence, pulls its palette, fonts and assets, exports design tokens, and gives a 0 to 100 Vibe audit score for how much of a page reads as raw AI output. It has more than 200 users and has reached version 1.0.0. We spoke about scoring something as subjective as an AI-made page, why the extension asks for no website permissions, and the last 10 percent of a release. Key takeaways: - The Vibe audit scores a page from 0 to 100 across six weighted categories using 100 heuristics. Each finding states the measurement behind it and offers a fix, and hand-written pages are calibrated to score low. - Every stack detection shows its evidence: a DOM marker, an asset URL, a header, a page global or a cookie name. - LoupeKit asks for no permission on any specific website. Its code runs only in the tab you click, only when you ask. - Every tool is free for 24 hours with no account. Pro is €6.99 a month or €49.99 for life. The Principal / Founder: - Ján Turský (Founder) Based: Bratislava, Slovakia Studied: Matej Bel University, Banská Bystrica Before: Android developer; 14+ years in mobile and web apps Also makes: Podcast To Transcript, TextArray, AddonHive, Engraving Studio Studio: Apptiary s.r.o. Website: loupekit.com LinkedIn: Ján Turský (https://www.linkedin.com/in/jantursky/) ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Turský: I was the middle child, with one older brother and one younger. The whole family works in IT. Q: What first got you into building things? Turský: Android. I liked development itself, and I liked how quickly you could see something new you had programmed actually working. I enjoyed it a lot and I still do. Q: What was your first job, and what did it teach you? Turský: I was an Android developer at a small company in Banská Bystrica, Slovakia. ### Part II: The road to LoupeKit Q: How did you come to build LoupeKit? Turský: I like the idea of bringing a large number of tools together in one place instead of having them scattered across many apps. Q: What was the moment you wanted to read any page the way you would read your own? Turský: I was looking for good tools to help me improve my own websites. Either I could not find the functionality, it was paid, or the content did not suit me. Q: What were the first pages you ran a Vibe audit on? Turský: My own websites, jantursky.com and apptiary.com. Q: What does building under Apptiary mean for how you pick projects? Turský: I wanted to start a business, and this was the option that appealed to me most. Q: What is it like to build browser extensions for three browsers? Turský: Chrome, Edge and Firefox all ship as real MV3 builds, not a Chrome extension with a Firefox promise. The engines disagree on permissions in small ways. Firefox's screenshot call, for example, accepts a different grant than Chrome's, so each build gets checked on its own. ### Part III: The product Q: Why the name LoupeKit? Turský: A loupe is the small magnifier you hold up to look closely at one thing. LoupeKit is the set of tools for doing that to a web page. Q: Why a Vibe audit for how much of a page is raw AI output? Turský: A stack detector tells you what a page is. I wanted a tool that also says how it was made. The audit scores a page from 0 to 100 across six weighted categories: HTML artifacts, CSS and utility-class verbosity, JavaScript runtime artifacts, copy patterns, accessibility and page weight. It is the one result people want to screenshot and compare. Q: Why name the tech stack with evidence? Turský: A bare label is easy to doubt. Every detection shows how it matched: a DOM marker, an asset URL, a header, a page global or a cookie name. You can check the claim yourself instead of trusting the tool. Q: Why export design tokens as a Tailwind theme? Turský: A palette you cannot paste into a project is only half useful. LoupeKit exports the colours to a tailwind.config.js or to CSS variables, and it can also check a page's colours and fonts against a Tailwind 4 @theme block you paste in. Q: Why read only the tab you click? Turský: Because the extension asks for no permission on any specific website. The inspection code goes into the tab you are on, at the moment you ask, through activeTab. Nothing runs in the background and nothing is collected across tabs. Audits run only when you press the button. Q: Why a 24 hour trial with all tools? Turský: Every tool is unlocked for the first 24 hours, with no account. You judge it on your own pages, and when the trial ends the panel locks until you buy Pro or enter a licence key. Q: Why build tools like a cookie editor with so few permissions? Turský: Security comes first, and so does asking for as few invasive permissions as possible. Q: What do you deliberately leave out of LoupeKit? Turský: Background collection across tabs, permissions for specific websites, and accounts for the trial. Live edits to a page stay in the page, can be undone, and are never saved or uploaded. AI provider keys stay in your browser and never reach my server. ### Part IV: Positioning Q: Who is LoupeKit for, and what job does it do for them? Describe the person who gets the most value. Turský: Developers and designers who need to know what a page in front of them is made of. Typical jobs: naming the stack behind a site, pulling its palette, fonts and SVGs, checking contrast, and finding out whether a page, their own included, reads as raw AI output. The most value goes to people who do this several times a week. Q: How do you view your positioning relative to Wappalyzer, WhatRuns and Hoverify? Is there another competitor you think of first? Turský: Stack detectors tell you what a page runs on. LoupeKit also scores how it was made, shows the evidence for each match and asks for no website permission at install. I think of it as an inspector that sits next to a detector rather than replacing one. Q: What is your core moat when stack detectors are everywhere? Turský: Three things together. The Vibe audit, a score that no stack detector produces. Empty host permissions in both built manifests, which anyone can verify in the manifest instead of taking a bullet point on faith. And evidence for every detection. Q: What are some of the use cases for LoupeKit that are not obvious and can add a lot of value to users? List as many as possible. Turský: Checking a client's or a competitor's site before a pitch. Auditing your own AI-built site before launch. Pulling a full palette, the font matrix and every SVG from a page you admire. Checking colours and fonts against your design tokens. Contrast checks measured against the real background behind the text. Previewing a page at other screen sizes. It is also a very good tool for checking accessibility and search engine optimization, and it has many other tools that can be really helpful. Q: What is the one detail in LoupeKit you are proudest of? Turský: LoupeKit is not only a separate extension. It also has shortcuts and a bar at the bottom of the page for quick, dynamic use, and the app scales well as it grows. ### Part V: Launch Q: What did you learn from the launch? Turský: That you have to prepare for the release: good graphics, good copy, and promotion across several channels. I am still learning the last part. Q: Which feedback from the launch surprised you? Turský: Most of the feedback came privately by email, from several people. They were good ideas, and I worked some of them in and adapted my work to them. ### Part VI: The hard part Q: How many people use LoupeKit today? Turský: Over 200 users at the moment. Q: What is a challenging time you had building LoupeKit? Turský: A stretch where I was not sure the app would get a good response or appeal to users. I still think it is a useful app that can help people and make their work easier. Q: Who believed in you before there was anything to show? What did they say or do? Turský: My wife believed in me, and she still does. Q: How do you score something as subjective as how AI made a page looks? Turský: I do not pretend it is objective. The audit runs 100 heuristics, and each finding states the measurement that triggered it and offers a concrete refactor. The weights differ by category, and the whole thing is calibrated so that hand-written pages score low. A tool that flags careful work is worse than no tool. ### Part VII: What it is all for Q: What would make you say LoupeKit has succeeded? Turský: Getting the release done. The last 10 percent always takes the longest, so I am glad to have reached version 1.0.0, the official production release. I also learned a lot along the way about programming, languages and technologies across several areas. Q: What does it feel like to see other people use LoupeKit? Turský: I like the idea of my software being used by someone else for a good purpose, and of that person enjoying using it. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Turský: Keep adding features and services to the product, listen to the community and support it, and build the requests that are reasonable. Q: And for the next 24 months? Turský: I am already working on my next product, podcasttotranscript.com. Q: Where do you want LoupeKit to be in five years? Turský: Five years is far off, but I would like to stay in the startup space and keep working on my own projects. LLMs and AI are popular right now, so why not learn more in that area and push further. Q: What are your hopes and fears for a post AGI world? Turský: I worry that the demand growing right now, along with all the low-quality content that drains the world's resources, will hit a point where the bubble bursts. I am also sad about all the pointless content being created, in images, videos and so on, that adds no value. About LoupeKit: LoupeKit is a browser extension for Chrome, Edge and Firefox that inspects any web page: it names the tech stack with evidence, extracts the palette, fonts and assets, exports design tokens to Tailwind or CSS variables, checks contrast and accessibility, and gives a 0 to 100 Vibe audit score for how much of a page reads as raw AI output. Every tool is free for 24 hours; Pro is €6.99 a month or €49.99 lifetime. LoupeKit is made by Apptiary in Bratislava, Slovakia. loupekit.com ## Appulse Index (India) URL: https://startup.talk/india/appulseindex/ Founders: Pringal Gigaiya Pringal Gigaiya grew up in Rajkot, Gujarat, and the first thing he built was a WordPress plugin. He studied Information Technology at V.V.P. Engineering College, started his career at INTUZ Solutions in Ahmedabad in 2014, and from 2016 worked remotely for companies in the US and the UK, completing an MTech in Computer Engineering at RK University along the way. After twelve years of building software for clients in India, the UK, Nigeria, Italy and the US, he decided to build something of his own. Appulse Index is a free, live index of iPhone apps moving through Apple’s App Store top 100 charts: climbers, fallers, new releases and new number ones, across 11 countries, 25 categories and 858 charts. He built the first version alone in about a month and launched it on 29 October 2026. We spoke about why he refuses to estimate downloads, the crawler that made 31% of his traffic, and how a product started by one developer could grow into a global business. Key takeaways: - Appulse Index tracks 858 App Store charts across 11 countries and 25 categories, about 30,600 apps, and collects new data every six hours. - Its 0 to 100 momentum score uses only chart positions. There are no download or revenue estimates, by design. - Pringal built and launched the first version alone in about a month. It is free and self-funded, with a freemium tier planned. - His targets: about 10,000 monthly active users and $1,000 in monthly recurring revenue within 12 months, then 50,000 users and $5,000 within 24 months. The Principal / Founder: - Pringal Gigaiya (Founder) Based: Ahmedabad, India Grew up: Rajkot, Gujarat Studied: IT, V.V.P. Engineering College; MTech, RK University Before: 12 years as a full-stack developer and solutions architect Funding: Self-funded Website: appulseindex.com LinkedIn: Pringal Gigaiya (https://www.linkedin.com/in/pringal-gigaiya/) ### Part I: Before any of this Q: Where did you grow up, and what was the first thing you built on a computer? Gigaiya: I grew up in Rajkot, Gujarat, and I'm currently based in Ahmedabad. My journey into technology started with a simple curiosity: understanding how websites work and how I could build something of my own. The first thing I built was a WordPress plugin. It wasn't a massive project, but it was a turning point for me. Seeing something I had written actually work on a live website gave me a sense of excitement and possibility. I realized that with code, an idea could become something real that people could use. That curiosity eventually turned into a career spanning 12+ years in software development, working across web applications, SaaS platforms, and complex backend systems. Today, as the founder of Appulse Index, I'm still driven by that same curiosity. The scale of what I build has changed, but the motivation hasn't: identify a problem, build a solution, and create something genuinely useful. Looking back, that small WordPress plugin was more than my first project. It was the beginning of my journey from developer to founder. Q: Who or what first got you interested in software? Gigaiya: I've always been naturally drawn to technology. Even before I started coding, I was fascinated by how software worked and how something built on a computer could solve real-world problems. What attracted me most was the endless possibility of creating something from scratch. Unlike many other fields, software gives you the freedom to turn an idea into a working product with just a computer, curiosity, and the willingness to learn. As I started exploring web development, that interest quickly became a passion. Every new technology I discovered made me want to learn more, experiment, and build something better. Even after 12+ years in development, that excitement hasn't changed. Technology keeps evolving, and that's what keeps me motivated. Today, with AI transforming how we build software, I feel there's never been a more exciting time to be a developer and founder. ### Part II: The road to going independent Q: You studied Information Technology at V.V.P. Engineering College from 2010 to 2014. What did you take from those years? Gigaiya: My years at V.V.P. Engineering College from 2010 to 2014 played an important role in shaping my technical foundation and the way I approach problem-solving today. Beyond learning programming languages, databases, and software engineering concepts, college taught me how to think logically, break complex problems into smaller parts, and find practical solutions. One of the biggest lessons I took from those years was that technology is a field where learning never really stops. What you learn in the classroom gives you a foundation, but real growth comes from experimenting, building projects, making mistakes, and continuously improving. Those four years also helped me develop an independent mindset. I realized that I didn't have to wait for someone to teach me everything. With curiosity and the right resources, I could explore new technologies and build things on my own. Looking back, college gave me more than an engineering degree. It gave me the confidence to enter the software industry in 2014 and a learning mindset that continues to guide me today, both as a developer and as the founder of Appulse Index. Q: Your first job was at INTUZ Solutions in Ahmedabad in 2014. What did it teach you? Gigaiya: Joining INTUZ Solutions in Ahmedabad in 2014 was my first real step into the professional software industry. It was where I started understanding the difference between writing code and building software that people actually depend on. Working on real-world projects taught me the importance of writing clean, maintainable code, understanding client requirements, meeting deadlines, and taking responsibility for what I delivered. I also learned that being a good developer isn't just about technical skills. Communication, teamwork, and the ability to solve unexpected problems are equally important. One of the most valuable lessons was learning to adapt. Every project came with different requirements and challenges, which pushed me to explore new technologies and think beyond a single solution. That experience helped shape my approach to software development. Even today, after more than 12 years in the industry, I carry those early lessons with me. INTUZ gave me something more valuable than just my first professional experience. It helped me develop the mindset of taking ownership, continuously learning, and focusing on building solutions that deliver real value. That same mindset eventually became an important part of my journey as a founder. Q: From 2016 you worked remotely for companies in the US and the UK, including Carolina Cabinet Warehouse and The Safety Supply Company. How did that start? Gigaiya: After gaining experience in my first job, I wanted to explore opportunities beyond my local market and work on products used by people internationally. Around 2016, I started working remotely with companies in the US and UK. It was an important transition because remote work required a different level of ownership, communication, and independence. Working with businesses like Carolina Cabinet Warehouse and The Safety Supply Company gave me exposure to different industries, business processes, and customer expectations. I wasn't just implementing features anymore. I was learning how software directly affects business operations, from e-commerce and inventory management to analytics and workflow automation. One of the biggest lessons was that location doesn't limit the value you can create. What matters is understanding the problem, communicating clearly, and delivering reliable solutions. Those experiences gave me the confidence to work independently with international teams and eventually build a product intended for a global audience. Q: You went back for an MTech in Computer Engineering at RK University in 2020 while working. Why? Gigaiya: By 2020, I had already spent several years working professionally in software development. I had gained practical experience building applications, solving technical challenges, and working with international clients. But I wanted to strengthen my understanding of computer engineering beyond day-to-day development. Pursuing an MTech while continuing to work was my way of combining academic learning with real-world experience. It gave me an opportunity to revisit technical concepts with a completely different perspective because I could connect what I was studying with problems I had already encountered professionally. Balancing work and higher education wasn't always easy, but it reinforced the importance of discipline, time management, and continuous learning. For me, education has never been just about earning a degree. It's about expanding the way I think and becoming better at solving increasingly complex problems. Q: Of all the projects you have delivered, which one taught you the most? Gigaiya: One of the most valuable learning experiences came from working on a healthcare technology platform at EVERSANA. Healthcare systems involve complex workflows, multiple external integrations, and a high expectation of reliability. One particular challenge involved improving an insurance verification integration that initially depended on a single provider and configuration. I worked on restructuring the integration to support multiple providers, including migrating from pVerify to Availity, while making the architecture more flexible for different API responses and configurations. What made the project interesting wasn't just the technical implementation. It was understanding how to design software that could adapt as business requirements changed without constantly rebuilding the underlying system. That experience reinforced an important lesson: good engineering isn't simply about making something work today. It's about building systems that remain maintainable, extensible, and reliable tomorrow. I carry that philosophy into every project I work on, including Appulse Index. ### Part III: From client work to Appulse Index Q: After twelve years of building for clients, why build a product of your own? Gigaiya: For more than twelve years, I've helped businesses turn their ideas into working products. I've worked across different industries, built applications from scratch, improved existing systems, and handled everything from database architecture to deployment. Over time, I started thinking about what it would be like to apply that experience to something of my own. Client work taught me how to build products, but I wanted to experience the complete journey of being a founder: identifying a problem, making product decisions, launching, understanding users, and continuously improving based on feedback. Appulse Index became an opportunity to do exactly that. I also wanted to build something that could create value beyond the hours I spent developing it. A product that could serve users globally and continue growing over time. For me, becoming a founder wasn't about leaving development behind. It was about taking everything I'd learned as a developer and applying it to a product where I was responsible for both the technology and the direction. Q: Why App Store charts? What was the moment you decided this was the thing to build? Gigaiya: I've always been interested in how digital products grow, especially mobile applications. Thousands of apps compete for attention, but occasionally a relatively unknown app suddenly climbs the rankings. I found that interesting because the App Store charts reveal changes in user attention that aren't always obvious from mainstream technology news. Most people see the apps already at the top. I became more interested in the ones moving toward the top, especially smaller apps gaining momentum in specific countries or categories. That led to a simple question: what if there were a place where anyone could discover these movements without manually checking hundreds of charts? As I explored the available App Store data, I realized there was an opportunity to organize public ranking information into something useful for developers, founders, marketers, and people researching app trends. That curiosity became Appulse Index. The idea wasn't to predict the next billion-dollar app. It was to make emerging app movements easier to discover, understand, and follow. Q: The name comes from "appulse", the astronomy term for two objects appearing to come close in the sky. How did you land on it? Gigaiya: I wanted a name that connected naturally with apps but also represented movement, discovery, and changing positions. The word "appulse" caught my attention because of its astronomical meaning, when two celestial objects appear close together in the sky. I liked the connection with what the product does. Apps are constantly moving through rankings, approaching competitors, overtaking them, or appearing alongside new names in the charts. The word also begins with "app", which made it feel particularly appropriate for an App Store intelligence platform. Adding "Index" made the purpose clearer: a place where these movements could be organized, tracked, and explored. Appulse Index felt distinctive, memorable, and broad enough to grow beyond its initial feature set. Q: Your roles at EVERSANA and Provab ended in June 2026. Is that when you started building Appulse Index, and how long did the first version take? Gigaiya: I started developing Appulse Index in September 2026, a few months after my roles at EVERSANA and Provab ended. The first version took approximately one month to build, and I launched it publicly in October 2026. After spending more than twelve years developing software for different businesses, I wanted to challenge myself to build and launch a product entirely on my own. I handled everything from product planning and UI design to backend development, database architecture, data collection, and deployment. My initial focus was simple: collect App Store chart data, track ranking movements, and make it easier for people to discover emerging applications. Rather than spending months building every possible feature, I wanted to launch quickly, learn from the experience, and improve the product continuously. That approach helped me move from an idea to a publicly available product in just one month. Since launching, I've continued adding features, improving performance, and expanding what Appulse Index can offer. Q: You design, build and run Appulse Index yourself. What does a normal day on it look like next to client work? Gigaiya: My days usually involve switching between software engineering, product development, and understanding what the data is telling me. I typically start by checking whether the data collection processes are running correctly, reviewing application performance, and looking for unusual movements in the App Store charts. Then I move into development work. That could mean improving an existing feature, optimizing database queries, fixing an issue, or experimenting with a new idea. I also spend time exploring the charts themselves. Sometimes an interesting ranking movement becomes an App Insights story or gives me an idea for a feature. Alongside that, I manage the operational side of the product, including hosting, infrastructure, performance, and costs. Balancing Appulse Index with client work requires prioritization. I can't build every idea immediately, so I focus on improvements that make the product more useful or more reliable. Being a solo founder means wearing many hats, but I enjoy that. It keeps me connected to every part of the product, from the underlying code to the experience users see. ### Part IV: The product Q: Who is Appulse Index for, and what job does it do for them? Describe the user who gets the most value. Gigaiya: Appulse Index is built for anyone who wants to understand what's happening in the mobile app market, particularly independent developers, startup founders, app marketers, product researchers, and people exploring new opportunities. The user who gets the most value is someone who wants to discover emerging apps before they become widely recognized. For example, an indie developer might use Appulse Index to identify rising apps in a particular category, study competing products, or explore opportunities in different countries. Instead of manually checking App Store rankings every day, users can see which apps are climbing, which are falling, and which new releases are gaining visibility. My goal is to make App Store discovery accessible to everyone, not just companies that can afford expensive market intelligence platforms. Q: How do you view your positioning relative to Sensor Tower and AppFigures? Is there another competitor you think of first? Gigaiya: I see Sensor Tower and AppFigures as established platforms serving important parts of the mobile app intelligence market. They offer extensive analytics, competitive research, and commercial intelligence. Appulse Index approaches the market from a different starting point. Rather than trying to provide every possible metric, I'm focused on making public App Store chart movements easy to discover and understand. The emphasis is on momentum, emerging applications, new releases, category movements, and geographical trends. AppFigures is probably the closer comparison because of its connection to app rankings and developer-focused analytics. However, I also think about simpler discovery tools and the App Store charts themselves as alternatives users might rely on. My goal isn't to replicate every feature of the larger platforms. It's to create a focused, accessible product that helps users quickly answer a specific question: which apps are gaining attention right now? Q: What is your core moat that prevents Sensor Tower, AppFigures or anyone else from replicating your feature set? Gigaiya: I don't believe any individual feature is impossible to replicate, especially in today's world where AI has made software development faster and more accessible. The real challenge is building a product that consistently delivers value and keeps improving. For Appulse Index, I'm focused on three things: reliable historical chart data, a discovery experience built around ranking movements, and making that information accessible without unnecessary complexity. Historical data becomes more useful as it accumulates. A ranking today tells you where an app stands, but its movement over weeks and months provides much more context. I also believe being a solo founder gives me an advantage in speed. I can identify a problem, experiment with a solution, and release improvements without multiple layers of decision-making. My long-term advantage has to come from execution, data quality, accumulated history, and understanding users better over time, rather than assuming competitors cannot build similar features. Q: Your momentum score runs from 0 to 100 and uses only chart positions. How did you arrive at that formula? Gigaiya: The idea behind the momentum score was to make App Store ranking movements easier to understand and compare. When you're tracking hundreds of charts across multiple countries and categories, looking at individual ranking changes can become overwhelming. I wanted to provide a simpler way for users to identify applications showing interesting movement. The score uses chart position data rather than estimated downloads or revenue. My intention was to keep the indicator grounded in publicly observable information instead of introducing speculative metrics. I chose a 0 to 100 scale because it's familiar and easy to interpret, even for users without a technical background. The score is designed as a discovery signal rather than a definitive measurement of an application's success. A high momentum score highlights an app worth investigating, but it doesn't necessarily mean the app is generating significant downloads or revenue. I see the scoring methodology as something that can continue improving as the platform accumulates more historical data and I learn how users interpret the results. Q: You refuse to show download or revenue estimates or to guess why an app moved. Why draw that line? Gigaiya: Because I believe there's an important difference between what the data shows and what we assume it means. Public App Store charts provide ranking information, but they don't reveal exact downloads, revenue, marketing spending, or the reasons behind every movement. It's possible to create estimates, but those estimates depend on assumptions that may not be accurate, particularly for smaller apps or individual countries. I would rather show users reliable information and clearly explain its limitations than present speculative numbers that look precise. For example, if an app climbs 50 positions, I can show that movement. But I shouldn't automatically claim it happened because of a marketing campaign, viral video, or sudden increase in downloads without supporting evidence. Trust is important for a data product. My approach is to separate verified observations from interpretation and give users enough context to form their own conclusions. As the product evolves, I'm open to exploring additional metrics, but transparency about methodology and uncertainty will remain essential. Q: The indie list scores developers from public clues. How do you check that the label is right? Gigaiya: Identifying independent developers from public information isn't always straightforward. An app might look like an indie project but belong to a larger company, while another app might be built by a small team operating under a registered business name. That's why I think it's important to treat publicly available signals as indicators rather than definitive proof. The goal of the indie discovery feature is to help users find potentially independent developers, not to make absolute claims about company ownership or team size. Where possible, the information should be checked against developer websites, publisher details, and other publicly available sources. Ambiguous cases need to be treated cautiously. I also see developer feedback and corrections as an important part of improving accuracy. For me, the priority is to make the feature useful while being transparent about the limitations of what public data can establish. Q: Foldcheck lists more than 2,000 apps ready for the iPhone Duo. Why did you build it? Gigaiya: FoldCheck came from thinking about how new hardware creates new opportunities for app developers. Whenever a new device format or screen experience emerges, developers need to understand which applications are adapting and how those applications are using the available space. I wanted to create a directory where people could discover apps designed for the iPhone Duo experience, including support for outer screens, larger inner displays, and split-view layouts. For users, it provides a way to find applications that make better use of the device. For developers, it creates an opportunity to showcase their work and discover how other applications are approaching the new form factor. It also fits the broader philosophy behind Appulse Index: helping people discover applications and opportunities that might otherwise be overlooked. What interests me most is that a change in hardware can create an entirely new category of software opportunities, even for very small development teams. Q: What are some of the use cases for Appulse Index that are not obvious and can add a lot of value to users? List as many as possible. Gigaiya: Most people initially see Appulse Index as a place to check app rankings, but I believe its value goes much further. Some of the use cases include: 1. Discovering emerging app ideas before they become mainstream. 2. Identifying smaller apps competing successfully against established brands. 3. Finding underserved categories with growing activity. 4. Studying how new applications perform after launch. 5. Tracking competitors across different countries. 6. Discovering apps that are successful in one market but less visible in another. 7. Identifying potential localization opportunities. 8. Comparing ranking movements across app categories. 9. Researching potential competitors before building a new product. 10. Finding independent developers and interesting niche applications. 11. Discovering potential acquisition or partnership candidates for further research. 12. Understanding seasonal patterns in app rankings. 13. Monitoring the impact of publicly announced product launches or updates. 14. Finding unusual ranking movements worth investigating. 15. Generating research topics for journalists, bloggers, and content creators. 16. Exploring opportunities created by new device formats through FoldCheck. 17. Building watchlists of applications worth following. 18. Identifying apps that maintain strong rankings over longer periods. 19. Discovering regional consumer preferences. 20. Learning which types of applications are attracting attention in different markets. One of my favorite use cases is simply exploring the charts without a specific app in mind. Sometimes the most interesting opportunities are the ones you weren't originally looking for. I want Appulse Index to become a place where developers and founders come not only to track competitors, but also to discover what they might build next. ### Part V: Going live Q: When did Appulse Index go live, and how did the first users find it? Gigaiya: I launched Appulse Index as an independent project, initially focusing on making the core chart tracking and discovery experience available publicly. My approach was to launch with a useful foundation and keep improving it rather than waiting until every planned feature was complete. The early discovery strategy focused on making individual app pages, category rankings, and insights accessible through search engines. I also started sharing interesting app movements and product updates through platforms such as LinkedIn and Product Hunt. Organic discovery is particularly important to me because every app has its own story, and people searching for those apps can potentially discover Appulse Index through the ranking information. The initial audience is still small, but the experience has been valuable. Building the product is one challenge; helping the right people discover and return to it is another. I'm treating this early stage as an opportunity to learn which features create the most value and where to focus next. Q: What is the most surprising thing the charts have shown you so far? Gigaiya: One of the most surprising things has been how quickly a very simple app can become highly visible. For example, I recently came across DuoFridge, a $1.99 app built around a fridge-related use case that reached the top of an App Store chart. It was a reminder that successful app ideas don't always have to be complicated. Sometimes a simple product that solves a specific problem at exactly the right moment can attract significant attention. I've also seen regional applications perform exceptionally well in individual countries while remaining relatively unknown elsewhere. These discoveries have changed the way I think about app opportunities. As developers, we sometimes focus heavily on complex technology and feature-rich products. But the charts regularly remind me that timing, relevance, simplicity, and understanding a particular audience can be just as important. That's one of the reasons I enjoy working on Appulse Index. The data constantly challenges my assumptions about what makes an app interesting. Q: You publish a weekly report every Monday and App Insights stories. Who writes them, and how do you pick the apps? Gigaiya: I manage the content direction myself, using Appulse Index's chart data as the starting point. The weekly reports are designed to summarize notable movements and help readers understand what's happening across the tracked App Store charts. For App Insights, I usually look for applications with an interesting story behind their rankings. That might be a newly released app climbing quickly, an independent developer reaching a top position, or a regional app gaining unexpected visibility. I don't want the stories to be just another list of rankings. I try to connect the data with useful observations for developers, founders, and people interested in mobile products. AI tools can help me organize research and prepare drafts, but the topics, editorial direction, and final review are my responsibility. The aim is to make the data more approachable and turn interesting chart movements into stories that people can learn from. ### Part VI: The hard part Q: What has been the hardest technical problem in tracking 858 charts every day? Gigaiya: The biggest challenge hasn't been collecting the data once. It's collecting it consistently, processing it efficiently, and making it useful over time. Tracking hundreds of charts across countries, categories, and ranking types creates a significant amount of data. Each collection cycle needs to identify applications, update rankings, preserve historical information, and calculate movements without introducing duplicates or inconsistencies. I built the platform using Next.js, PostgreSQL, and scheduled background processes. The collection system runs every six hours, so database efficiency and reliable processing are particularly important. Another challenge is presenting historical data quickly. As the number of applications and ranking records grows, even relatively simple queries can become expensive if the database isn't structured properly. Working on this has reinforced the importance of designing for scale early while avoiding unnecessary complexity. The challenge is balancing three things: data accuracy, performance, and infrastructure cost. As a solo founder funding the product myself, I have to pay attention to all three. Q: You posted that one crawler made 31% of the requests on a Next.js site with 30,000 pages and pushed up your Vercel bill. Was that Appulse Index, and what did you change? Gigaiya: Yes, that experience came from Appulse Index, and it taught me an important lesson about running a data-heavy website. The platform has thousands of publicly accessible pages, including individual app profiles and category rankings. That makes it useful for search engines, but it also attracts a significant amount of automated traffic. At one point, I noticed that hosting usage was increasing even though actual visitor numbers weren't growing at the same rate. After investigating the request data, I found that a single crawler was responsible for approximately 31% of requests. The surprising part was how directly that traffic affected infrastructure costs. On usage-based hosting platforms, automated requests can consume resources just like genuine visitors. I started reviewing caching behavior, unnecessary server work, crawler traffic, and the infrastructure costs associated with each request. One important change was reducing excessive incremental static regeneration activity, which had contributed to unnecessary writes. The experience also pushed me to evaluate infrastructure where costs would be more predictable. My biggest takeaway was that building a scalable product isn't only about handling more users. It's also about controlling the resources consumed by traffic that may never become users. Q: Everything is free, with no ads. How do you pay for it, and do you plan a paid tier? Gigaiya: Right now, Appulse Index is self-funded. I cover the infrastructure and operating costs myself. I wanted to make the platform accessible from the beginning because I believe useful app market information shouldn't be limited to businesses with large research budgets. At this stage, my priority is building a reliable product, attracting the right audience, and understanding which features users find genuinely valuable. I do plan to introduce a freemium model eventually. The idea is to keep essential app discovery and chart information available for free while offering more advanced features for users who need deeper analysis, expanded historical data, alerts, exports, or professional research capabilities. I don't want monetization to become a barrier to discovering useful information. My goal is to build a sustainable business where free users continue receiving value and paying users support the development of more advanced capabilities. For now, I'm investing in the product with a long-term perspective rather than trying to monetize every feature immediately. Q: What is a challenging time you had building this? Gigaiya: One of the most challenging moments was realizing that the technical cost of running Appulse Index could grow much faster than its actual audience. As a developer, I'm comfortable solving application problems, optimizing queries, and building new features. But running my own product forced me to think differently about infrastructure expenses and long-term sustainability. I encountered a situation where automated crawler traffic was generating millions of requests and increasing hosting costs, even though the platform was still in its early growth stage. It was frustrating because the resources were being consumed without a corresponding increase in genuine users. I had to investigate the traffic, understand how different hosting features were being billed, optimize the application, and reconsider the infrastructure strategy. The experience reminded me that being a founder requires more than technical ability. You have to understand the business impact of every technical decision. It also reinforced something I've believed throughout my career: when a problem appears, the best response is to understand its root cause and build a better solution. ### Part VII: What it is all for Q: Who believed in this before there was anything to show? What did they say or do? Gigaiya: In the beginning, it was really just me. Appulse Index started as a personal idea, without investors, a co-founder, or a team behind it. After spending more than twelve years building software for other businesses, I believed I had the experience and technical ability to create something of my own. I didn't know whether the product would attract users or eventually become a successful business, but I knew I wanted to give the idea a real chance. That belief was enough to get started. I invested my own time, skills, and money into developing the first version. Within a month, I had turned the idea into a working product and launched it publicly. Building something independently teaches you a lot about responsibility and persistence. There isn't always someone else to make decisions or solve problems. You have to trust your judgment, learn quickly, and keep moving forward. For me, Appulse Index represents that willingness to take an idea seriously and act on it, even when there's no guarantee of success. Q: What would make you say Appulse Index has succeeded? Gigaiya: For me, success would mean Appulse Index becomes a genuinely useful resource that developers, founders, and product researchers return to regularly. I would love to reach a point where someone thinking about building a mobile app naturally checks Appulse Index to understand the market, explore opportunities, and study competing applications. Of course, building a financially sustainable business is important. I want the product to generate enough revenue to support its infrastructure, continued development, and eventually a team. But revenue alone wouldn't define success. If an independent developer discovers an opportunity through Appulse Index, builds something useful, and creates a successful business, that would be incredibly meaningful to me. My long-term goal is to build something that helps other people create better products. If Appulse Index becomes a trusted part of that journey, I'll consider it a success. ### Part VIII: The road ahead Q: Alerts are marked "coming soon". What else is next: more countries, Google Play, an API? Gigaiya: There are several directions I want to explore, but I'm trying to approach development in stages. One of the immediate priorities is alerts. I want users to be able to follow applications and know when something meaningful happens, such as a major ranking movement or a new chart milestone. I'm also interested in expanding country coverage, improving historical analysis, and making it easier to compare applications across markets. Google Play is a natural direction to explore because it would make Appulse Index useful across both major mobile ecosystems. However, that expansion depends on finding a reliable and sustainable approach to collecting and maintaining the data. An API is another possibility, particularly for developers, researchers, and businesses that want to integrate ranking information into their own tools. I'm also continuing to improve opportunity discovery, app insights, and the overall research experience. My priority is to build features that solve real user problems rather than expand the product simply to increase the feature count. Q: What are your milestones for the next 12 months, in users and revenue? Gigaiya: Over the next 12 months, my goal is to establish Appulse Index as a useful and trusted platform for mobile app discovery and market research. I'm aiming to reach approximately 10,000 monthly active users, with a strong focus on attracting developers, founders, and researchers who return regularly to explore chart movements and discover new opportunities. On the revenue side, my target is to introduce a freemium model and work toward $1,000 in monthly recurring revenue. These are ambitious targets for a self-funded product, but I believe they're achievable with consistent development, organic growth, and a clear focus on user needs. Beyond the numbers, I want to understand which features provide the most value, improve user retention, and build a sustainable foundation for long-term growth. My priority isn't just attracting visitors. It's creating a product that people find useful enough to return to regularly. Q: And for the next 24 months? Gigaiya: Within 24 months, I'd like Appulse Index to grow from an independent project into a recognized platform within the mobile app development and research community. My target is to reach approximately 50,000 monthly active users and $5,000 in monthly recurring revenue. By that stage, I hope to have expanded the platform's capabilities with broader market coverage, advanced analytics, personalized alerts, deeper historical insights, and additional tools for developers and businesses. I'd also like Appulse Index to become financially self-sustaining, covering its operating costs and generating enough revenue to support continued development. If growth supports it, I would consider bringing in a small team to accelerate product development and improve the user experience. However, I don't want to grow simply for the sake of numbers. My focus will remain on building something useful, reliable, and accessible. Ultimately, I'd like Appulse Index to demonstrate that a product started by one developer can grow into a sustainable global business. Q: Where do you want Appulse Index to be in five years? Gigaiya: In five years, I'd like Appulse Index to become a globally recognized platform for discovering mobile app trends, understanding market movements, and identifying new opportunities. I envision a product that covers multiple app ecosystems, provides meaningful historical insights, and helps everyone from independent developers to established businesses make better-informed decisions. I also want Appulse Index to remain approachable. Market intelligence can become complicated and expensive, but I believe there will always be room for a platform that makes useful information accessible and easy to understand. From a business perspective, I'd like it to be profitable, sustainable, and supported by a small, highly capable team. More importantly, I want it to have a clear identity. When someone asks, "What's gaining momentum in the app market, and what opportunities are emerging?", I want Appulse Index to be one of the first places they think of. That would be a meaningful achievement for something that started as a solo developer's idea. Q: What are your hopes and fears for a post AGI world? Gigaiya: As someone who has worked in software development for more than twelve years, I find the progress in artificial intelligence both exciting and thought-provoking. My biggest hope is that advanced AI will make technology more accessible and allow people to solve problems that previously required enormous resources, specialized knowledge, or large teams. We're already seeing how AI can accelerate software development. I believe future systems could make it possible for individuals and small teams to create products that once required entire organizations. That could open up extraordinary opportunities for entrepreneurship, education, healthcare, and scientific discovery. At the same time, I have concerns about how quickly society can adapt. Job displacement, concentration of power, misinformation, privacy, and the responsible use of increasingly capable systems are serious challenges. I also think that as software becomes easier to create, the value of simply writing code will change. Understanding real problems, making thoughtful decisions, earning trust, and building useful experiences will become even more important. My hope is that AI becomes a tool that expands human potential rather than replacing human purpose. As a developer and founder, I want to keep learning, adapting, and using these technologies responsibly to build things that make a positive difference. About Appulse Index: Appulse Index is a free, live index of iPhone apps moving through Apple's App Store top 100 charts: climbers, fallers, new releases and new number ones, by country and category. It covers 11 countries, 25 categories and 858 charts, about 30,600 apps, with a 0 to 100 momentum score built only from chart positions. A free account unlocks full lists, longer rank history and a 200-app watchlist. Appulse Index is built by Pringal Gigaiya in Ahmedabad, India. appulseindex.com ## Skillbox (United Kingdom) URL: https://startup.talk/uk/skillbox/ Founders: Adenekan Wonderful, Shedrack Akintayo Skillbox is built by two cofounders, Adenekan Wonderful and Shedrack Akintayo. Adenekan learned to build for the web in a free community class in Lagos, where a teacher told the room that what they were learning had made Google the company it was. He worked out 1% of that and decided that was all he needed. He walked into more than 30 companies asking to intern, and the day after he finished school he started at Switch.ng. Years later, as an engineer in London on a UK Global Talent visa, he found himself copying AI coding skills around and restaging his setups by hand. Together with Shedrack, who brings deep experience in cloud, DevOps, security and documentation, he built Skillbox, one library for those skills that follows their sources, checks every version and syncs the right set to every tool. It launched on 11 November 2026. We spoke about the update that started it, why a security check has to be tied to the exact version, and why good architecture still takes experience. Key takeaways: - Skillbox keeps AI coding skills in one library, follows their GitHub or GitLab sources, shows the diff when something changes, and delivers the right set to Claude Code, Cursor, Codex and each machine. - Every version is checked. Skillbox Trust decides whether a version can sync, and Cisco Skill Scanner adds a second, independent set of findings. A pass is not a guarantee, so people still decide what runs. - Trim shortens noisy test and build output before the agent reads it, and keeps the full log on the machine. - About 50 people installed the desktop app in the first two or three hours after launch. Today it has a few hundred users and about 3,000 synced skills. The Principals / Founders: - Adenekan Wonderful (Cofounder) Based: London, United Kingdom From: Lagos, Nigeria Before: Engineer at Magic, Audiomack and DoorDash; founding engineer at NEX Also built: Piggment, Scoutbar and Pause.do Mentor: Engineering mentor on ADPList - Shedrack Akintayo (Cofounder) Based: Manchester, United Kingdom Background: Cloud, DevOps, security and documentation; docs lead for Kubernetes and cloud native Community: CNCF Ambassador - Find Skillbox Website: skillbox.so Download: Skillbox for desktop (https://skillbox.so/downloads) ### Part I: Before any of this Q: Where did you grow up, and what first got you into programming? Adenekan Wonderful: I grew up in Lagos, Nigeria. It was an amazing place to grow up, though my family was not wealthy. From early on I was looking for something that could set me apart from the other kids, and programming became that. I started around 16 or 17. Being able to sit at a computer and see something I had written actually change what was on the screen blew my mind. I did not know much yet, so I just kept at it, over and over. Q: How did you learn to build for the web, and what did those early years teach you? Adenekan Wonderful: I learned web development at a free community class near where I lived. People in the area ran those classes to give kids another path, and I was lucky enough to get into one. It changed my life. The moment that stayed with me was when my teacher said that what we were learning that day was what had made Google the huge company it was. I sat there and worked out 1% of that, and thought, I just need 1%. That was one of my biggest drivers. If people could make that kind of money from tech, I was going to double down and learn. Q: What was your first job as a developer? Adenekan Wonderful: It was not a formal developer job at first. In Lagos, a lot of people finish school and then spend a year at home waiting for college admission. My brother had done that, and I wanted to do something different. I walked into more than 30 companies around my school, asking to intern, with only some HTML and CSS to show. Around the 26th or 27th company, the CEO liked what I had already made and asked me to start the next day. So the day after I finished school, I started as an intern at Switch.ng. I did not spend that year at home. Q: You are a mentor on ADPList. What do you tell people who ask how to become a senior engineer? Adenekan Wonderful: I tell them that seniority is less about years in the chair and more about how you think. A senior engineer can step back, look at a problem from a wider angle, and explain it clearly. A lot of that comes from having made many mistakes, and from knowing what is worth trying and what is not. I see that pattern over and over. ### Part II: From frontend engineer to maker Q: You moved to London on a UK Global Talent visa. What was that path like? Adenekan Wonderful: It was a unique experience: fairly smooth, but taxing. I had been lucky enough to work on many projects and with many people back home, and that was a lot of what I showed the UK government. The hard part was going back through years of work and remembering the impact I had actually made. That was also one of the first times I really learned to document what I was doing as I went. Q: You have shipped Piggment, Scoutbar, Pause.do and now Skillbox. What do those products have in common? Adenekan Wonderful: They all started as a problem I had myself. I try to solve it, and if the solution works exceptionally well for me, I look at whether it is a common problem for other people. Only then do I try to turn it into a product. I am my first customer. I also have a high bar for the applications I use and pay for, and I hold myself to that same bar on everything I ship. That is a big part of why I like building products: I am building for myself first, and I will not put my name on something I would not want to use. Q: What did you learn from the ones that did not take off? Adenekan Wonderful: The ones that did not take off taught me not to get so obsessed with making something perfect before launch that I never put it in front of people. It is easy to build a polished solution in private and miss the window. You want something refined enough to stand up, but not so over-finished that customers never get the chance to shape it. They are the ones who help you find product-market fit as you go. Q: Who did you build Skillbox with? Adenekan Wonderful: Shedrack Akintayo is my cofounder, and he was one of the most important people in getting Skillbox to where it is. He brings a huge amount of experience in documentation, security, cloud and DevOps. I come more from the product side. Together it has been a combo from heaven. He helped shape the product itself, not just the infrastructure around it. Q: What has it been like building Skillbox with a cofounder? Adenekan Wonderful: It has been a combo from heaven, genuinely. I come from the product side. Shedrack comes from cloud, DevOps, security and documentation, and he has an insane amount of experience in those areas. He was one of the people who actually helped shape Skillbox, not just support it. We have also been using the product ourselves every day, and that shared daily-driver habit has been a useful test: if the two of us cannot live without it, we have something worth passing on. ### Part III: Skillbox Q: Where did the idea for Skillbox come from? Adenekan Wonderful: It came from trying to keep skills in sync myself. I was copying them around and restaging setups, and it was messy. I wanted the skills I trusted on my machine to stay current without me having to chase every change by hand. Q: Why the name Skillbox? Adenekan Wonderful: The idea is in the name. Your skills are keyed in a box, and the box is your machine. I also did not want to spend so long on the name that it became another reason not to launch. So Skillbox it was. Q: Your skills were scattered across repos, project folders and tools. What was the moment you decided to fix it? Adenekan Wonderful: The moment that really frustrated me was when a well-known developer left a comment on Matt Pocock’s grilling skill, complaining about the em dashes. Matt updated the skill, and I had to go and pull the new version myself. I remember thinking, why is this not just coming upstream automatically, especially if I have auto-updates on, from developers I already trust? That was the frustration that pushed me to build Skillbox. Q: Why scan every version of a skill for prompt injection? Adenekan Wonderful: A skill is not just a prompt. It can include instructions, scripts, dependencies and bundled files, and any of those can tell an agent to fetch secrets, run remote code, or ignore your rules. Copying a skill into Cursor or Claude Code is easy. Knowing what this exact version contains, and stopping a bad one from reaching every machine, is not. That is why we check each version, not the skill’s name. The result is bound to the content hash. A later update is a new check. Allowing one flagged version does not approve the next one. An incoming version stays separate from the copy already on your machines until you decide what happens. We look for prompt injection, data exfiltration, secrets, obfuscated code, remote execution, destructive commands, and known vulnerabilities on pinned dependencies. A passing check is not a guarantee that a skill is harmless. You still decide what may run. Q: How do Skillbox Trust and Cisco Skill Scanner work together? Adenekan Wonderful: Skillbox Trust is our required check: static rules and known-vulnerability lookup on pinned dependencies, and it is what actually holds delivery. A version that is waiting, needs review, is blocked, or could not be fully inspected stays in the library. It does not sync until Trust can decide. Cisco Skill Scanner sits beside that as a second, independent set of eyes. It looks for prompt injection, exfiltration, command injection, secrets and obfuscation, and labels findings in STRIDE and OWASP terms. It runs isolated and offline. Cisco findings can still surface issues after Trust has passed, but they cannot grant delivery or override a Trust block. So we built the gate ourselves, and we use Cisco as extra evidence, not as a substitute. A Trust pass is not a sign that Cisco found nothing, and a clean scan is not a guarantee of safety. Q: What does Trim do, and why did you add it to a skill library? Adenekan Wonderful: Agents often run tests, builds, installs, Git and search commands that print thousands of lines they never use. That output still fills the context window, sessions compact sooner, and work gets lost. Trim shortens noisy command output before the agent reads it. The command still runs. The full output stays on your machine, and the agent can read it back if it needs to. Errors always come through. If Trim cannot tell whether an output is safe to shorten, it leaves that output in full. All of that work happens on the device. Prompts, code, transcripts, command output and file paths never leave it. Skillbox only receives numeric totals and setup status, so we can show usage without seeing the work itself. We added it because a skill library that delivers instructions is only half the problem. If those skills then fill the session with noise, you have not really helped. Trim is how we keep the agent’s context useful. It is part of the Individual plan, from £6.99 a month, and getting it into more teams is a big part of what we want to do next. Q: Why a free plan plus a paid tier from £6.99 a month? Adenekan Wonderful: We wanted the core of Skillbox — one library, sync across your tools and machines, source tracking, and Skillbox Trust — to be free, with no card required, for as long as you like. Security is not an add-on on any plan. The free plan covers up to three devices and 100 followed sources. Individual is £6.99 a month after a seven-day trial, and adds Trim, Skillbox AI and Try in chat, with no device or source limits. Team is £12.99 per active member per month, and adds shared workspaces, team standards, workspace usage and a weekly report for admins. The idea is simple: free to sync, and you pay for what saves you money. ### Part IV: The product Q: Who is Skillbox for, and what job does it do for them? Describe the user who gets the most value. Adenekan Wonderful: Skillbox is for people who already use AI coding agents — Claude Code, Cursor, Codex and others — and whose skills have ended up scattered. One came from GitHub. Another lives in a project folder. Three more are installed in Cursor but missing from Claude Code. A teammate has the version everyone else should be using, and eventually nobody is sure what they have, which copy is current, or why one agent behaves differently from another. The job is to give those skills a single library: import them from your computer, GitHub or GitLab, or pick one carefully from Explore; follow the source; review diffs when something changes; scan each version; and deliver a smaller active set to the tools and projects where it belongs. The person who gets the most value is someone who cares which version is actually running, wants the same setup on every machine and tool, and wants a teammate’s good skill to have a proper home rather than being passed around in chat. Teams that want shared standards across a workspace, without forcing an automatic install on every person, get a lot from it too. Q: How do you view your positioning relative to skills.sh, rulesync and Anthropic's own skills? Is there another competitor you think of first? Adenekan Wonderful: skills.sh, rulesync and Anthropic’s own skills are all part of this landscape. Anthropic defined a lot of the skill format that tools now follow. Directories and rule-sync tools help you find or copy skills. We are trying to be the library you actually live in: one place to keep skills, follow their GitHub or GitLab sources, review diffs when something changes, scan each version, and deliver the right set to the right project and machine. The competitor I think of first, though, is Tessl. We looked at them early on. They are building what they call a software factory: an agent-enablement layer for building, testing and distributing skills, with a strong focus on evals — measuring whether a skill actually changes what an agent does. That is important work. For the problem we were trying to solve, the everyday job of keeping a personal or team library in sync, reviewing incoming changes, and deciding what reaches your machines, that was not the UX we needed. So Tessl is still the name that comes to mind first when I think about the space, even though we are pointed at a slightly different job. Q: What is your core moat that prevents Anthropic or anyone else from replicating your feature set? Adenekan Wonderful: I would not claim that a feature list is impossible to copy. Anthropic and others could ship pieces of this. What is hard to reproduce quickly is the combination we had to get right: following a source without silently overwriting the version you already use, binding a security check to a content hash so an approval does not follow the skill’s name, pinning one person to a known version while streaming an update to someone else, and doing that across teams with proper access control. Under the hood we designed it so many clients can share the same version tags, which is how we think about running this at a much larger scale — hundreds of thousands of skills, not just a few thousand. The other half is the feel of the product: the polish, the onboarding, the way updates and reviews work, so people actually trust it as a daily driver. Shedrack’s background in security, cloud and DevOps was a big part of getting that architecture right. After a point you cannot prompt your way through those details. You need experience to make them work. Q: What are some of the use cases for Skillbox that are not obvious and can add a lot of value to users? List as many as possible. Adenekan Wonderful: - Capture a session that went well and turn the approach into a reusable skill, with the names, paths and private details from that conversation left out. It lands in your library and stays private until you publish it, and it is there the next time you work. - Follow a GitHub or GitLab source and review the exact diff before an update reaches your machines, instead of pulling the new version by hand. - Pin one teammate to a known-good version while someone else receives updates automatically. - Give each project its own active set of skills, so a frontend repo does not load the same setup as a release repo, then keep that set consistent across the machines you work on. - Pause unused installs and get a nudge when two skills look like they overlap, without deleting them from the library. - Preview a skill in an isolated chat before you rely on it. - Require a skill as a team standard so a project’s conventions stay in place, while sharing still does not auto-install for everyone. - Keep executables off machines until someone opts in, even if the rest of the skill looks fine. An exception on one version does not cover the next update. - Use Trim so test and build output stops eating the agent’s context, with the full log still on the machine. - For teams outside engineering: we are already hearing from companies that want the same shared-skill idea for sales and marketing, not only for developers. - Browse Explore, inspect a public skill’s instructions and files, and add it deliberately, rather than installing from a README in a hurry. Q: How well can anyone trust a skill from a public repository today? Adenekan Wonderful: Not very well, on its own. A public repository can change, and a skill can include scripts and instructions that are easy to miss. Anyone can publish something that looks useful. Skillbox helps by checking each version, showing you the files and the diffs, and holding delivery when Trust needs a decision. Cisco’s scanner adds another set of findings. None of that is a guarantee. A pass means we did not find a known threat pattern in that version, not that the skill is harmless. You should still read what you are about to run, especially if it came from a source you do not already trust. Trust developers you do trust, review updates from everyone else, and do not let an exception on one version silently cover the next. Q: What is the one detail in Skillbox you are proudest of? Adenekan Wonderful: Two things, and I care about both. The first is the polish: the onboarding, the type, the colour, the way the app feels to use. I care about the micro-animations, the stagger when you open a dropdown, the timing that just feels right. Most people would never measure it, but they feel it. The second is the architecture itself: how we handle security attestation, tagging, syncing skills across teams, and access control. These are small, unglamorous details. After a point you cannot prompt your way to them. You need experience to make them work. ### Part V: Launch Q: Skillbox launched on 11 November. What happened in the first hours? Adenekan Wonderful: The first hours were amazing. We had a huge amount of support, almost immediately. Shortly after we went live we were sitting at number six or seven. People from the Product Hunt community were reaching out to me and to Shedrack to say they loved what we had built. That kind of organic response means a lot. Q: Which feedback from the launch surprised you? Adenekan Wonderful: I assumed that because Skillbox is a desktop app, there would be a real barrier to actually using it. I thought people would visit the site, click around, and leave. Within two or three hours we had about 50 installs, and people were signing up and syncing their skills. I had not expected that. I was so happy. It was the first clear sign that we had built something that made sense to people other than us. Q: How did you prepare for the launch? Adenekan Wonderful: Shedrack and I launched it together. I had also been through Product Hunt before: a previous product of mine landed at number seven, so I had a sense of how the site works and I was not overly anxious about it. I still prepared carefully. I did not want to be overconfident. We used the same approach I had used to launch earlier products. Q: Skillbox is a desktop app. What did you learn about launching one? Adenekan Wonderful: I went in expecting the install step to be a real barrier. I thought people would browse the site and leave. What surprised me was how quickly people signed up, installed it, and started syncing skills — about 50 installs in the first two or three hours. If the product solves a problem people already have on their machines, a desktop app is not the obstacle I feared it would be. The harder work is still making the first hour feel polished enough that they trust it. ### Part VI: The hard part Q: How many people use Skillbox today? Adenekan Wonderful: We are at a few hundred users, and about 3,000 synced skills. That is a lot when you think about what each skill represents, and it is still a size where we can actually hear from people. Users have emailed me about bugs, and we fixed them quickly. I love that. The architecture is built so we can manage skills at a much larger scale, but I would rather have honest numbers and real feedback than a figure we cannot learn from. Q: What was the hardest stretch of building Skillbox? Adenekan Wonderful: The hard stretch was figuring out the architecture that would actually work: the middle ground between getting the UX right and not giving people too much to deal with. We wanted something powerful enough for teams, versioning and security, without making everyday use feel heavy. Finding that balance took time. Q: Who believed in what you were building before there was anything to show? What did they say or do? Adenekan Wonderful: Shedrack believed in it early, and he did more than cheer from the side. He came in with the cloud, security, DevOps and documentation experience I did not have, and he helped shape the product. For my part I was already living with the problem, so I was my own first believer. And once we launched, the people who installed it in those first hours, and the users who take the time to email us when something is wrong, have meant a great deal. ### Part VII: What it is all for Q: What would make you say Skillbox has succeeded? Adenekan Wonderful: For me and for Shedrack, Skillbox has already become a daily driver. We have been using it every day for weeks, and it is a strange, good feeling to realise we have built something we ourselves cannot live without. Success is passing that same behaviour to the next 10,000 developers: people actually using it every day, and actively syncing their skills. That is the number I would look at. Q: If agents write and share their own skills, who should review them? Adenekan Wonderful: A human should. If an agent writes a skill, the person or team who is going to run it needs to review it before it reaches their machines, the same way they would review a skill from a public repo. In Skillbox, a captured skill lands in your library and stays private until you publish it. We help pull the reusable method apart from the names, paths and private details that belonged only to that session, and we still check the version. Trust can hold delivery, and executables still need an explicit opt-in. None of that replaces a person reading the instructions and the files. A passing check is not a sign-off. The more agents write and share their own skills, the more important that human review step becomes, especially inside a team. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Adenekan Wonderful: We want to start monetising Trim and get Skillbox into the hands of more teams. We already have interest from companies in San Francisco, including some YC companies, that want team-wide skills across sales, marketing and engineering. That is the next stretch: go from a daily driver for individuals to a library teams can actually run on. Q: And for the next 24 months? Adenekan Wonderful: In the next 24 months we want to sign more logos. Same direction, more teams actually running Skillbox as part of how they work. Q: Where do you want Skillbox to be in five years? Adenekan Wonderful: I want Skillbox to become a layer for how agents authenticate and communicate, not only a library for skills. You would install it in your organisation, connect something like Linear, and everyone who should have access would have it. Skills, MCPs, and context management sitting together as the knowledge and permission layer for agents. Q: What are your hopes and fears for a post AGI world? Adenekan Wonderful: I do not really have fears about a post-AGI world. Knowledge is power, and if we already have knowledge in our pockets, I genuinely do not see what we should be afraid of. I think it makes us faster as humans. Everything in life is a bit scary, and everything carries risk, but that does not stop us living every day. The risks are still there. We just live our lives. About Skillbox: Skillbox is one library for AI coding skills, synced across Claude Code, Cursor, Codex and your machines. It follows GitHub and GitLab sources, shows diffs, checks every version with Skillbox Trust and Cisco Skill Scanner, and includes Trim to keep agent context clean. Free to start; Individual from £6.99 a month. skillbox.so ## Scribble Cards (USA) URL: https://startup.talk/usa/scribble-cards/ Founders: Luke Emery Luke Emery grew up in Sidmouth, a quiet town on England’s southwest coast where most young people leave for the big cities. He washed dishes at 13, trained as a carpenter, studied architecture, ran a renovation company in London and then taught himself online marketing. His pet-portrait business, Van Woof, scaled to about 65 remote staff and 200 custom portraits a day in the run-up to Christmas. Then a podcast about handwritten letters clicked with his interest in robots. Scribble Cards has robots write each card with a real pen, then stamps and mails it across the US. It is self-funded, sends more than 30,000 cards a month, and lets customers send a card straight from ChatGPT or Claude. We spoke about making machine handwriting look human, why credits never expire, and why a handwritten note matters more in an age of AI. Key takeaways: - Robots hold a real pen. Scribble Cards built its own software that varies each letter and plans the whole card, and customers approve a preview before anything is sent. - It sends more than 30,000 cards a month to a few hundred business customers and thousands of consumers, and it is entirely self-funded. - There are no subscriptions or setup fees: one price per card, bulk discounts, and credits that never expire. - Cards can be sent from ChatGPT or Claude, through an MCP server, or automatically from a CRM via Zapier when a deal closes. The Principal / Founder: - Luke Emery (Founder) Based: Austin, Texas Grew up: Sidmouth, England Studied: Architecture Also runs: Van Woof, custom pet portraits Funding: Self-funded Website: scribblecards.com ### Part I: Before any of this Q: Where did you grow up, and what got you into art and business? Emery: I was born in London, but I grew up in Sidmouth, a small town on the southwest coast of England. It's a very quiet, rural area with small towns and a lot of nature. Most of the people who live there are older retirees, so there weren't many jobs, and most young people leave for London or one of the other big cities when they finish high school. I never heard of startups growing up, and I never met anyone who worked in an online business. The only business owners I knew owned shops, restaurants or construction companies. I was always interested in starting my own business. My first one was actually a construction company in London; we did small renovation projects. I decided to start an online business because I'd had enough of construction. I didn't like commuting, and I didn't like working on a building site, even in a managerial role. I found it boring and uninteresting, so about 10 years ago I started teaching myself online marketing and website design. As for art, I was always quite good at it growing up. Then when I studied architecture at university, there was a big emphasis on art and design. I improved my art skills there and definitely picked up good design skills at architecture school. Q: What did you study, and what was your first job? Emery: I studied architecture at university. From about 16 or 17, I'd been working in construction, first in laboring roles and then training as a carpenter, so becoming an architect seemed like a natural fit. After three years of university, I decided it probably wasn't the career for me and went into construction management instead. I've worked since I was 13, in so many different jobs. My first was washing dishes at an Italian restaurant. I'd get there at about 6 pm on a weeknight and finish around 11 or 11:30, and I always felt pretty rich for my age with all that extra pocket money. All through my late teens and 20s, I worked every random job you can imagine: every job on a building site, detailing cars, working on an oyster farm. If you can think of it, I probably did it. Q: You started vanwoof.com and other art businesses before this. What did those teach you? Emery: Vanwoof.com taught me a lot about managing large remote teams on time-sensitive work, because every portrait was an individual piece. 90% of the revenue comes in the run-up to Christmas, so we would scale up to about 65 remote employees over that period, pushing out 200 custom pet portraits a day. ### Part II: From art to robots Q: Where did the idea for Scribble Cards come from? Emery: I was looking for a new business to start, and I'd been getting interested in robots and robotics, even though I'm not a technical person, especially with hardware. Then I was listening to a podcast and heard someone talking about handwritten letters, and it clicked. In the age of AI, getting a handwritten note into someone's hands is a much better way to reach them than trying to get through by email. And it paired up really nicely with robots, because you could get robots to do the handwriting. Q: How did you go from an art business to running writing robots? Emery: I actually sold Vanwoof.com at one point and recently bought it back. Going from Van Woof to Scribble Cards felt like a very natural change, because they're both essentially personalized gifts. They're just packaged slightly differently. Q: Scribble became Scribble Cards. Why the new name? Emery: Another company in a similar industry was also using the name Scribble, and there are just too many companies calling themselves Scribble, so I wanted to stand apart. We mainly sell cards, so I went with Scribble Cards. Q: Who works on Scribble Cards with you, and how do you split the work? Emery: I have a number of people working with me, and they all work remotely. Some have been with me for years across other businesses I've run. We have virtual assistants who handle some of the customer service and order management. Then we have an operations person in the U.S. who does the printing, manages the writing machines and handles the packing. And we have some software developers and hardware developers. ### Part III: The product Q: Who is Scribble Cards for, and what job does it do for them? Describe the customer who gets the most value. Emery: Scribble Cards is mainly for businesses, although we also have a lot of consumers buying one-off cards for birthdays and thank-you notes. A lot of our business customers send cards to their clients or employees on their birthdays. Q: How do you view your positioning relative to Handwrytten, Simply Noted, LettrLabs, SendOutCards and Postable? Is there another competitor you think of first? Emery: Our product is really good. Our cards have the look and feel of a card you'd buy in a gift shop. We have some of the best handwriting technology available, so it looks like a real person wrote it, and we have a huge library of handwriting styles, most of which haven't even been released yet. Our software makes setting up cards really simple: sending one card or a thousand is just as easy, and it's been really well designed. We also have ChatGPT and Claude integrations, or you can connect through our MCP server, so your AI can send cards for you. Q: What is your core moat that prevents Handwrytten, Simply Noted or anyone else from replicating your feature set? Emery: Right now, I believe our software is a lot better than theirs, and we're constantly improving it and making sure our integrations all work really well. I'm just so focused on every customer having a great experience. At Van Woof, we were dealing with people's pets, and quite often the pet had died. That made me obsessed with going above and beyond for customers. If they're struggling with something, we help them, and we don't charge extra for it. We don't have subscriptions or setup fees. There's one price per card, with a bulk discount, and our credits never run out. It's all about customers having a great experience with us and keeping things as simple as possible. Q: What are some of the use cases for Scribble Cards that are not obvious and can add a lot of value to users? List as many as possible. Emery: The number one use case, and a massive value unlock, is for companies with leads they either never ended up working with or only worked with once. Sending them a holiday card or a birthday card puts you back on their radar. Some others: - Real estate agents sending past buyers a card on the anniversary of moving into their home. - Online stores putting a handwritten thank-you in the mail after someone's first order, or to their best customers. - Following up with people you met at a conference or event, instead of another email. - Sending a note before or after a big sales meeting so you stand out. - Winning back customers who cancelled or haven't ordered in a while. - Asking happy customers for a review or a referral. - Welcoming new hires before their first day, and marking work anniversaries. - Nonprofits thanking donors, so they give again. - Recruiters thanking candidates and the clients who hired them. - Vets, dentists and other local businesses sending sympathy, birthday or check-in cards to patients. - Property managers welcoming new tenants. - Thanking podcast guests, partners or anyone who has helped you out. - Hooking it up to a CRM through Zapier or AI, so a card goes out automatically when a deal closes. Q: A robot holds a real pen. How do you make the writing look human and not printed? Emery: We built our own software to make the handwriting look like a real person wrote it. A lot of companies use the same handwriting software, and I personally don't think it looks that real; it basically just tilts the words at a slight angle. Ours has loads of variations of each letter, and it plots out how the handwriting would look across the whole card. We show it to the customer first, so they can see it and decide whether they approve. Nothing is hidden from the customer before it goes out to whoever they're sending it to, and they can see it happening in real time. Then, to make it look unprinted, we use a standard pen, the kind people use to write by hand, so it really does look like handwritten text. You wouldn't be able to tell it was written by a robot. Q: You can send cards from ChatGPT and Claude. Why build for AI assistants so early, and how are people using it? Emery: I built it for ChatGPT and Claude because I'm a heavy AI assistant user, and I build things I would definitely use myself. We're seeing a lot of customers use it because they already have ChatGPT and Claude connected to all their other important apps, like their CRM and their email. Going into ChatGPT and saying, "Hey, can you send a handwritten card to Andy saying thank you?" takes a couple of seconds, and they don't even have to log into our app. It finds the right card for them and deducts it from their credits. It's really simple and a real unlock for our customers. Q: Why credits that never expire, and no monthly subscription? Emery: I like to keep things as simple and straightforward as possible for my customers. Once someone buys from us, they shouldn't feel pressured to upgrade or use up credits within a certain time. They can use our service however they want. If they want to lock in a bulk discount early and are willing to pay us a large amount up front, they should be able to keep that value for as long as they want. ### Part IV: Launch Q: When did Scribble Cards go live, and what happened in the first weeks? Emery: We first launched in August 2025. It was really an MVP back then. I was working with a couple of small customers while we built out the software and hardware, and it was very manual and hands-on. People would send me spreadsheets, and I'd work from those. Since then, we've just kept iterating and improving the hardware and software, and for the last three months we've been in a really good place with the quality of both. Q: How did you find your first customers? Emery: Our first customers came from my contact list. I know a lot of people in business, so I reached out to the ones I thought might be interested in the service. Q: What did early customers ask for that surprised you? Emery: Honestly, nothing surprised me. There's not much that's surprising about handwritten cards. ### Part V: The hard part Q: How is Scribble Cards funded today, and are you raising? Emery: It's completely self-funded. I've made enough money from my other businesses to fund it all myself. I don't like taking on investors or business partners, because I like the freedom to do whatever I want. Q: How many cards have you sent so far, and how many customers do you have? Emery: Right now, we send over 30,000 cards a month. We have a few hundred business customers and thousands of consumer customers. Q: Holiday season must be your busiest time. How do you keep up with demand? Emery: We built demand management into our operations software. It spots any high-risk days coming up that would need a lot of writing in one day and brings those jobs forward. Q: What is a challenging time you had building this company? Emery: Honestly, I like challenges, so nothing about building this company has really stressed me out. ### Part VI: What it is all for Q: What is the best reaction you have heard from someone who got one of your cards? Emery: I haven't heard from anyone who got one of our cards, because our customers are sending them to their own customers, so we're a couple of steps removed. The recipient shouldn't know we exist. That's the whole point of us: it should look like that person sent a handwritten card themselves. We're just a silent actor in the background. Q: Who believed in Scribble Cards before there was anything to show? What did they say or do? Emery: When I tell my entrepreneurial friends about the idea, they think it's very random, but they know I can take something a bit weird and random and make it work. People I know are always quite interested and think it's a good idea; they just think I come up with weird and wonderful ideas. Q: In a world of email and AI, why does a handwritten card still matter? Emery: I think handwritten cards matter now more than ever. People's inboxes are getting fuller, there's so much AI-written content, and fewer and fewer words are written by humans. That makes handwriting, handwritten letters and physical things more important than ever before. Q: What would make you say Scribble Cards has succeeded? Emery: I'd like to be sending a few million cards a year. I also have some dream customers I'd like to work with, and landing them would be a success too. ### Part VII: The road ahead Q: What comes next for Scribble Cards? Do you plan to mail outside the US? Emery: Yes, eventually we'll mail outside the U.S. I also plan to open an operations hub in the UK. I was born there and understand the business environment, so it just makes sense. Q: What are your milestones for the next 12 months, in customers and revenue? Emery: We're still a young company and growing fast, so I'd like to see the number of cards we send grow by 10% to 20% a month. Q: And for the next 24 months? Emery: Again, consistent growth, with a focus on winning new customers and keeping them happy so we have high retention. Q: Where do you want Scribble Cards to be in five years? Emery: I'd like everything to be about 50 times the size it is now: our operations, the number of customers we have and the number of cards we send. Q: What are your hopes and fears for a post AGI world? Emery: I'm an AGI optimist. I think it will only make all of our lives better. I see what we're going through now as a transitional period. I read a lot of economic history books, and similar things have happened before. Even in my lifetime, I can remember big changes, and people's jobs just changed to fit them. For instance, when I was 18 or 19, marketing was all about TV ads, newspaper ads and the written word. Within 10 years, it had moved completely online, and then onto social media. The difference this time is that the change is affecting everyone at once, rather than one particular industry. We're all going through it together, so I don't think it's a problem. I think it will just end up with everyone doing more interesting things. About Scribble Cards: Scribble Cards writes cards by robot with a real pen, then stamps and mails them across the US. Choose from 30 handwriting styles and 1,180 business card designs or your own artwork, set up recurring birthday and anniversary cards, or send from ChatGPT, Claude, Zapier or the API. One card is $6.99, credits never expire, and there is no subscription. scribblecards.com ## OkTreat (Canada) URL: https://startup.talk/canada/oktreat/ Founders: Patrick Leong Every time Patrick Leong travelled, he paid for a full month of boarding for Toto, his shy, leash-happy dog, and then spent the trip wondering how he was settling in. No photos, no messages. That silence became OkTreat, a Vancouver pet-care community where you look after a neighbour’s pet, earn TreatCoins, and spend them on care for your own months later, with anyone in the network. Leong is not a software engineer by training. He taught himself to build websites for his immigration consulting business, and when the developers he hired delivered a first version of OkTreat that fell short, he took over the code himself, with AI as his tutor. He launched in November 2026. We spoke about bartering time, a lost dog that rode with him for three days, and why he still walks with every new sitter before he leaves Toto behind. Key takeaways: - TreatCoins let owners give care when they have time and get it back later, from a different member, so nobody needs the same favour on the same dates. - A day of boarding is 1,440 TreatCoins, one coin per minute, though sitters set their own rates. - A lost dog he travelled with for three days led to OkTreat Smart Tags, which let any finder call the owner with a scan or a tap. - Leong is bootstrapped and says repeat exchanges in one neighbourhood matter more than raw sign-ups. The Principal / Founder: - Patrick Leong (Founder and CEO) Based: Vancouver, Canada Before: Programme Director in education; immigration consulting Code: Self-taught: Joomla, WordPress, HTML, CSS and PHP Website: oktreat.com App Store: OkTreat for iPhone (https://apps.apple.com/ca/app/oktreat-pet-sitting-boarding/id6479255523) Google Play: OkTreat for Android (https://play.google.com/store/apps/details?id=com.oktreat.app) YouTube: youtube.com/@oktreat Instagram: instagram.com/oktreatapp Facebook: facebook.com/oktreat X/Twitter: x.com/oktreat ### Part I: Before any of this Q: What first got you into building software and products? Leong: I wanted to build a website for my immigration consulting business. I had a clear idea of how it should look and what visitors should be able to do, but I found it hard to explain those details to someone else. I decided to learn how to build it myself so I could try things out and make changes as I went. I had no technical background, so I taught myself Joomla and WordPress and picked up the basics of HTML, CSS, and PHP along the way. It took time, but I built the site I had in mind. That experience gave me the confidence to take on a bigger project when I started OkTreat. ### Part II: The dog sitter bill Q: What did paying for a month of boarding for Toto teach you? Leong: When I travelled for a month, I used to pay for a full month of boarding for Toto. A month of boarding was a big expense, but what stayed with me was how little I knew about how he was doing while I was away. I didn’t get any photos, so I spent a lot of the trip wondering how he was settling in. That experience shaped what I wanted to build with OkTreat. I wanted a more affordable way to arrange care and a way to get to know the person looking after Toto before I left. Even a quick photo or message while I was away would have made a difference. Q: What is your dog like, and why was it so important to you that someone you trust looks after him? Leong: Toto is shy around strangers, but once he’s on his leash and out for a walk, he’s full of energy. I want someone who’ll give him time to warm up and enjoy getting outside with him. That’s why I like to meet a sitter before I leave. Even a short walk together can show me how they handle him on the leash and how comfortable he’s becoming around them. I can travel with more peace of mind when Toto has already had a chance to get to know the person looking after him. Q: When did the idea of trading care instead of paying for it come to you? Leong: About three years ago, I was thinking about what it cost to board Toto for a month while I travelled. I’d been on the paying side of pet care, and I started to think about what I could offer in return. When I’m home, I have time to look after someone else’s dog. Could I use that time to earn care for Toto on my next trip? The idea clicked when I realised the exchange didn’t have to be between the same two people. I could look after one neighbour’s dog now and get help from another neighbour months later. That’s why I built TreatCoins into OkTreat. They let you give care when you have time and get care back when you need it. Q: What was your background before OkTreat, and are you building it alone? Leong: I’m a Programme Director in education, and my background also includes immigration consulting. About three years ago, I started exploring whether pet owners could exchange their time by looking after each other’s pets. I researched how the exchange might work and asked a lot of people for advice. I hired developers to build the first version of OkTreat, but the app they delivered fell short of what I had in mind. I took over development and learned as I went. Today, I lead OkTreat and work directly on the app myself, using AI to help me understand unfamiliar code, troubleshoot problems, and test ideas. Q: How did Vancouver shape the idea? Leong: Living in Vancouver helped me see both the problem and a possible solution. It’s an expensive city, and when you’re planning a trip, paying for several weeks of pet care can add a lot to the cost. At the same time, I saw so many people walking their dogs in my neighbourhood. It made me think about how many of us were facing the same problem, often just a few streets apart. I wondered what would happen if we had an easier way to get to know each other and take turns looking after our pets. I could look after someone’s dog while they were away, earn TreatCoins, and use them to book care for Toto when I travelled. That’s how Vancouver shaped OkTreat. It made me see that the people who needed pet care could also be the people who provided it. ### Part III: The product Q: Why the name OkTreat? Leong: A treat is a small way to show your pet you care. I wanted the name OkTreat to capture that same feeling and extend it to how pet owners can help each other. With OkTreat, you can look after another member’s pet while they’re away, earn TreatCoins, and use them to book care for your own pet when you travel. That’s the idea behind the line we used on Product Hunt: Give pet care. Earn TreatCoins. Get care back. Q: How does the TreatCoins system work, and why a credit instead of cash? Leong: When you look after another member's pet, you earn TreatCoins. You can spend them later to book care for your own pet. A credit lets the exchange work across the community, even when two owners need help on different dates. Q: How do sitters set their rates in TreatCoins? Leong: Sitters can set their own rates in TreatCoins. I designed the system around a reference rate of 1,440 TreatCoins for 24 hours of boarding, based on one TreatCoin per minute. I used about $60 per day as a cash benchmark when designing that reference rate. The actual booking price depends on the rate the sitter sets. Q: Why let people chat and meet before booking? Leong: Every pet has a routine. A quick conversation can surface things a profile might miss, such as how a dog behaves around other pets or what helps it settle at night. Meeting beforehand also gives the pet a chance to get comfortable with the sitter. Q: How do you build trust between strangers who look after each other's pets? Leong: Trust starts before the booking. OkTreat uses Stripe Identity to verify sitters’ identities. Owners can also review a sitter’s profile, chat with them about their pet’s routine, and arrange a meet-and-greet before deciding whether to book. For me, those steps work together. Identity verification helps owners know who they’re speaking with. A conversation and a chance to see how the sitter interacts with their pet help them decide whether it’s a good fit. Q: Why Smart Tags with QR and NFC, and how do lost-pet alerts work? Leong: In November 2025, while I was travelling, I found a dog on the street and had no way to contact its owners. The dog ended up travelling with me for three days. On the last day of my trip, I left the dog with a local pet shop owner, who helped me report it to the local police. They found the owners right away. I kept thinking about how much sooner that dog could have got home if I’d had a way to call them. That experience inspired the OkTreat Smart Tag. If someone finds a pet, they can scan the QR code or tap the tag with a compatible phone to open the pet’s digital profile and call the owner. The profile can also hold vaccination records, vet details, and whether the pet is microchipped. If the finder chooses to share their location, the owner can see where the tag was scanned. There’s also a separate lost-pet alert. When an owner marks their pet as lost in OkTreat, the app is designed to notify members within 25 km so they can keep an eye out. Q: What is AI sitter matching, and how far along is it? Leong: AI sitter matching is already built into OkTreat. It compares your pet’s temperament and care needs with a sitter’s experience and availability. If your dog is shy around strangers, for example, OkTreat can suggest someone who has cared for shy dogs and is free on the dates you need. I designed the matching system to evolve over time. The first recommendation is based on what the pet owner and sitter tell us about themselves. As more bookings are completed, feedback about how the care went can help us understand which combinations work well and refine future suggestions. The goal is for each round of real bookings to teach us something that makes the next match better. Q: What is Pals, and why include a feature for solo travellers? Leong: PALS grew out of the community I was already building with OkTreat. Pet owners were getting to know each other through a shared love of animals, and I wanted to give that same community a way to meet and make friends while travelling. Once I’d arranged care for Toto, I could go on my trip. But travelling solo, I sometimes wished I had someone to share a meal with, go for a walk with, or explore a new city with. With PALS, you can meet locals and other travellers in the same city and make plans around something you both enjoy. That might be a dog walk, a coffee, or a visit to a place you both want to see. It started with dog owners and is open to solo travellers with or without pets. I also want people to feel more comfortable meeting someone new. That’s why I want PALS to use a live camera check to help verify that the person matches their profile photos. The idea is to give people more confidence about who they’re meeting before they make plans in person. Q: How do you solve the cold start problem of needing both sitters and owners in one city? Leong: I'd start neighbourhood by neighbourhood. A member can be both an owner and a sitter, so someone who joins looking for care can also help another member. The early goal is to have enough people close enough to each other to make a real booking practical. ### Part IV: Positioning Q: Who is OkTreat for, and what job does it do for them? Describe the user who gets the most value. Leong: OkTreat is for pet owners who sometimes need someone to look after their pet and are willing to do the same for others. A member might look after a neighbour's dog one weekend, then spend the TreatCoins on care for their own dog when they take a trip. Q: How do you view your positioning relative to Rover, Wag and TrustedHousesitters? Is there another competitor you think of first? Leong: People already have several ways to arrange pet care, from hiring a sitter to asking friends. OkTreat's approach is reciprocal care across a community. You can help one member today and use the credits with a different member later. Q: What is your core moat that prevents a big pet care marketplace from replicating your feature set? Leong: I think the biggest barrier is the business model. A marketplace that earns a fee on each paid booking has little incentive to replace those bookings with an exchange of time. If customers start trading care they would otherwise pay for, the marketplace could lose booking fees. Making reciprocal care central to the business would mean changing how it earns money and why people use it. I built OkTreat around that exchange from day one. You can look after a neighbour’s dog, earn TreatCoins, and use them to book care for your own pet later. For that to work, you need enough people nearby who enjoy caring for pets and trust each other enough to keep coming back. A competitor could copy the credits. It would still have to build that community neighbourhood by neighbourhood while risking revenue from its existing bookings. Q: What are some of the use cases for OkTreat that are not obvious and can add a lot of value to users? List as many as possible. Leong: I started OkTreat because I needed care for Toto while I travelled, but it’s just as useful close to home. If you have a free weekend, you can look after a neighbour’s dog and earn TreatCoins. You can put those credits toward care later when your work schedule changes, you have a family commitment, or you’re heading out of town. The person who helps you can be a different member, so you don’t need to find someone who needs the same favour at the same time. You can also book a short stay before a longer trip. That gives your dog time to get comfortable with the sitter and gives you both a chance to talk through meals, walks, and bedtime. A trial stay can be especially helpful for a pet that takes time to settle somewhere new. The Smart Tag is useful between bookings, too. Your pet’s digital profile can hold vet details, care instructions, and whether your pet is microchipped. If your dog slips out through an open gate, the person who finds them can scan the QR code or tap the tag with a compatible phone to call you. They can also choose to share where they scanned it, giving you a place to start looking. Q: What is the one detail in OkTreat you are proudest of? Leong: I’m proudest that I turned a simple idea into something that works: you can barter your time for pet care. I can look after another member’s dog for a weekend, earn TreatCoins, and use them months later to book care for Toto. The person who helps me can be someone else in the community. We don’t have to need care at the same time to help each other. AI matching helps owners find someone who understands their pet’s needs, but the exchange itself is what I’m proudest of. The time I spend helping another pet owner today can help me take care of Toto when I need it. ### Part V: Launch Q: What did you learn from the launch? Leong: The launch taught me how important it is to explain OkTreat in plain language. “Reciprocal pet care” describes the model, but a real example makes it easier to picture. You look after someone’s dog for a weekend, earn TreatCoins, and use them months later to book care for your own dog. The person who helps you can be someone else in the community. That’s why I’d lead with the exchange itself when introducing OkTreat. Once people understand how giving their time now can help them get care later, they can decide whether it would work for them. Q: Which feedback from the launch surprised you? Leong: What surprised me was how much people liked the idea of earning pet care by giving it. The part that seemed to click was that you could look after someone’s dog when you had a free weekend, earn TreatCoins, and use them to book care for your own pet months later. The person who helps you can be someone else in the community. That flexibility is what makes the exchange practical. Q: What did the launch do for sign-ups and bookings? Leong: It’s still early. So far, the Product Hunt launch has brought feedback on the idea of exchanging pet care. I haven’t seen a measurable increase in sign-ups or bookings yet. ### Part VI: The hard part Q: What is a challenging time you had building this company? Leong: One challenge has been making the exchange easy to understand. People are used to thinking of themselves as either an owner or a sitter. With OkTreat, the same person can do both. The app has to make each role clear while keeping the booking experience simple. Q: What is the hardest part of running a marketplace where people trade care? Leong: The hardest part is getting enough reliable care close by when people need it. A member needs someone who is available on the right dates and comfortable with their pet. Repeat exchanges in the same neighbourhood are a better sign of progress than sign-ups alone. ### Part VII: What it is all for Q: What would make you say OkTreat has succeeded? Leong: I'd call it a success when members regularly book care with TreatCoins they've earned and recommend the app because they found people they'd trust with their pets again. Q: What does a community that looks after each other's pets mean to you? Leong: It means being able to make plans without scrambling for pet care every time. It also means knowing people nearby well enough to return the favour when they need help. ### Part VIII: The road ahead Q: What are your milestones for the next 12 months, in users and revenue? Leong: Over the next year, I want to grow repeat bookings in the communities we serve, improve sitter matching, and make it easier for members to get help through the app. I’ll track completed care exchanges, how often members come back to book again, and revenue to see where we’re making progress. I also want to make OkTreat more useful throughout a trip. Once someone has arranged care for their pet, PALS can help them meet locals or other solo travellers and make plans together. I’d like to build on that with features travellers can use on the road, such as finding local activities, organising day trips, and getting mobile data through an eSIM. Q: And for the next 24 months? Leong: Over two years, I'd like to take what works in our strongest communities into more cities. I also want to keep improving Smart Tags and develop PALS around the kinds of meetups and activities people actually use. Q: Where do you want OkTreat to be in five years? Leong: In five years, I'd like OkTreat to be a service pet owners can rely on in many cities. I want someone to be able to plan a trip knowing they have a practical way to arrange care, and to feel that their own time spent helping other owners counts. Q: What are your hopes and fears for a post AGI world? Leong: My hope is that more people get the chance to build something they’ve had in mind for years. I came from immigration consulting and education, and AI helped me understand unfamiliar code and take over development of OkTreat. I’d like more people with firsthand experience of a problem to have the tools to do something about it. My worry is that we’ll start trusting AI’s recommendations more than our own judgement. In OkTreat, AI can help me find a sitter who has experience with shy dogs like Toto. I still want to take a walk with that person and see how Toto responds before I decide. I hope we use more capable AI to make those decisions easier while keeping people responsible for them. About OkTreat: OkTreat is a reciprocal pet-care community based in Vancouver. Members look after each other’s pets and earn TreatCoins to book care for their own. It also makes Smart Tags, QR and NFC pet ID tags, and Pals, which helps solo travellers meet locals. oktreat.com ## Dots (Philippines) URL: https://startup.talk/philippines/dots/ Founders: Jakob Edhel Puri Jakob Edhel Puri grew up in the Philippines taking apart electronics and tinkering with computers. Whenever a task on a screen felt clunky, his first instinct was to build a faster way around it. Today he studies engineering, from communications and electronics to advanced mathematics, and builds Mac apps. Dots started at his own desk. Studying and coding on a MacBook, he kept switching between his code editor, a browser, YouTube lectures and screenshot tools, and lost his focus every time. Then he looked up at the notch, sitting in the middle of his screen doing almost nothing, and turned it into a workspace: launchers, a video player you can pop out, scrolling screenshots, a teleprompter under the camera and local file sharing, with a $5 one-time Pro instead of a subscription. We spoke about building and selling Mac apps from the Philippines, what he leaves out on purpose, and why he cares about local-first software. Key takeaways: - Dots turns the MacBook notch into a workspace you reach by moving the cursor up. On Macs without a notch and on external displays it anchors to the top-centre edge, so the habit stays the same. - Tools come out of the notch: detach a video player, resize it anywhere on screen and snap it back without stopping playback. - Everything runs locally. There are no mandatory accounts, file conversion happens on the Mac, and Send nearby moves files without a third-party server. - Dots is free to use, Pro is a $5 one-time licence, and around 100 to 300 people use it today. The Principal / Founder: - Jakob Edhel Puri (Founder) Based: Philippines Studies: Engineering: communications, electronics and advanced mathematics Also builds: Local DevOps and observability tools Website: dotsapp.cloudaeri.com ### Part I: Before any of this Q: Where did you grow up, and what was childhood like? Puri: I grew up in the Philippines, in an environment where curiosity was always the best entertainment. Growing up, I was naturally drawn to how things worked under the hood—whether that meant taking apart electronics, tinkering with computers, or figuring out how to solve problems with whatever tools were available. It taught me early on to be resourceful and self-reliant. Q: What first got you into building things? Puri: A mix of impatience and curiosity. Whenever I hit a repetitive task or a clunky workflow on a computer, my first instinct wasn't to accept it—it was to figure out if I could script or build a faster way around it. That habit eventually turned into a passion for software engineering, electronics, and building tools from scratch. Q: What was your first job, and what did it teach you? Puri: My earliest professional work involved taking on short, high-intensity software engineering sprints—turning rough prototypes into production-ready applications and integrating local AI workflows. It taught me two critical lessons: speed matters, and shipping something reliable that actually solves a user's bottleneck is infinitely more valuable than over-engineering in isolation. ### Part II: The road to Dots Q: What do you do besides building apps, and how did you start? Puri: Outside of indie app development, I study engineering—diving into everything from communications and electronics to advanced mathematics—and I build local DevOps and observability tools. I started building apps simply because I wanted better tools for my own daily workflow as a student and developer. Q: You describe yourself as building apps with AI. How does that change how you work? Puri: Working with AI compresses the distance between an idea and a working prototype. Instead of spending days boilerplate-coding or getting stuck on unfamiliar native APIs, I can iterate on architecture, test edge cases, and refine UI interactions in hours. It lets a solo builder operate with the output of a small studio—while keeping full control over local, privacy-respecting code. Q: What was the moment you decided the notch deserved more than a camera? Puri: I was studying and coding on my MacBook, constantly Command-Tabbing between my IDE, a browser for research, YouTube lectures, and screenshot tools. Every time I switched windows, I lost my flow state. I looked up at the top of my screen and realized the notch was sitting right in the center of my visual field doing nothing—while most notch apps only used it for basic music controls. That was the moment I decided to turn it into a real workspace. Q: What did the first working version do, and who used it first? Puri: The earliest version was a lightweight top-center shelf that let me pin quick tools and switch contexts without stealing focus from my active window. I was the first daily user, followed closely by university classmates and groupmates who needed a faster way to juggle research, lectures, and screen captures during crunch time. Q: What is it like to build and sell Mac apps from the Philippines? Puri: It’s a superpower with a few unique hurdles. Building from the Philippines gives you global leverage—you can build world-class software from your bedroom and sell it to users in the US, Europe, and Asia. The challenge is navigating international payment processors, tax compliance, and timezone gaps for launches, while realizing that marketing and distribution are often twice as hard as writing the code. ### Part III: The product Q: Why the name Dots? Puri: "Dots" represents small, understated focal points that connect your entire workflow together. A dot is minimal, unobtrusive, and takes up almost no space—yet when you connect them, you get a complete picture. It fits an app that lives quietly at the top edge of your screen until you need it. Q: Why the notch as the home for a workspace? Puri: Because your eyes naturally rest near the top third of the screen, and the notch is hardware real estate you already have. Putting a workspace there means you don't have to clutter your desktop with floating widgets or lose your place by switching full-screen spaces. You just glide your cursor up, grab what you need, and drop right back into your work. Q: Why a teleprompter under the camera? Puri: When you're recording a product demo, a class presentation, or a video update, looking down at notes on your screen makes it obvious you're reading. Placing a teleprompter directly beneath the camera lets you speak naturally while maintaining eye contact with the lens. Q: Why scrolling screenshots and a citation converter? Puri: Those came directly from real friction in studying and building. When you're researching papers, documenting code, or capturing long threads, stitching screenshots manually or reformatting citations breaks your momentum. Having scrolling capture and instant format/citation conversion built right into the notch turns a 5-minute interruption into a 3-second action. Q: Why local file sharing with Send nearby? Puri: Because emailing files to yourself or uploading sensitive documents to a cloud drive just to move them across the room is absurd. Local peer-to-peer sharing keeps transfers instant, works offline, and ensures your files never touch a third-party server. Q: Why a $5 one time Pro rather than a subscription? Puri: People are exhausted by subscription fatigue—especially for utilities. A utility should feel like a trusty tool you buy once and own forever. Keeping a generous free tier and pricing Pro at a $5 one-time license removes all friction for someone trying it out while building genuine goodwill. Q: Why mic auto mute after silence? Puri: It's a small safety net for screen recordings. If you're recording a long walkthrough or study session and step away or pause to think, auto-muting Dots' recording mic after 30, 60, or 120 seconds of silence prevents awkward background noise without touching your actual meeting apps. Q: What do you deliberately leave out of Dots? Puri: Cloud bloat, mandatory accounts, heavy background telemetry, and gimmicky animations. Everything in Dots is designed to run natively, stay lightweight on system resources, and keep your data local on your Mac. ### Part IV: Positioning Q: Who is Dots for, and what job does it do for them? Describe the person who gets the most value. Puri: Dots is built for students, developers, and creators who live inside deep-focus windows—like an IDE, a Figma canvas, or a research paper—and hate losing their train of thought to window management. The person who gets the most value is someone who juggles multiple mediums at once: watching a lecture or tutorial while coding, taking quick annotated captures, running Pomodoro sprints, and firing off custom shortcuts without ever leaving their primary workspace. Q: How do you view your positioning relative to Alfred, NotchSpace and OmniNotch? Is there another competitor you think of first? Puri: Alfred is a keyboard-driven launcher; it's incredible, but it's transient—it disappears the moment you execute a command. Most notch utilities like NotchSpace or OmniNotch focus heavily on media playback, aesthetics, or simple file shelves. Dots is positioned as an active, persistent micro-workspace. You can keep a video playing, run a timer, annotate a live screen, convert files, and trigger developer workflows all from one unified hub. Q: What is your core moat in a crowded space of notch apps? Puri: Anyone can wrap a media player around the notch in a weekend, but building a cohesive suite—detached picture-in-picture video, scrolling capture, screen drawing, local conversion, SSH/dev tools, and a teleprompter—that feels native and fast is much harder to replicate. Q: What are some of the use cases for Dots that are not obvious and can add a lot of value to users? List as many as possible. Puri: Picture-in-picture lecture & coding mode: Dragging a YouTube tutorial onto the notch, detaching it to resize while coding, and snapping it back into the notch without restarting playback. Live meeting whiteboard: Dragging the Draw tool out of the Capture pane during a screen-share to sketch arrows and highlight UI bugs directly over any app. Zero-cloud file conversion: Dropping images, audio, or video onto the notch to locally convert them (e.g., to PDF, PNG, M4A, or MP4) without uploading private files to random web converters. Remote server triggers: Running quick SSH services, agent checks, or developer scripts straight from the top edge of the screen without opening a terminal window. Camera-level script reading: Using the built-in teleprompter for Loom updates or asynchronous standups so your eyes stay locked on the camera. Q: What is the one detail in Dots you are proudest of? Puri: How you can detach a video player or tool from the notch, resize and move it anywhere on your screen, and return it back into the notch without losing state or interrupting playback. It makes the notch feel alive rather than rigid. ### Part V: Launch Q: What did you learn from the launch? Puri: Launching on Product Hunt in October and November 2026 taught me that building the product is only half the battle—distribution is a completely different discipline. Launch day spikes are great for validation, with days hitting over 50 downloads, but sustainable growth comes from talking directly to users in niche communities where people are actively complaining about the exact friction you solve. Q: Which feedback from the launch surprised you? Puri: How many people without a notched MacBook—or people using external monitors—still wanted to use Dots. That pushed me to make sure the experience worked seamlessly on non-notched displays and external monitors, turning it from a "notch gimmick" into a universal top-edge workspace. ### Part VI: The hard part Q: How many people use Dots today? Puri: Right now, we’re in our early growth stage with around 100 to 300 users, and we're seeing steady, organic adoption as more people discover it. Q: What is a challenging time you had building Dots? Puri: Balancing deep technical polish with the grind of marketing while juggling university engineering coursework. There were moments where fixing subtle macOS window-level bugs across multi-monitor setups took days of trial and error, followed immediately by the realization that nobody knows your app exists unless you actively put it in front of them. Q: Who believed in you before there was anything to show? What did they say or do? Puri: My close peers and classmates. Even when Dots was just a rough prototype running locally on my MacBook, they tested early builds, gave honest feedback on what felt clunky, and encouraged me to ship it publicly instead of keeping it as a personal tool. Q: How do you design for Macs with and without a notch? Puri: By treating the notch as an anchor point rather than a hard dependency. On notched MacBooks, Dots hugs the hardware cutout seamlessly; on non-notched Macs and external displays, it anchors cleanly to the top-center edge of the menu bar so the gesture and muscle memory remain identical everywhere. ### Part VII: What it is all for Q: What would make you say Dots has succeeded? Puri: When Dots becomes one of the first three apps a Mac user installs on a brand-new machine—not because it's flashy, but because working without it feels noticeably slower. Q: What does a day look like when your whole workspace sits at the top of the screen? Puri: Your desktop stays completely clean. You open your main work window full-screen, start a Pomodoro timer in the notch, keep a reference video or documentation tucked at the top edge, snap and annotate screenshots on the fly, and share files locally—all without ever breaking your visual focus. ### Part VIII: The road ahead Q: What are your hopes and fears for a post AGI world? Puri: My hope is that superintelligent tools democratize creation completely—allowing a single person anywhere in the world to build, cure, and invent things that used to require massive institutions. My fear is that we trade away human agency, privacy, and craftsmanship, letting centralized cloud systems own all our data and thinking for us. That's why I care so much about local-first software and tools that augment human focus rather than replacing human intent. About Dots: Dots is a customizable workspace built into the Mac’s notch, with launchers, web spaces, shortcuts, media controls, notes, a Pomodoro timer and file tools. It works on Macs with and without a notch. Free to use; Pro is a $5 one-time licence for one Mac. dotsapp.cloudaeri.com ## Alphabet (USA) URL: https://startup.talk/usa/alphabet/ Founders: Larry Page, Sergey Brin Founded: 1998 (incorporated September 1998; search project began 1996 as BackRub); Based in: Mountain View, California; Sector: Internet search, advertising, cloud, AI In May 2013, just before Google's developer conference, Larry Page posted a note explaining something people had long noticed: the chief executive of Google spoke in a strained, raspy voice. A bad cold around 1999, he wrote, had left him hoarse, and it never fully cleared. A doctor later diagnosed paralysis of his left vocal cord. Another cold, in the summer of 2012, showed his right cord moved poorly too. He said having both affected is very rare, and that no cause was found. Fourteen years of that, and the world had not been told. He grew up in Lansing, Michigan, in a house full of computers. His father was a computer science PhD and a Michigan State professor; his mother taught programming. His parents divorced when he was eight. When he was six, his father brought home an Exidy Sorcerer, and he mastered it. He has said that from around the age of 12 he knew he would start a company. Sergey Brin's start was harder. Born in Moscow in 1973, he was six when his family left the Soviet Union, after his father decided antisemitism made staying impossible. They reached the United States in October 1979. What almost nobody connects to the voice is the year. 1999 was also the year Kleiner Perkins and Sequoia put $25 million into Google, and asked for a grown-up to run it. Page and Brin turned down several candidates. Sequoia's Michael Moritz reportedly warned he could ask for $12.5 million back if no CEO was hired. John Doerr proposed Eric Schmidt, who bought $1 million of preferred stock to show he meant it. The two had already tried to get out once. They offered the search engine to Excite for about a million dollars, and Vinod Khosla has said he pushed the price down to $750,000. Excite said no. In August 1998 Andy Bechtolsheim had written a $100,000 cheque, and they had to incorporate to cash it. Their first office was a garage in Menlo Park. Google went public on 19 August 2004 at $85 a share, raising $1.67 billion. Since October 2015 it has sat inside Alphabet, which booked about $403 billion in 2025 revenue and was worth about $3.46 trillion on 6 November 2026. He named it in public in 2013, after leading a company worth hundreds of billions with a voice he could not fix. Some of the hardest parts of building something never reach the pitch deck. They show up much later, in a short post. ## Amazon (USA) URL: https://startup.talk/usa/amazon/ Founders: Jeff Bezos Founded: 1994 (founded 5 July 1994 in Bellevue, WA as Cadabra; site launched 16 July 1995); Based in: Seattle, Washington; Sector: E-commerce, cloud computing (AWS), logistics, streaming Jeff Bezos has said that when he told his boss at the hedge fund D. E. Shaw about an online bookstore, the reply was not a cheer. The boss thought it would be a better idea for someone who did not already have a good job. Bezos was thirty and a senior vice president. He had been born Jeffrey Preston Jorgensen in Albuquerque in 1964 to a mother of seventeen and a father of nineteen, then adopted by Mike Bezos, a Cuban immigrant. He grew up in Houston and Miami and graduated from Princeton in 1986, summa cum laude. So the decision to go was a hard one, and he made it by turning it into a question. He pictured himself old and asked which he would regret more: skipping a Wall Street bonus or missing the start of the Internet. He and his wife, MacKenzie Scott, drove from New York to Seattle in 1994 and wrote the plan on the way. What almost nobody connects to the Everything Store is a car ride at ten. In 2010 he told Princeton's graduating class about it. His grandparents took him along on summer road trips, and in the back seat he worked out, from an advertisement, that each puff of his grandmother's cigarette cost her two minutes of life. He announced the total: nine years. He expected praise for the arithmetic. She burst into tears. His grandfather pulled over, walked around and opened Jeff's door. He told him, as Bezos recalled it, that it is harder to be kind than clever. Bezos said he did not know whether he was in trouble. The early money was personal too. His parents put in an estimated $300,000. Amazon began on 5 July 1994 in a rented garage in Bellevue, as Cadabra, and he told early investors there was a 70 percent chance it would fail. The site opened on 16 July 1995. When publishers asked why a shop would carry bad reviews, he said he wanted every book there, the good, the bad and the ugly. The company went public on 15 May 1997 at $18 a share. In 2002 it launched Amazon Web Services, and the spare computing power became a business of its own. In 2025 Amazon reported revenue of $716.9 billion and net income of $77.67 billion. Yet when he stood before those graduates in 2010, the story he chose was not about any of that. It was about being a clever boy who made someone he loved cry. He told them cleverness is a gift and kindness is a choice. For a new founder the quiet lesson sits under the big one: you can be right about the numbers and still wrong about the person across from you. ## SpaceX (USA) URL: https://startup.talk/usa/spacex/ Founders: Elon Musk Founded: 2002 (14 March 2002, El Segundo, California); Based in: Starbase, Texas (earlier Hawthorne, California); Sector: Aerospace, launch services, satellite internet (Starlink), AI (xAI) "We were running on fumes at that point." That is how Elon Musk described 2008 to Scott Pelley on 60 Minutes. He called it the worst year of his life. The details are bleak. Falcon 1, SpaceX's small rocket, had failed on its first three launches. Tesla was also short of cash. Most of what he had made from PayPal had gone into the two companies, and later reporting says he stayed on friends' couches and borrowed money to cover rent. That last detail comes from secondary coverage, not from Musk himself. What he said on camera was blunter: a fourth failure would have been "absolutely game over." How did a man with that much money end up there? He was born in Pretoria in 1971, left South Africa for Canada as a young man, and finished degrees in physics and economics at the University of Pennsylvania. Compaq bought Zip2 in 1999 and his share was about $22 million. When eBay bought PayPal in 2002 for $1.5 billion in stock, he received about $175.8 million as the largest shareholder. Then came the Moscow trip. In autumn 2001 he flew there to buy a rocket, planning to put a greenhouse on Mars as a stunt. His Russian contacts were not interested, and Michael Griffin, who went with him, watched them grow less friendly with each meeting. On the flight home Musk decided he could build the rockets himself. SpaceX was founded on 14 March 2002 in El Segundo. Musk put in about $100 million. He hired Tom Mueller, Gwynne Shotwell and engineers from TRW and Boeing, and interviewed employees himself. Then the failures, from 2006 to 2008, which nearly ended the company just as Tesla needed rescue too. The fourth Falcon 1 reached orbit on 28 September 2008. That December NASA awarded SpaceX a $1.6 billion contract to resupply the space station. Asked whether the contract saved the company, Musk answered, "Yeah, they did." What followed was a ladder of firsts: Falcon 9 and Dragon in 2010, an upright booster landing in December 2015, a reflown rocket in 2017. In January 2015 SpaceX raised $1 billion at about a $12 billion valuation and announced Starlink. It acquired xAI in February 2026 and listed on Nasdaq as SPCX on 12 June 2026, with $18.7 billion of 2025 revenue and a $4.9 billion net loss. Its market value was about $1.68 trillion on 6 November 2026. The quiet fact is how close the margin was. A company now worth trillions depended on one rocket working, in the last months its founder could pay for it. ## Meta Platforms (USA) URL: https://startup.talk/usa/meta-platforms/ Founders: Mark Zuckerberg, Eduardo Saverin, Dustin Moskovitz, Andrew McCollum, Chris Hughes Founded: 2004 (4 February 2004, Cambridge, Massachusetts); Based in: Menlo Park, California; Sector: Social media, advertising, AI, virtual and augmented reality Mark Zuckerberg's father filmed him opening the email that said he had been accepted to Harvard. Years later Zuckerberg told the graduating class of 2017 that he still believes getting in is the thing his parents are most proud of. He was born in White Plains in 1984 and raised in Dobbs Ferry. He went to Ardsley High School, moved to Phillips Exeter Academy, captained the fencing team, and took a graduate computer course on Thursday evenings at Mercy College. He arrived at Harvard to study psychology and computer science. Then came the weekend. As a sophomore he built Facemash, pairing photos from dorm books and asking visitors to pick who was hotter. It overwhelmed a network switch, the university shut it down and students complained about their pictures. He apologised publicly and faced the administrative board. What the speech added, and few profiles mention, is who was there. When he faced possible expulsion, he said, his parents came to help him pack. Friends threw him a going-away party. He told the graduates he had done it to himself. He was not expelled. On 4 February 2004 he launched TheFacebook with Eduardo Saverin, Andrew McCollum, Dustin Moskovitz and Chris Hughes. Zuckerberg left Harvard in his second year. He and Saverin each put in $1,000, then about $18,000 more apiece. That summer in Palo Alto, Peter Thiel paid $500,001 for 10.2 percent. Accel put in $12.7 million in 2005 at a $98 million valuation, and the company paid $200,000 for the facebook.com domain. The low point was not money. In the same speech he said that when the fight over selling the company split people around him, he felt alone. He wondered if he was just wrong, an imposter, a 22-year-old who did not know how the world worked. In September 2006 talks with Yahoo had reached as much as $1 billion. The company stayed independent. The cost came from the people nearest the start. Facebook sued Saverin over a 2005 stock agreement, he countersued, and they settled in 2009. The 2012 IPO at $38 a share raised about $16 billion, but the stock fell 16.5 percent in the first week and lawsuits followed. In 2025 the company, renamed Meta in October 2021, reported revenue of $201 billion. For a first-time founder, the part worth keeping is small. The same parents who filmed the acceptance packed the boxes, and he remembered both. Early on, the people who stay when you have failed matter more than the next product. ## Tesla (USA) URL: https://startup.talk/usa/tesla/ Founders: Martin Eberhard, Marc Tarpenning, Ian Wright, JB Straubel, Elon Musk Founded: 2003 (incorporated 1 July 2003; Musk joined with Series A in Feb 2004); Based in: Austin, Texas; Sector: Electric vehicles, energy storage, autonomy and AI The round closed at 6pm on Christmas Eve 2008. Elon Musk wrote years later that it was the one that kept Tesla alive. To understand why, start with Martin Eberhard, who kept seeing a Toyota Prius parked beside a luxury sports car in Palo Alto driveways. He grew up in Kensington in the Berkeley Hills and studied computer and electrical engineering at Illinois. In 1996 he and Marc Tarpenning started NuvoMedia, which made the Rocket eBook, and they sold it to Gemstar in 2000. Hunting for lithium-ion cells for the reader left him thinking about electric cars. Alan Cocconi, who built the AC Propulsion tzero, declined to make a production car. So Eberhard and Tarpenning incorporated Tesla Motors on 1 July 2003. Money followed. In February 2004 the company raised $7.5 million, $6.5 million of it from Musk, who became chairman. JB Straubel joined in May 2004. A $13 million round came in 2005. Then the founders fell out. In 2007 the board, led by Musk, asked Eberhard to step down as chief executive. He left in January 2008, and sued Musk for libel, slander and breach of contract in June 2009. He dropped it in August, and a September settlement let all five men call themselves co-founders. Musk took over as CEO in October 2008. What almost nobody remembers about that Christmas Eve is how personal the money was. In a November 2020 post, Musk wrote that he had put in all the money he had, did not own a house, and had to borrow from friends to pay rent. Tesla had cut about 10 percent of its staff that year, and a site called Tesla Death Watch counted down its days. The December round of about $40 million kept it alive. By January 2009 Musk had put in $70 million of his own, and in May 2009 Daimler paid a reported $50 million for a stake of under 10 percent. Musk later said that deal saved Tesla. The company listed on 29 June 2010 at $17 a share, the first American car company IPO since Ford in 1956. Years later came 'production hell' around the Model 3. In 2025 Tesla reported revenue of $94.83 billion. The founder who started it left with a lawsuit and a settlement. The one who saved it had no house and borrowed rent. For a founder about to put in everything, the lesson is plain: you can survive losing your savings, but settle early who will be standing next to you, and under what name. ## Anthropic (USA) URL: https://startup.talk/usa/anthropic/ Founders: Dario Amodei, Daniela Amodei, Jared Kaplan, Jack Clark, Chris Olah, Sam McCandlish, Tom Brown, Ben Mann Founded: 2021 (incorporated 26 Jan 2021); Based in: San Francisco, California; Sector: Artificial intelligence (frontier AI models, Claude) Dario Amodei was in his early twenties when his father died. Riccardo, a leather craftsman, had been ill for years, and in 2006 the illness won. Dario had been heading for theoretical physics. After the funeral he moved his graduate work at Princeton toward biology. What almost nobody outside his circle knew was the cruel detail in the timing. Dario has said his father died because cures arrived a few years too late. Within about four years, a breakthrough took the illness from roughly half of patients dying to about 95 percent being curable. He was never going to get that time back. He grew up in San Francisco, born in 1983, the son of Riccardo and a library project manager named Elena Engel. He went to Lowell High School, made the US physics olympiad team in 2000, and later studied at Caltech, Stanford and Princeton. His sister Daniela, four years younger, took the humanities route: English literature at UC Santa Cruz, a classical flute scholarship, then Stripe in 2013. Biology led Dario to the retina and then to computers. He worked at Baidu and Google Brain, and in 2016 joined OpenAI, where he worked on GPT-2 and GPT-3. Daniela followed in 2018 and ran safety and policy. In January 2021 they walked out together with Jared Kaplan, Jack Clark, Chris Olah, Sam McCandlish, Tom Brown and others, and incorporated Anthropic as a public benefit corporation. Dario has said he gets angry when people call him a doomer; his warnings, he argues, exist so that society does not have to slow down. Safety talk needs money, and the money was not always clean. In 2022 the company raised $580 million, $500 million of it from FTX under Sam Bankman-Fried. Amazon invested $8 billion in all, and Google followed. The valuation climbed from $61.5 billion in March 2025 to $183 billion that September. The bill came due. Anthropic agreed to pay $1.5 billion to authors whose pirated books had been used in training. In early 2026 it refused to drop limits on mass domestic surveillance and fully autonomous weapons; the president ordered federal agencies to stop using its models and the Pentagon labelled it a supply chain risk. A judge granted a temporary injunction in March. A May 2026 round valued it at $965 billion. Revenue was $4.6 billion in 2025 and $16.3 billion in the first half of 2026, and its IPO filing went public on 26 October, with a listing reported planned for November. The father's story is the one thread that never changed. A founder's reason does not have to be a strategy. It only has to be a loss that still hurts, because that is what you will still be working for on the bad days. ## OpenAI (USA) URL: https://startup.talk/usa/openai/ Founders: Sam Altman, Greg Brockman, Ilya Sutskever, Elon Musk, Trevor Blackwell, Wojciech Zaremba, John Schulman, Andrej Karpathy, Durk Kingma, Vicki Cheung, Pamela Vagata Founded: 2015 (December); Based in: San Francisco, California; Sector: Artificial intelligence (ChatGPT, GPT models) On 17 November 2023 the board of OpenAI fired its chief executive. Sam Altman was told he had not been consistently candid. His co-founder Greg Brockman quit. Among the votes was Ilya Sutskever, who four years earlier had officiated Brockman's wedding in the OpenAI offices. It was not the first time a board had come for Altman. He was born in Chicago in 1985, the eldest of four, and the family moved to Clayton, Missouri, in 1989. At eight he got a Macintosh and began coding. At 17 he came out as gay at his high school, speaking up after some students objected to a National Coming Out Day speaker. He went to Stanford for two years, and at 19 he left to start Loopt, a location app. What almost nobody remembers is how Loopt went. He raised more than $30 million in venture capital as its chief executive, yet the app never gained much traction. Loopt's board tried twice to remove him as CEO; his supporters said the attempts were not serious. Green Dot bought the company in March 2012 for $43.4 million, a modest outcome after seven years. OpenAI began in December 2015 as a nonprofit in Brockman's living room. The pledge from Altman, Brockman, Elon Musk, Reid Hoffman, Peter Thiel and others was about $1 billion, but only about $133 million had arrived by 2021. Musk resigned in 2018. In 2019 OpenAI became a capped-profit company, Microsoft put in $1 billion and Altman became CEO. ChatGPT launched in November 2022, and Microsoft added $10 billion two months later. The firing came after that. Hundreds of staff threatened to leave with him, and Microsoft's chief said Altman would join Microsoft. Five days later Altman was back and the board was replaced. Sutskever later said he regretted taking part. A former board member said Altman had withheld information, which his side disputed. The fights went on. OpenAI finished converting into a public benefit corporation in October 2025. Musk sued over the change and lost in May 2026. In 2025 it reported revenue of about $13.1 billion and an operating loss of about $20.9 billion, and a spring 2026 round valued it at $852 billion. For a founder, the quiet thing is this. Altman had already lived through a board that wanted him out, and he knew what it was to be the face of a company whose results did not match its pitch. Whoever holds the final vote can use it, so know who that is before you need to. ## Palantir (USA) URL: https://startup.talk/usa/palantir/ Founders: Peter Thiel, Alex Karp, Joe Lonsdale, Stephen Cohen, Nathan Gettings Founded: 2003 (6 May 2003); Based in: Aventura (Miami area), Florida since Feb 2026; formerly Denver, Colorado (518 17th St.); Sector: Data analytics and AI software for government and enterprise Alex Karp has said he never learned to drive: he was too poor, and then he was too rich. The line sounds like a joke, but the first half was true for a long time. He was born in New York in 1967 and grew up in Philadelphia, the son of a Jewish pediatrician and an African American artist, both civil rights activists who took him to protests. He has said he had dyslexia from an early age, and that he felt different at every institution he met. He followed his father to Central High School, then took a philosophy degree at Haverford and a law degree at Stanford, where he met Peter Thiel. Law was not the plan. He wanted to be a social theorist. So he went to Frankfurt and wrote a PhD in German, though his adviser offered English, and Thiel alone backed the move. He worked as a research associate at the Sigmund Freud Institute and lived in Germany from the mid-1990s to 2001. What almost nobody knows is how he paid for that stretch. He used a $12,000 inheritance from his paternal grandfather to start the Caedmon Group, a money-management firm in London. His goal, he has said, was to save $250,000 so he could return to Berlin. His biographer Michael Steinberger questioned how $12,000 could build a firm that attracted wealthy investors. Then Thiel brought him in as chief executive in 2004. Palantir had started in May 2003 around Joe Lonsdale, Stephen Cohen and Nathan Gettings, who wanted to apply PayPal-style fraud detection to national security. Thiel named it after the seeing stones in Tolkien, and his fund covered most of the early cost, about $30 million. The CIA's venture arm, In-Q-Tel, added about $2 million. Venture capitalists mostly stayed away, and Palantir refused to give investors board seats. Lonsdale left in 2009. The price was twenty years without profit. Intelligence and military agencies adopted the software, and by 2013 it was valued at about $9 billion. It listed directly on the NYSE on 30 September 2020 and posted its first GAAP profit in early 2023. It has been criticised for work with US immigration enforcement and a 2023 NHS England data contract, and in 2026 France's domestic intelligence service said it would replace it. In 2025 revenue was $4.48 billion and net income $1.63 billion, and Karp's net worth passed $18 billion. The man who once wanted $250,000 to go back to Berlin now runs a company that is hard to ignore. A founder can take one thing from it: the goal at the start can be small and personal, and still lead somewhere large. ## Netflix (USA) URL: https://startup.talk/usa/netflix/ Founders: Reed Hastings, Marc Randolph Founded: 1997 (29 Aug 1997; site launched 14 Apr 1998); Based in: Los Gatos, California; Sector: Streaming entertainment Marc Randolph spent a long commute pitching business ideas to his boss, and most of them were bad. Personalized baseball bats. Shampoo delivered by mail. Reed Hastings, who ran Pure Atria, kept listening anyway. Randolph was born in Chappaqua, New York, the eldest child of an Austrian-born nuclear engineer turned financial adviser and a mother who ran her own real estate firm. He worked summers as one of the youngest instructors at the National Outdoor Leadership School, helped start the US edition of MacUser, and co-founded the mail-order firm MacWarehouse. In late 1996 Pure Atria bought the nine-person startup he had joined, and Hastings, a Boston-born maths graduate who had taught in a rural Swaziland school with the Peace Corps, kept him on as a vice president. Then came a test. They mailed a music CD inside a greeting card, and it arrived unbroken. Randolph has called that the aha moment. A disc could travel by post. What few people remember is who else paid for the early days. Randolph had to pitch his own mother. In accounts of his story she was the last investor, after he had raised about $100,000 from others, and she put in $25,000. Those figures come from secondary reports, and he has joked that she was delighted he was finally running a company she understood. Her money sat beside Hastings's own and that of Steve Kahn, founder of Integrity QA. Netflix launched in April 1998, charged per rental, and added a monthly plan in 1999. That year Randolph, who had named the company, gave the top job to Hastings. In September 2000 the two offered the business to Blockbuster for $50 million. Its chief executive said the dot-com hysteria was overblown. Amazon had once offered $14 to $16 million, which Hastings, who owned about 70 percent, refused. Randolph left in 2002, the year Netflix went public at $15 a share. He has said early-stage work is triage and that he is not good at the other ninety-seven things. Hastings stayed and, in September 2011, announced a separate DVD service called Qwikster. About 800,000 of 12 million subscribers cancelled before he reversed course. In 2025 Netflix reported revenue of about $45.2 billion and some 325 million paid memberships. The part worth carrying away is how small the first proof was. An envelope, a greeting card and a mother's cheque were enough to start something that Blockbuster would not buy for $50 million. ## Palo Alto Networks (USA) URL: https://startup.talk/usa/palo-alto-networks/ Founders: Nir Zuk, Rajiv Batra Founded: 2005 (1 Mar 2005); Based in: Santa Clara, California; Sector: Cybersecurity (firewalls, cloud and AI security) Nir Zuk waited a year for an answer that never came. Inside Juniper Networks he had proposed building a better firewall and asked for three things: ownership of the project, a $10 million budget and 25 handpicked engineers. Nothing. So he gave two weeks' notice. He was not new to being ignored. Zuk grew up in Rehovot, Israel, where his parents bought him a Dragon 64 computer. He insisted on the 30-megabyte hard drive instead of the 20. He taught himself to program, wrote commercial software as a teenager, and wrote computer viruses. What he rarely leads with is that he has no college degree. He started university mathematics during five years in an army unit that wanted whiz kids who could program, and he never finished. He told an interviewer in 2010 that he was one of the first virus developers in the world. In 1994 he joined Check Point, helped build stateful inspection, and in January 1997 moved to the United States for it. Then the company decided not to release Floodgate, a product his American team had built. He has said Israeli engineers were angry. He also said he likes to develop new technology, and bureaucracy made that hard. He left in March 1999. He founded OneSecure that month. After the dot-com crash he turned from services to a product he calls the first intrusion prevention system. NetScreen bought it in mid-2002, then Juniper bought NetScreen and planned to retire its products. That was the setting for his unanswered request. Palo Alto Networks was incorporated on 1 March 2005 with $9.4 million. Zuk says the hardest part of a struggling company is what happens to all the people you hired, and that a company dies when it slides from a can-do to a cannot-do attitude. It listed in July 2012, raising $260 million, then bought its way into a platform, including Expanse for $1.25 billion in 2020. In the year to July 2025 it earned $1.13 billion on $9.22 billion of revenue, and its market value was about $258 billion on 6 November 2026. Zuk retired as chief technology officer in August 2025. The quiet lesson sits in that unanswered request. He did not fight for a yes. He took the silence as the answer and left, and the firewall nobody would build became the whole company. ## CrowdStrike (USA) URL: https://startup.talk/usa/crowdstrike/ Founders: George Kurtz, Dmitri Alperovitch, Gregg Marston Founded: 2011; Based in: Austin, Texas (moved from Sunnyvale, California in Dec 2021); Sector: Cybersecurity (cloud endpoint protection, threat intelligence) At 13, Dmitri Alperovitch left Moscow. It was 1994, and his father, a nuclear physicist, had been granted a visa to Canada. About a year later the family landed in Chattanooga, Tennessee, where his father took a job at the Tennessee Valley Authority. A boy from Moscow was now in a Tennessee high school. There, still a student, he and his father started an encryption business together. It is a small fact, tucked into magazine profiles, and it says security began for him as a family trade. He went on to Georgia Tech, took a computer science degree in 2001 and a master's in information security in 2003, and at the email-security firm CipherTrust helped invent TrustedSource, a way of scoring senders' reputations. Then he ran threat research at McAfee. George Kurtz came from another direction. He grew up in Parsippany, New Jersey, says he started writing games on a Commodore in fourth grade, and studied accounting at Seton Hall. He began as a CPA at Price Waterhouse, until in 1993 the firm opened a security division and he moved into penetration testing, being paid to break into clients' systems. In 1999 he started Foundstone and co-wrote Hacking Exposed. McAfee bought Foundstone in 2004, and when Intel bought McAfee in 2011, he left. At Warburg Pincus he began on a cloud-based way to protect machines. Warburg put in $25 million. Alperovitch and the finance chief Gregg Marston joined him, and Falcon launched in June 2013. CrowdStrike's reports then helped bring US charges against five Chinese military hackers in 2014 and linked North Korea to the Sony hack. A December 2016 report on a Ukrainian artillery app was disputed, the company's first public bruise. The company went public on Nasdaq on 12 June 2019, priced at $34 against an expected $28 to $30, and opened at $63.50. Alperovitch left in 2020 and Kurtz stayed, and the company joined the S&P 500 in 2024. On 19 July 2024 a faulty update crashed about 8.5 million Windows computers. Airlines grounded planes. Delta later sued for roughly half a billion dollars. Even so, revenue reached about $4.81 billion in the year ended January 2026. The lesson sits in that July afternoon. Trust in security is built slowly, and a company can spend it in a day. ## Arista Networks (USA) URL: https://startup.talk/usa/arista-networks/ Founders: Andy Bechtolsheim, David Cheriton, Ken Duda, Jayshree Ullal Founded: 2004 (Arastra; renamed/incorporated as Arista in 2008); Based in: Santa Clara, California; Sector: Cloud and AI data-centre networking Before he was a billionaire's partner, David Cheriton was a teenager in Edmonton who had applied to the University of Alberta twice over: once for mathematics, once for music. Mathematics said yes. The music programme said no. Born in Vancouver in 1951, he went on to a mathematics degree at UBC in 1973 and a PhD in computer science at Waterloo in 1978, and became a Stanford professor. It is a small rejection. It matters because of where he ended up. In 1998 Cheriton and Andy Bechtolsheim were both early believers in a Stanford search project that did not yet legally exist as a company. Bechtolsheim, a German-born engineer who at sixteen had programmed an industrial controller in binary because he had no assembler, wrote the first cheque. Their pages disagree on whether Cheriton put in $100,000 or $200,000. Either way, Google made them rich. Cheriton has since been named on a list of frugal billionaires, known for avoiding big houses and expensive cars. Bechtolsheim had a second reputation: he built a Sun workstation, became Sun's first employee in 1982, then founded Granite Systems, a gigabit Ethernet start-up that Cisco bought for $220 million in 1996. He owned 60 percent. So in 2004 or 2005 the two men started Arastra with Ken Duda, who had written the switch kernel for Cisco's gigabit unit. The idea was a data centre switch that ran software on an unmodified Linux kernel. Big cloud customers wanted something they could program, not a sealed box. In October 2008 they brought in Jayshree Ullal as chief executive. She was born in London in 1961, raised in New Delhi by a physicist father, trained as an electrical engineer at San Francisco State, and had spent about 15 years at Cisco. Success drew a fight. In December 2014, six months after Arista listed on the New York Stock Exchange, Cisco filed two patent lawsuits. An import ban was upheld before it was reversed in 2016 after product changes and after two Cisco patents were overturned; a jury also rejected Cisco's damages claim that December. Then, in August 2018, Arista agreed to pay Cisco $400 million, ending the claims and Arista's own antitrust suit. For four years a young company had spent its energy in court against the firm its founders and chief executive came from. Arista kept buying capability, from Metamako to Big Switch, Pluribus and Broadcom's VeloCloud business. It reported revenue of about $9.0 billion for 2025, and on 6 November 2026 its market value was about $201 billion. The pattern is easy to miss. None of the founders was a stranger to rejection or to Cisco. They simply kept finding the next door, and walked through it with a better product. ## Databricks (USA) URL: https://startup.talk/usa/databricks/ Founders: Ali Ghodsi, Matei Zaharia, Ion Stoica, Andy Konwinski, Arsalan Tavakoli-Shiraji, Patrick Wendell, Reynold Xin Founded: 2013; Based in: San Francisco, California; Sector: Data and AI platform (lakehouse, Apache Spark) Ali Ghodsi was born in Tehran in December 1978, and by the time he was a small boy his family was leaving. It was 1983, the Iran-Iraq war was escalating, and they settled in Stockholm. His parents were physicians. In Sweden their qualifications did not simply carry over; they had to go through recertification and rebuild their careers. Ghodsi grew up feeling like an outsider. That is the detail that rarely appears in the Databricks story. The company is usually told as a tale of seven academics and a free piece of software. Before any of that, one of them had watched his parents start over. He took his master's and an MBA at Mid-Sweden University in 2002 and a PhD at KTH in Stockholm in 2006. A paper from his thesis, Dominant Resource Fairness, became foundational for Apache Mesos. He co-founded a peer-to-peer company, Peerialism, in Stockholm and worked as an assistant professor at KTH. He later found Swedish academia slow and the atmosphere suffocating. In 2009 he took a visiting scholar post at UC Berkeley's AMPLab. There he met Matei Zaharia, a Romanian-Canadian student who had built Apache Spark that year as a faster alternative to MapReduce, and worked alongside professors Ion Stoica and Scott Shenker. Spark spread because it was free. Free software does not pay salaries, so in 2013 seven people from the project, Ghodsi, Zaharia, Stoica, Andy Konwinski, Arsalan Tavakoli-Shiraji, Patrick Wendell and Reynold Xin, started Databricks. Andreessen Horowitz led a $13.9 million Series A. Stoica, who had arrived in the US from Romania in 1994, was the first CEO. Ghodsi, then running engineering and product, took over in January 2016. The public record on those early years is thin on struggle. What it shows is the order of moves: be available on every large cloud, including Azure from 2017; keep formats open; then build the lakehouse. The money grew to match, with a $62 billion valuation in December 2024. In February 2026 the company reported a $5.4 billion revenue run rate, and by August 2026 about $7 billion, with a Coatue-led round at $190 billion. It is still private. A founder who has been the outsider already knows how little a credential guarantees. Ghodsi's parents had degrees and had to prove themselves again. He built a company on technology the whole field could use for free, and trusted that being useful to everyone would come back to him. ## Salesforce (USA) URL: https://startup.talk/usa/salesforce/ Founders: Marc Benioff, Parker Harris, Dave Moellenhoff, Frank Dominguez Founded: 1999 (founded 8 March 1999); Based in: San Francisco, California; Sector: Cloud software (CRM, enterprise AI) As a teenager, Marc Benioff cleaned a jewelry store, and he did not enjoy it. Across the street was a Radio Shack with a computer in it. He taught himself to write a program there called How to Juggle, and he sold it for $75. That was the first money he made from software. At 15 he founded Liberty Software to make games for the Atari 8-bit, including King Arthur's Heir. By 16 the royalties were reportedly $1,500 a month, enough to help pay for college. Before that he had repaired antennas and CB radios for neighbours. He was born in San Francisco in 1964, grew up in Hillsborough, and studied business at the University of Southern California, graduating in 1986. He took a summer internship at Apple after talking to Guy Kawasaki, then joined Oracle. He was Rookie of the Year at 23 and, about three years later, its youngest vice president, earning more than $1 million a year. Then, at 31, he took six months off. He travelled to India and Hawaii, and the trip pushed him toward yoga and Eastern religions and toward the idea that software should be rented over the internet. Larry Ellison, his boss, let him take the break. On 8 March 1999 he founded Salesforce with three developers, Parker Harris, Dave Moellenhoff and Frank Dominguez. Ellison and CNET founder Halsey Minor were early investors. The timing was bad. The dot-com crash arrived and Salesforce laid off 20 percent of its staff in the early 2000s. Benioff hired actors with signs saying the end of software to heckle Siebel Systems at its own conference. The company listed in June 2004, raising $110 million. Revenue passed $1 billion in 2009, and by 2022 Salesforce had overtaken SAP as the largest enterprise applications company by revenue. The price of size came later. In January 2023 about 8,000 jobs, roughly 10 percent, were cut, with Benioff blaming pandemic over-hiring. He later called it a bad idea. For the year ended 31 January 2026 revenue was $41.5 billion. What stays with you is the order. A kid who disliked cleaning a shop found a computer across the road. Salesforce began as that same move, grown up: go where the machine is. ## Intuitive Surgical (USA) URL: https://startup.talk/usa/intuitive-surgical/ Founders: Frederic Moll, John Freund, Robert Younge Founded: 1995; Based in: Sunnyvale, California; Sector: Robotic surgery (da Vinci system), medical devices A young surgeon in residency at Virginia Mason in the early 1980s watched how large a cut a patient needed just to let him in. Frederic Moll said later he was struck by the size of the incision and the injury done merely to get inside the body. It bothered him enough to change his career. He grew up in Seattle, the child of two pediatricians, and went to Lakeside School, where Bill Gates and Paul Allen were schoolmates. He studied at Berkeley and the University of Washington, then left residency to develop the safety trocar, a device that helped make laparoscopic surgery possible. By 1994 he was at Guidant, and he had found a prototype at SRI International. Nobody else wanted it. In the early 1990s SRI had shown the system to many venture capitalists and, as one company history puts it, no one expressed serious interest until Moll. Moll himself first saw little commercial value in it, though he thought it had wow factor. What almost nobody remembers is that its first purpose was not the one he wanted. The system was meant for remote surgery on wounded soldiers. Moll saw a different use, and Guidant, his employer, would not back it. So in 1995 he left. Mayfield Fund, which had backed him before, paired him with John Freund, a Harvard-trained executive who negotiated the SRI licence, and with Robert Younge, an engineer. They incorporated in November 1995 and raised about $5 million from Mayfield and Sierra Ventures. Then reality. Early sales were slow. The system was expensive and its tools could be used only a few times, costing about $1,500 an operation. Gallbladder surgery, the first target, did not pan out because cheaper laparoscopic tools existed. In 1999 the company sold ten systems in Europe. In 2000, a month before its IPO, rival Computer Motion sued for patent infringement, which made customers nervous. The IPO raised $46 million anyway, and the FDA cleared da Vinci for general laparoscopic surgery that year. The two companies merged in 2003. Success has had a price. A da Vinci cost about $1.5 million, and Intuitive paid $43 million in 2018 to settle shareholder suits over unreported injuries. Moll left in 2002 to found Hansen Medical. For 2025 Intuitive reported revenue of about $10.1 billion. The close is a small one. Moll did not invent the machine. He kept looking at a wound until he could see what the machine was for. ## Stripe (USA) URL: https://startup.talk/usa/stripe/ Founders: Patrick Collison, John Collison Founded: 2010 (founded in Palo Alto); Based in: South San Francisco, California (dual HQ with Dublin, Ireland; Dublin HQ opened Oct 2025); Sector: Payments infrastructure, fintech The deal closed on Good Friday 2008, and one of the two brothers who had just sold their company was still seventeen with an exam ahead of him. John Collison had not yet sat his Leaving Certificate. His brother Patrick, nineteen, was about to move to Vancouver as a director of engineering. The buyer was a Canadian firm, Live Current Media, and the price reported by the Irish broadcaster RTE was more than three million euros. It had been a year of refusals and leaps. In early 2007 the brothers had started a company in Limerick that made software for heavy eBay users. They called it Shuppa. They asked Enterprise Ireland, a state agency, for money. It said no. Y Combinator showed interest instead, and that summer they moved to California. On their American investors' advice they merged with a similar UK start-up, built by the Oxford graduates Harjeet and Kulveer Taggar, and the result was Auctomatic. Its whole life lasted about ten months. Patrick had grown up in Dromineer, County Tipperary, with a microbiologist mother and an electronic engineer father. He took his first computer course at eight, was programming at ten, and at sixteen won Ireland's BT Young Scientist exhibition with a language called Croma. He started at MIT and left in 2009. John scored eight A1s in the Leaving Cert, started at Harvard in September 2009 and left within a year. Then came Stripe. It began in 2010 in Palo Alto with Patrick complaining, as one profile tells it, about how hard it was to take payments on the web. They processed their first transaction within two weeks of starting to build their own system. More than a thousand developers were on a waiting list before the public launch in September 2011. In May 2011 they raised $2 million from Peter Thiel, Elon Musk, Sequoia, SV Angel and Andreessen Horowitz. In 2016 a round valued Stripe at $9.2 billion, and by March 2021 at $95 billion. Then the market turned. The valuation was cut to $50 billion in the March 2023 round. In November 2022 Stripe laid off 14 percent of its staff, and in January 2025 about 300 more, with criticism of a cartoon sent to those people. A tender offer in February 2026 valued the company at $159 billion, and it said it handled $1.9 trillion in payments in 2025. What the brothers show is that a no from the wrong door is only information. In 2007 a funding agency passed on them. A year later they had sold their first company and were already planning the next one. ## Uber (USA) URL: https://startup.talk/usa/uber/ Founders: Garrett Camp, Travis Kalanick Founded: 2009 (founded March 2009 as Ubercab; renamed Uber in 2011); Based in: San Francisco, California (Mission Bay); Sector: Ride-hailing, delivery, mobility In the autumn of 2001, Travis Kalanick moved back into his parents' house in Northridge. He was 25. His company, Red Swoosh, had just used about $110,000 of its employees' withheld payroll taxes to keep the lights on, his co-founder had walked out, and some staff had gone months without a paycheck. Living at home, he later complained, "sucked". He had grown up in that same neighbourhood. His father, Donald, was a civil engineer for the city of Los Angeles; his mother, Bonnie, sold retail advertising for the Los Angeles Daily News. As a teenager he sold Cutco knives door-to-door, and at 18 he started a test-prep company. In 1998 he dropped out of UCLA to join Scour, a file-sharing service. In 2000 the film, music and publishing trade groups sued Scour for $250 billion, and it filed for bankruptcy that September. Red Swoosh was what he called his revenge business. What almost nobody remembers is how close it came to a darker ending. Advisers warned that using the payroll money could amount to tax fraud; accounts differ on who made the call. A second funding round repaid the IRS, and no one was prosecuted. He apparently went more than three years without a salary. By 2002 only he and one engineer remained. Mark Cuban put in $1.8 million in 2005, and Akamai bought the company in 2007 for about $19 million, leaving Kalanick roughly $2 million after tax. Uber came from someone else's itch. Garrett Camp, the StumbleUpon founder, could not find a taxi on a snowy night in Paris, after he and Kalanick had already spent $800 on a private driver on New Year's Eve. Ubercab was founded in March 2009, with a prototype built by Camp, Oscar Salazar and Conrad Whelan. Ryan Graves became first employee in February 2010 and CEO that May; Kalanick took over in December 2010. San Francisco taxi operators objected to the name, and in 2011 it became Uber. The price arrived in 2017. On 19 February a former engineer, Susan Fowler, published an account of being propositioned by a manager and threatened when she reported it. More than 20 employees were fired. Investors led by Benchmark demanded his exit. At the end of May his mother was killed in a boating accident that seriously injured his father; on 20 June he resigned as CEO. Dara Khosrowshahi replaced him that August. Kalanick later sold about 90 percent of his shares, around $2.5 billion. Uber listed in May 2019. In 2025 it reported $52 billion in revenue, and its market value was about $140 billion on 6 November 2026. The habits that carry a founder through a near-death year are not always the habits that run a company of thousands, and the first set can be hard to put down. ## Marvell Technology (USA) URL: https://startup.talk/usa/marvell-technology/ Founders: Sehat Sutardja, Weili Dai, Pantas Sutardja Founded: 1995; Based in: Santa Clara, California; Sector: Semiconductors (data infrastructure, networking, storage) In 1979, at seventeen, Weili Dai walked into Abraham Lincoln High School in San Francisco knowing very little English. She had left Shanghai, where she played badminton and junior semi-professional basketball, a sport she later credited with building her confidence and her drive to win. Her father was an engineer, her mother a nurse. After one year at Lincoln she went on to UC Berkeley and studied computer science. At Berkeley she met Sehat Sutardja, born in Jakarta in 1961, a certified radio repair technician by thirteen, in the United States since 1980 and then a doctoral student. The couple remember the meeting as one in a campus elevator. They married in 1985. He worked at Micro Linear and Integrated Information Technology; she worked as a software engineer. What almost nobody pictures when they hear the name Marvell is a kitchen table. Dai has said the company was planned around hers in 1995, with Sehat and his brother Pantas. There was no venture round. The early money came from friends and family, plus $200,000 the founders earned by licensing a chip design. Only two years later did Diosdado Banatao put in $1 million. Sehat took the technology; Weili took products, markets and the running of the company. They began with a CMOS read channel for disk drives, a bet that cheaper chips could replace costly ones, and Seagate became the first customer. On 27 June 2000 Marvell listed on Nasdaq, selling 6 million shares at $15 and raising $90 million. Then the table tilted. In 2006 the SEC opened an inquiry into backdated stock options. The company paid a $10 million fine in 2008, and Dai paid SEC fines and gave up her post as chief operating officer and her seat on the board. In 2012 a jury awarded Carnegie Mellon $1.17 billion over disk-drive patents; Marvell settled in 2016 for $750 million. The end came that April. Activist investor Starboard Value had built a stake of about 7 percent, and the company's own investigation found no fraud, though it found pressure to hit revenue targets and some revenue booked too early. Sehat and Weili were out. Matt Murphy became CEO in July. The couple moved to Las Vegas. In 2021 Sehat co-founded Silicon Box, a chip-packaging company in Singapore, with Dai and Byung Joon Han. He died on 18 September 2024, at 63. Marvell went on without them: about $8.2 billion in revenue in fiscal 2026, and a market value near $120 billion on 6 November 2026. A company can begin at a kitchen table with a licensing cheque and a family's savings, and still end up decided in a boardroom the founders no longer control. ## AppLovin (USA) URL: https://startup.talk/usa/applovin/ Founders: Adam Foroughi, John Krystynak, Andrew Karam Founded: 2012; Based in: Palo Alto, California (Stanford Research Park); Sector: Mobile advertising technology, app monetization, AI ad engine In the spring of 2012, Adam Foroughi pitched the top venture capitalists in Silicon Valley. He wanted $1 million for roughly a quarter of the company, a valuation of about $4 million. They all said no. He has since said he thinks they doubted advertising and mobile gaming more than they doubted him. He was already a veteran of failure and of success. Foroughi was born in Sardasht, Iran, in 1980 and grew up in Laguna Niguel, California. He studied economics at Berkeley, worked as a derivatives trader, then co-founded two advertising businesses, one in 2005 and one in 2008, and sold both around 2010. Then he moved to Palo Alto and tried to build apps for consumers. In 2011 he built a dating app and a fashion app. Both failed. He has called them two terrible ideas. What few people know is where the hunger comes from. Foroughi was four when his family was uprooted out of Iran. His father, he has said, had built one of the most successful real estate development companies in the country, employing thousands of people. The family lost all of it when they left. He has said he saw a part of his father lost, and that he has felt an obligation ever since to prove he can succeed. That feeling, he says, never went away. He looks for people with a chip on their shoulder, and counts himself among them. So when the venture firms passed, he did not stop. With John Krystynak and Andrew Karam he kept AppLovin going for about eighteen months as a bootstrapped company, with $4 million from angel investors, out of sight until 2014. Early customers included OpenTable and Spotify, and by 2016 the company had reached number ten on Deloitte's Fast 500. The next scare came in 2016, when Orient Hontai Capital agreed to pay $1.42 billion for a majority stake. The deal needed a review in Washington, and the review took over a year. Foroughi has said he did not want to be made an example of if it blew up. It was abandoned. In 2018 KKR put in $400 million for a minority stake. On 15 April 2021 AppLovin listed on Nasdaq at $70 a share, valuing it at about $24 billion. Then the stock fell about 92 percent, to a market value he recalls at around $3.8 billion, and he has described feeling that the whole world thought the business was bad. It recovered. In 2025 revenue was $5.48 billion and net income $3.33 billion, and it joined the S&P 500 in September. A short seller attacked in February 2025, and by 6 November 2026 the market value was near $120 billion. Foroughi has said he still wakes up anxious and checks that the business is healthy. The boy who watched his father lose a company is now the man who checks his own. ## ServiceNow (USA) URL: https://startup.talk/usa/servicenow/ Founders: Fred Luddy Founded: 2004 (founded as Glidesoft in 2003; incorporated in California in 2004); Based in: Santa Clara, California; Sector: Enterprise workflow software, IT service management, AI At sixteen, Fred Luddy took his parents' station wagon and drove to California. A profile of his life reports he had no plan. He camped along the way, picked strawberries and washed cars for cash. He grew up in New Castle, Indiana, the son of an accountant and a Catholic schoolteacher. He got his GED, then, as a teenager, took a job at a folding-door factory in town and taught himself Fortran overnight. In 1973 he wrote programs for the economics dean at Indiana University for $3.50 an hour, often working fifty or sixty hours a week. He dropped out after his freshman year and did not go back. What he built next was a career in other people's companies. Peregrine Systems hired him as chief technology officer in 1990. He stayed thirteen years, helped build its AssetCenter product and watched the company buy Remedy in a deal worth about $1 billion in stock. His own stake peaked near $35 million. Then Peregrine's executives were found to have inflated revenue and invented deals. The company collapsed in 2002, and Luddy, then 49, was in effect broke. He had sworn, the profile says, never to touch IT service management again. The promise lasted about nine months. Two weeks before his fiftieth birthday he incorporated Glidesoft in California. He has said he felt he could not wait any longer. For a long time he was the only person on the payroll, and at first he gave the software away to gather feedback. In mid-2005 JMI Equity put in $2.5 million, enough to hire five people. In 2006 it became Service-now.com. By 2007 it reported about $13 million in revenue and said it was cash-flow positive. The handoff came in 2011, when Frank Slootman took over as CEO. Luddy has said he did not have the skills to be CEO. Late that year VMware offered $2.5 billion. Luddy, by his own later account, was inclined to take it, because it was a chance to not be bankrupt. The board, pushed by investor Doug Leone, said no. The company listed in June 2012 with a $210 million IPO. In 2025 ServiceNow reported $13.28 billion in revenue and $1.748 billion in net income, and its market value was $106.24 billion on 5 November 2026. Luddy later gave $60 million to Indiana University, the school he left after a year. It now carries his name. A man who once ran away to California in a borrowed car, and went broke at forty-nine, ended up with his name on the building of the school he quit. ## Airbnb (USA) URL: https://startup.talk/usa/airbnb/ Founders: Brian Chesky, Joe Gebbia, Nathan Blecharczyk Founded: 2008 (incorporated 27 June 2008; site launched 11 Aug 2008; idea began Oct 2007); Based in: San Francisco, California; Sector: Online travel and short-term home rental marketplace When Brian Chesky arrived in San Francisco in October 2007, his friend Joe Gebbia told him the rent was $1,150. It was more than he had expected. The idea they would soon have for paying it began with three designers named Michael, Kat and Amol, a sold-out design conference, and an air mattress on the living-room floor. Chesky, born in Niskayuna, New York, to two social workers, had graduated in industrial design from the Rhode Island School of Design in 2004 and taken a job at a Los Angeles design firm, 3DID. Gebbia, from Georgia, studied graphic and industrial design, worked at Chronicle Books and started a green-design site, Ecolect. They brought in Nathan Blecharczyk, a Harvard computer science graduate who had run a web-hosting business as a teenager, as CTO in February 2008. The company was incorporated that June as AirBed & Breakfast, and the site went live in August. Chesky expected it to take off. It did not. As he later put it, they got two customers, and he was one of them. What almost nobody remembers is what his own mother said. When he told her about it, she offered to send money, and called it crazy to have strangers staying in your home. He has recalled her reaction in interviews. She was not alone. He sought $150,000 for 10 percent of the company and was turned down by many investors, and he has said the founders probably did not look very impressive. Their credit-card debt, by various accounts, ran to $20,000 to $25,000 each. In 2008 he and Gebbia nearly stopped. He has described going to bed with his heart pounding, and called it probably rock bottom. So they did something odd. During the 2008 presidential race they redrew cereal boxes as Obama O's and Cap'n McCain's and sold them at the Democratic convention. Estimates of the take run from $20,000 to $30,000. It was enough to get Paul Graham's attention, and in January 2009 Y Combinator took the three in for $20,000 and 6 percent. By March the site had 10,000 users and 2,500 listings, and the name was shortened to Airbnb. In April, Sequoia put in $600,000. The company grew to a $10 billion valuation by 2014. In April 2020 travel collapsed and Silver Lake and Sixth Street put in a combined $2 billion. Airbnb listed on Nasdaq on 10 December 2020 at $68 a share and set aside $238 million of shares for its hosts. In 2025 it reported $12.2 billion in revenue and $2.51 billion in net income, and on 6 November 2026 it was worth about $102 billion. The people closest to him said it was crazy, and for a while the numbers agreed. He kept the rent paid and the site live until a stranger said yes. ## Equinix (USA) URL: https://startup.talk/usa/equinix/ Founders: Jay Adelson, Albert (Al) Avery Founded: 1998 (June 1998; company history page says 22 June 1998; Adelson's page says initially called Quark Communications); Based in: Redwood City, California; Sector: Data centres, colocation and interconnection (REIT) Before he built the buildings that now carry a large share of the internet, Jay Adelson worked his way in through a temp agency. A podcast page for his 2019 Something Ventured interview marks that as the start of his path. It fits a career that had no straight line. Adelson grew up in Southfield, Michigan, studied film and broadcasting at Boston University with some computer science alongside, and began working in post-production sound. In 1993 he moved to San Francisco and went into internet plumbing. He joined the early ISP Netcom as an installation coordinator and rose to director of network operations. He was running Netcom's network in 1995, when the hacker Kevin Mitnick was caught. By late 1996 he was at Digital Equipment Corporation, building and running the Palo Alto Internet Exchange, or PAIX. The idea was simple: carriers needed a neutral building where they could hand traffic to each other. Adelson saw that it could be copied. Every big city could have a telecom hotel. With Al Avery, another Digital facilities manager, he left in June 1998 and founded Equinix. Early on the company was reportedly called Quark Communications. He was the first chief technology officer, overseeing how the buildings were designed and built. Looking back on the early years in 2013, he said only that "it was a crazy time". The timing was unkind. Equinix listed in August 2000, just as telecom customers began to fail. How it got through is not well documented, but what followed is: in 2002 it merged with I-STT, a subsidiary of Singapore Technologies Telemedia, and with Pihana Pacific, which gave it a base in Asia-Pacific. Europe came with the 2007 purchase of IXEurope. When Steve Smith left Hewlett-Packard to run Equinix that year, it had roughly $400 million of annual revenue and a market value of about $6 billion. Adelson did not stay to see it. He stepped back to a strategic advisor role in 2005, ran Digg from February 2005 and left in April 2010 after disagreements with Kevin Rose and the board. Later he started a venture firm, Center Electric, named after a company his great-grandfather founded in Detroit in the 1940s. Equinix opened its 50th data centre in London in 2010, bought Switch and Data, and became a real estate investment trust in 2015. In 2016 it closed the roughly $3.8 billion TelecityGroup deal. In 2025 it reported $9.22 billion in revenue, and on 6 November 2026 its market value was around $99 billion. Adelson did not invent a new technology. He copied something that had worked once, in one building, and asked why it should not work everywhere. ## Moderna (USA) URL: https://startup.talk/usa/moderna/ Founders: Derrick Rossi, Noubar Afeyan, Robert Langer, Timothy Springer, Kenneth Chien Founded: 2010 (September 2010 as ModeRNA Therapeutics; Afeyan's page says 2009); Based in: Cambridge, Massachusetts; Sector: Biotechnology: mRNA medicines and vaccines One evening at home, after a lot of unsuccessful brainstorming, Derrick Rossi took the dash out of the name his lab used for its work. Mod-RNA, for modified messenger RNA, became Moderna by turning the dash into an e. He has told the story in an interview with the New York Stem Cell Foundation. Rossi was born in Toronto, the youngest of five in a Maltese immigrant family. His father worked in auto body shops and his mother co-owned a Maltese bakery. He grew up around animals, he has said, because an older brother kept bringing strange pets through the front door, snakes among them. He took a bachelor's and a master's in molecular genetics at the University of Toronto, and in high school a biology course had blown him away. What is rarely mentioned is that his path ran through a false start. After Toronto he began a graduate programme in Paris and left it early. He later finished a PhD at the University of Helsinki in 2003, then went to Stanford to work with Irving Weissman on stem cell ageing. By 2007 he was at Harvard Medical School, trying to turn cells into other cell types without inserting cancer-linked DNA. In 2008 his lab found that modified mRNA could do the job, and the 2010 paper was named one of Time's top ten medical discoveries of the year. The idea needed more than a lab. Noubar Afeyan, born in Beirut to an Armenian family whose grandfather survived the genocide, had left Lebanon in 1975 and gone on to McGill and MIT. He had founded his first company in 1987, days before Black Monday, and later Flagship Pioneering. He has said Armenians have a practiced ability to restart. In September 2010 Rossi, Afeyan, Robert Langer, Timothy Springer and Kenneth Chien founded ModeRNA Therapeutics. In 2011 Stephane Bancel became CEO. Rossi left the board in 2014. By then AstraZeneca had paid $240 million upfront for a partnership and DARPA had given $25 million. The company raised about $2 billion in venture money from 2011 to 2017 while STAT described a secretive culture. In 2017 Alexion ended its partnership after safety issues stalled work before human trials. On 7 December 2018 Moderna listed at $23 a share and raised $621 million, the biggest biotech IPO to that point. Then came COVID-19. Operation Warp Speed gave $955 million, and the vaccine was authorised in December 2020. Bancel's stake of about 9 percent passed $1 billion in April 2020. After the pandemic, demand fell: fiscal 2025 revenue was about $1.94 billion against a loss near $2.8 billion. In 2023 Moderna paid $400 million to settle a patent dispute with the NIH, Dartmouth and Scripps. Rossi has described the hard obstacles in moving a discovery to the clinic as a concrete wall that can take years to get through. Persistence can do it, he says, but you have to ask whether you have the time, the resources and the funding to keep going. ## Snowflake (USA) URL: https://startup.talk/usa/snowflake/ Founders: Benoit Dageville, Thierry Cruanes, Marcin Zukowski Founded: 2012 (23 July 2012, San Mateo, California); Based in: Menlo Park, California; Sector: Cloud data platform and AI data cloud Benoit Dageville bought a whiteboard from Amazon and put it in the living room of his apartment. His wife, he later joked to a New Zealand interviewer, was not happy about it. That whiteboard was the first office of Snowflake. Dageville and Thierry Cruanes were French database engineers who had met at Oracle, the company whose warehouse software defined the market. In 2012 they quit to build something Oracle's design could not do. A third engineer joined them: Marcin Zukowski, who had moved to Amsterdam for the last year of his studies at the Vrije Universiteit, earned a PhD at the Dutch research institute CWI, and co-founded the database company Vectorwise. Older warehouses tied storage and computing together, so customers paid for capacity they rarely used. The three wanted to pull the two apart and build for the cloud. Mike Speiser of Sutter Hill Ventures became the first chief executive. Snowflake was incorporated on 23 July 2012 in San Mateo, and Sutter Hill put in $5 million. What the later success hides is how close it came to never starting. Dageville has said the strength of his friendship with Cruanes was the main reason they nearly did not go ahead: they feared that arguing over a company could damage it. They went ahead anyway. He says the company has since made the friendship better. Handing over control was the second hard choice. After the first product in 2014, the founders brought in Bob Muglia, a longtime Microsoft executive, as chief executive so they could stay with the engine. In May 2019 Frank Slootman, who had run ServiceNow, took over. Slootman retired on 28 February 2024, and Sridhar Ramaswamy succeeded him. Money followed the product. A $263 million round in January 2018 valued the company at $1.5 billion, and a $450 million round led by Sequoia that October valued it at $3.5 billion. In September 2020 Snowflake went public, raising $3.4 billion, one of the largest software listings ever. Even that day was not simple joy. Cruanes described a sinking feeling at the valuation, and Dageville said he would have preferred a lower one. For the year ended 31 January 2026 the company reported revenue of about $4.7 billion and a net loss of about $1.33 billion, and its market value was around $95 billion on 6 November 2026. Both men still write code. The whiteboard story is a small one, but it holds the lesson: the partnership you protect at the start is the asset the company is built on. ## Fortinet (USA) URL: https://startup.talk/usa/fortinet/ Founders: Ken Xie, Michael Xie Founded: 2000 (as Appligation, Inc.; first product FortiGate released 2002); Based in: Sunnyvale, California; Sector: Cybersecurity: network firewalls, secure networking, security services Ken Xie was six foot five and wanted to play professional volleyball. That is not the origin story anyone expects from the founder of a cybersecurity giant. Born in Beijing in 1963, he told Forbes in 2017 that he had hoped for a sports career in China. His parents, both academics at Tsinghua University, had a different plan: a PhD at Stanford, then home to a professorship. He followed the first part. Xie studied electrical engineering at Tsinghua and went to Stanford for a master's. What happened next was not in the parents' plan. At Stanford he met students who wanted to start businesses, something he said he had never experienced growing up in China. While still a student he started Systems Integration Solutions in 1993 to help small businesses get online securely. Three years later he co-founded NetScreen with Yan Ke and Feng Deng, and built the first firewall and VPN box around a custom chip, an ASIC, in his Palo Alto garage. He became a US citizen in 1998. Juniper bought NetScreen in 2004 for $4 billion in stock. By then Xie had already left. In 2000 he quit and began again with his brother Michael, an engineer who had studied at the University of Manitoba. The company began as Appligation, then ApSecure, and finally Fortinet, from Fortified Networks. He took a second shot at the same idea: security built into the network hardware, not bolted onto it. The early years were lean. Fortinet raised $13 million of private money between 2000 and 2003, and its first product, FortiGate, shipped in 2002. Opponents arrived fast. Trend Micro filed a patent complaint in 2004, a German court issued an injunction in 2005 over open-source code, and a report that year alleged its products were used for internet censorship in Myanmar. The first profitable quarter came in the third quarter of 2008, and Fortinet went public on Nasdaq in November 2009. Reputation trouble followed. In 2019 a whistleblower case over mislabelled Chinese-made equipment sold to US government users was settled, and security flaws in its devices drew attention in 2021, 2025 and 2026. ProPublica reported in 2023 that Xie had claimed over $30 million in tax deductions by giving shares to a family foundation, which spent $3 million on a house he lived in. Bloomberg valued his 7.4 percent stake at about $1 billion in 2019. In 2025 Fortinet reported $6.8 billion in revenue, and on 6 November 2026 its market value was about $90 billion. Xie's parents wanted a scholar and got a founder. Sometimes the plan a family makes for you is the reason you leave to find your own. ## Cloudflare (USA) URL: https://startup.talk/usa/cloudflare/ Founders: Matthew Prince, Michelle Zatlyn, Lee Holloway Founded: 2009 (26 July 2009); Based in: San Francisco, California; Sector: Internet infrastructure, CDN, network security, edge computing About three months in, Lee Holloway told Matthew Prince that he understood why they needed Michelle Zatlyn, but he was not sure why they needed Prince. Prince repeated the line himself, on a stage at Startup School in October 2014, with Zatlyn beside him. Holloway was the engineer who built the first prototype. Prince was born in Salt Lake City in 1974 and raised in Park City. His father was a journalist and restaurateur; his mother owned gift stores and sneaked him into University of Utah computer science classes as a child. He had an Apple II Plus at seven. He studied at Trinity College, then law at the University of Chicago, and taught cyberlaw before running a small company, Unspam, that supported Project Honey Pot. In 2008 the Department of Homeland Security called to ask what its spam data was worth. Holloway, who studied computer science at UC Santa Cruz, was Unspam's chief software architect. Zatlyn, from Prince Albert, Saskatchewan, studied chemistry at McGill and was at Harvard Business School when the three entered a business plan prize and won $50,000. Holloway built the prototype; the other two wrote the plan. Cloudflare was founded on 26 July 2009. The early product was rough. Prince said the first release was one he was "literally embarrassed" by: its first ten test users all had their sites knocked offline. They had no money for servers, so Zatlyn emailed about 200 people near Palo Alto, and roughly 70 percent answered with a machine or a lead. She drove around collecting them, and none of them worked. They made servers out of the parts. Then came the cost. In 2012 a hacker group took over accounts belonging to Prince. In 2015 Holloway stepped down. The 2019 IPO filing explained why: he had frontotemporal dementia, a rare neurological disease. Prince and Zatlyn called him the genius who architected the platform, and the company named its IPO Project Holloway. Trusts linked to him held about 3.2 percent of the company. The wider company faced hard years too: the Cloudbleed bug in 2017, and fierce criticism for hosting sites such as The Daily Stormer before dropping them. Cloudflare listed on the New York Stock Exchange on 13 September 2019 at $15 a share. In 2025 it reported about $2.17 billion in revenue and a net loss of $102 million, and in November 2026 its market value was about $87 billion. Holloway's doubt about whether the company needed its CEO is the part worth keeping. The man who asked it built the platform everyone else came to depend on, and then had to leave it. ## Carvana (USA) URL: https://startup.talk/usa/carvana/ Founders: Ernest Garcia III, Ryan Keeton, Ben Huston Founded: 2012; Based in: Tempe, Arizona; Sector: Online used-car retail, auto finance, logistics When Ernest Garcia III was a small boy, his father stood in a courtroom and pleaded guilty to a felony. In October 1990 Ernest Garcia II, a Tucson real estate developer, admitted bank fraud for acting as a straw borrower in the collapse of Charles Keating's Lincoln Savings and Loan. He served three years of probation, and he and his firm filed for bankruptcy. The younger Garcia was born around 1982 and so would have been about eight. What his father did next is the story. In 1991 the elder Garcia bought a bankrupt rent-a-car franchise, Ugly Duckling, for under $1 million, merged it with his finance company and turned it into a used-car seller for buyers with weak credit. He took it public in 1996 and private again in 2002 as DriveTime. Garcia III studied management science and engineering at Stanford, graduating in 2005. He started at RBS Greenwich Capital and joined DriveTime in 2007. By then he was working inside a business his family had rebuilt out of a ruin. In 2012 he started Carvana inside it with Ryan Keeton and Ben Huston, and DriveTime paid for the first round. They opened a small operation on Marietta Street in Atlanta and a one-storey prototype in 2013. In November 2015 a coin-operated glass tower of cars opened in Nashville, turning an anxious purchase into something customers photographed. By August 2016 Carvana had raised about $460 million and operated in 15 markets. Then the family name was tested again, in public. On 28 April 2017 Carvana priced its IPO at $15 and raised about $225 million, then opened at $13.50 and closed at $11.10. Garcia and his father owned about 71 percent of the company. The worst came in 2022. Carvana bought the auction chain ADESA for $2.2 billion just as used-car demand cooled, cut about 12 percent of staff, then thousands more, and Bloomberg estimated Garcia's wealth fell 98 percent. For about eighteen months, CNBC reported, the company worked through its debt and people wondered whether it would survive. In January 2025 Hindenburg Research published a report on dealings between Carvana and DriveTime, which it said favoured Carvana. It survived. In 2025 Carvana reported about $20.3 billion in revenue and net income of about $1.41 billion, and its market value was about $80 billion in early November 2026. Garcia's story is not only that of a wealthy heir with a clever gimmick. It is a family that knew how fast a name can collapse, and built a company at the same edge. ## DoorDash (USA) URL: https://startup.talk/usa/doordash/ Founders: Tony Xu, Andy Fang, Stanley Tang, Evan Moore Founded: 2013 (PaloAltoDelivery.com launched January 2013; incorporated as DoorDash June 2013); Based in: San Francisco, California; Sector: Local delivery, food delivery, logistics For a Chinese family newly arrived in Champaign, Illinois, in 1989, a trip to McDonald's was a luxury. Tony Xu wrote that himself in 2017. He was five, and the family had come, in his words, with only a few hundred dollars. He was born Xu Xun in Nanjing. His father was a graduate student in aeronautical engineering and applied math at the University of Illinois. His mother, Julie Cao, had been a doctor in China, but the United States did not recognise her licence. Money was tight and the family relied on federal assistance, as CNBC reported from Xu's own account. His mother took jobs wherever she could, and Xu started a lawn-mowing business at nine. He washed dishes at the Chinese restaurant where his mother worked. He and his father chose the name Tony, from the TV show Who's the Boss?, at the immigration office. What most people hear about DoorDash is that it grew out of a Stanford class project. What is less told is the line Xu gave CNBC on 9 December 2020, the day the company listed: his mother worked three jobs a day for 12 years, one of them at a local Chinese restaurant, before she opened her own clinic. That is the wound under the company. His IPO letter said DoorDash exists to help people like his mother, who came here with a dream to make it on their own. Xu studied at Berkeley, worked at Square, RedLaser and McKinsey, then went to Stanford's business school. There he met Andy Fang and Stanley Tang, who had roomed together and built a social calendar app. In January 2013 they launched PaloAltoDelivery.com with Evan Moore. They often made the deliveries themselves. The first order, the AP reported, was pad thai with prawns and spring rolls. In summer 2013 Y Combinator gave them $120,000 for 7 percent. The price came later. A 2017 lawsuit from couriers claimed misclassification, and DoorDash agreed to a tentative $100 million settlement in 2022. A New York City pay settlement reached $131.5 million. In 2019 a data breach affected 4.9 million people, and in 2022 the company cut about 1,250 corporate jobs. On 9 December 2020 DoorDash sold shares at $102 and raised $3.37 billion. Xu recalled to Bloomberg TV delivering hummus out of his Honda. In 2025 the company reported $13.7 billion in revenue and $935 million in net income, and its market value was about $78 billion in November 2026. His parents came to this country, Xu said, to make a better life for him. A founder can carry that debt quietly, and decide what to build with it. ## Robinhood (USA) URL: https://startup.talk/usa/robinhood/ Founders: Vlad Tenev, Baiju Bhatt Founded: 2013 (April 2013); Based in: Menlo Park, California; Sector: Online brokerage, retail investing, crypto, fintech Vlad Tenev has said his parents' status as visa holders always seemed fragile. He was an only child, and by his own account he felt the pressure to make their sacrifice count. He was born in Varna, Bulgaria, in 1987, under a communist government. His father, an economics professor, went to study at the University of Delaware when Vlad was three. His mother followed a year later, and about six months after that, five-year-old Vlad flew to New York to join them. They lived in Newark, Delaware, and then in student housing in College Park, Maryland, while his father worked on a PhD. Money was scarce, with few books or toys. Tenev went with his father to the University of Maryland computer lab and taught himself to use the machines. At seven he was given Stephen Hawking's A Brief History of Time. He went on to Thomas Jefferson High School for Science and Technology, then Stanford, where he met Baiju Bhatt, who grew up in Poquoson, Virginia, the son of Gujarati immigrants; his father worked at NASA's Langley Research Center. Tenev did a master's in mathematics at UCLA and left a PhD plan to work with Bhatt. They built trading software for Wall Street first, Celeris in 2010 and Chronos Research after it. The 2008 crisis and the Occupy protests pushed them the other way, toward a broker for everyone else. Tenev's wife suggested the name. Reportedly they pitched about 75 investors, and to win Tim Draper's backing they agreed to take no salaries until the broker licence was approved. Index Ventures led a $3 million seed in December 2013. Rounds followed, up to a $6 billion valuation in May 2018 that made both founders billionaires. Then the cost arrived. In June 2020 a 20-year-old student, Alex Kearns, died by suicide after reportedly believing he owed a huge negative balance; his parents later sued. On 28 January 2021 Robinhood restricted GameStop buying after its clearinghouse demanded collateral rising from $700 million to $3.7 billion, and Tenev was in front of a House committee on 18 February. The IPO came on 29 July 2021 at $38 a share, raising about $2.1 billion. Bhatt left his executive role in March 2024, and Robinhood joined the S&P 500 in September 2025. It reported net income of about $1.88 billion on $4.47 billion revenue in 2025, and a market value of about $70.6 billion on 5 November 2026. A boy who grew up knowing his family's place here was fragile built a company on the promise that everyone gets access. When that promise was tested, the people on the other side of the screen were the ones who paid for it. ## Datadog (USA) URL: https://startup.talk/usa/datadog/ Founders: Olivier Pomel, Alexis Le-Quoc Founded: 2010; Based in: New York City, New York; Sector: Cloud monitoring, observability, security software The company's name came from a database everyone feared. At Wireless Generation, the New York education-software firm where both founders worked, there was an Oracle system the team nicknamed Datadog17. Olivier Pomel told RTP Global that when it came time to name their own company, they dropped the 17. Pomel grew up in France and studied computer science at the Ecole Centrale Paris, from 1996 to 1999. What fewer people know is that he is one of the original authors of VLC, the video player that millions of people still use, a line Datadog puts on his own official biography. He went on to work at IBM Research, where by one account he and Alexis Le-Quoc worked together in the mid-2000s, and then at Wireless Generation, where Pomel ran technology and Le-Quoc ran operations. They worked there together for about nine years. The idea came from friction they had lived with: developers wrote code, system administrators ran it, and when something broke, the two groups blamed each other. Pomel said the frustration of switching between monitoring tools was the origin of Datadog. In 2010 they incorporated in New York City. For the first six months they wrote no code. They talked to potential customers, because, Pomel has said, the biggest fear was building the wrong thing. Then came the hard part. Much of the West Coast smart money passed. Pomel remembers kind rejections such as too early for us, or great idea, wrong time. One VC dismissed monitoring as a crowded market in his first pitch. Building infrastructure from New York was seen as a weakness. They stayed in Airbnbs and watched unit economics. A seed round came from NYC Seed, Contour Venture Partners and IA Ventures, then $6.2 million in 2012, led jointly by Index Ventures and RTP. Pomel had no MBA and no prior experience as a chief executive. For a long time Datadog sold only month-to-month, so bad news surfaced fast. The first head of product also ran sales, and the first dedicated salesperson came only after the business had modest recurring revenue. Buyers noticed. Before the IPO, Cisco reportedly offered more than $7 billion. Datadog said no, and on 19 September 2019 it sold shares at $27 on Nasdaq. The stock jumped about 37 percent that day. In 2025 Datadog reported revenue of $3.43 billion and net income of $108 million, and its market value was about $62.7 billion on 5 November 2026. Pomel calls an IPO a starting line, not a finish line. A founder who has suffered a problem for nine years inside someone else's company can survive a lot of polite no's. ## Anduril Industries (USA) URL: https://startup.talk/usa/anduril-industries/ Founders: Palmer Luckey, Trae Stephens, Brian Schimpf, Matt Grimm, Joe Chen Founded: 2017; Based in: Costa Mesa, California; Sector: Defense technology, autonomous systems, AI software (Lattice) Palmer Luckey was on a trip to Disney World with his girlfriend when Facebook fired him, in early 2017. He has said he was afraid of becoming another Eduardo, the co-founder who got pushed out of the story. Months earlier a news story had tied him to a $10,000 donation to a pro-Trump group, and the company had told him to stay off social media and out of the office. Later reporting found no evidence the group produced hateful content. Facebook says the exit had nothing to do with politics. Luckey says it did. He was twenty-four. Three years before, Facebook had paid about $2 billion for Oculus, and he had used part of his share to buy his parents a house. That detail lands harder once you know where he started. He was born in Long Beach in 1992, the eldest of four, to a car salesman and a mother who homeschooled all of them. The family lived on the ground floor of a small multifamily home. He was taking college classes by fifteen, teaching himself electronics, and funding his hardware habit by scrubbing decks and repairing engines at the Long Beach Sailing Center. What almost nobody knows is where he lived while he built the first Rift. His parents had pushed him out of the house, and he moved into a 19-foot camper trailer in their driveway. He tore out the bathroom and the kitchen to make room for gear. He used a public restroom near a laundromat and ate frozen burritos from Jack in the Box. In 2012 his parents sold the trailer and told him he could sleep in the garage. That same year Sony offered him $70,000 a year to work at its Santa Monica lab. He said no, and the Oculus Kickstarter raised more than $2.4 million. So the man Facebook let go had already lived with almost nothing, and had already turned down a salary to bet on himself. Anduril came out of that. He met Trae Stephens, a former Palantir employee at Founders Fund, in 2014, and in 2017 they started the company in Irvine with Palantir and Oculus alumni, naming it after Aragorn's sword. Their first pitch was cheap border surveillance for Homeland Security. Critics called it tech's most controversial startup in 2019, and in 2024 China sanctioned it. The money followed the contracts. In August 2024 the valuation was $14 billion. In May 2026 a roughly $5 billion round priced it near $61 billion, on about $2.2 billion of 2025 revenue. Luckey has said he is driven by vengeance toward Facebook. Whatever you make of that, the pattern is plain. The year he lost the big job was the year he started the company that now makes fighter drones. Being pushed out cost him a place, and it gave him a market. ## Ramp (USA) URL: https://startup.talk/usa/ramp/ Founders: Eric Glyman, Karim Atiyeh, Gene Lee Founded: 2019 (March 2019; launched publicly Feb 2020); Based in: New York, New York; Sector: Fintech: corporate cards, spend management, bill pay, AI finance agents Before Eric Glyman ran a $44 billion company, he worked the floor of a discount clothing store in Las Vegas and watched families fall for the word sale. They bought, then days later saw the same items drop another 25 percent. He has told the story as his first lesson in how many businesses profit from customers not paying attention. He grew up in Las Vegas and went to Harvard, where he met Karim Atiyeh, a Beirut-born engineer who would later move to New York for a consulting job. For a while Glyman worked at a firm that helped struggling companies restructure through bankruptcy. He saw how good businesses die from sloppy spending. In 2013 a flight he had booked dropped by about $100, and he wished some software had told him. That idea became Paribus, which he, Atiyeh and engineer Gene Lee launched in 2014. It watched online purchases and asked stores for refunds when prices fell. It raised about $2 million and drew around 700,000 users. Glyman has said both Paribus and, later, Ramp were dismissed at first. He has also described seeing 80 percent of Paribus revenue vanish overnight. Capital One bought Paribus in 2016. What almost nobody remembers is what the exit felt like from inside. Glyman has said that after the team launched a late-shipment refund feature, his manager called him in. He had expected praise. He got pushback, because the bank had processes for everything. Founders who once moved in days now worked in a company where a good idea had to pass through many hands. They stayed until 2019 and watched a card business from the inside. That March, the three started again. They spoke with about 100 finance experts before building anything, and kept hearing the same complaints: collecting receipts, logging expenses, chasing employees. Ramp launched in February 2020 with a roughly $15 million Series A led by Founders Fund. The pace was fast, then it turned. A $115 million round in April 2021 made it a $1.6 billion company, the fastest New York startup to reach that mark. In early 2022 it had $100 million in annualized revenue and an $8.1 billion valuation. In August 2023 a $300 million round marked it down to $5.8 billion. It kept selling. Annualized revenue reached about $1 billion in August 2025, and in June 2026 the valuation hit $44 billion, when Glyman and Atiyeh became co-CEOs. The thread running through it is a small, stubborn one. A young man watched shoppers get nudged by a sale sign, and years later built a card whose pitch is that it helps customers spend less. The lesson for founders is to keep the irritation that started you, even after the money arrives. ## Cognition AI (USA) URL: https://startup.talk/usa/cognition-ai/ Founders: Scott Wu, Steven Hao, Walden Yan Founded: 2023; Based in: San Francisco, California; Sector: AI software engineering agents (Devin, Windsurf/Devin Desktop) In 1989 Scott Wu's father landed in the United States with about $80. Wu has told the story of a cab ride that year: his father tipped a dollar or two and realised it was roughly three percent of everything he owned. Wu was born in 1996 and grew up in suburban Baton Rouge with an older brother, Neal. He has said he was a loner and was bullied, starting around sixth grade, until competitions put him in a room with kids like him. At eight he entered a seventh-grade math division and was not called at the award ceremony. He has said he was furious about it. He went on to win gold at the International Olympiad in Informatics in 2012, 2013 and 2014, first overall in the last year, and helped a team take gold at the 2016 world programming finals. At seventeen he left high school after junior year to work in California. He did about two years at Harvard, studying economics, and has said he sort of knew he would not graduate. In the summer of 2017 he moved to San Francisco and started Lunchclub with Vlad Novakovski. It began as a scheduling app and became a service matching professionals for meetings. He left in June 2022 and has described it as a company that went okay but was not a smashing success. What almost nobody knows is what happened next. In 2020 his mother became seriously ill and he moved back to Louisiana for most of the pandemic to be with her. He returned again in 2023 as she got worse. She died on 6 October 2023 of stage-four lung cancer. Six weeks later, on 17 November, he was at lunch in New York pitching an AI lab when Sam Altman was fired. He has said the news pushed his team to move immediately. He cancelled Thanksgiving with his father, rented a house in Burlingame and invited friends to build with him. The house ran for two weeks, then a second in December, then on through January. A conference room at Cognition still carries that house's address. The team was Wu, Steven Hao and Walden Yan, all IOI gold medallists. Devin launched in March 2024 at 13 percent on the SWE-bench test and drew harsh criticism. Founders Fund backed it at $350 million, then $2 billion in April. In July 2025 Cognition agreed to buy Windsurf after Google hired its leaders. On 8 September 2026 it announced a Series E of more than $2 billion at a $48 billion valuation. Wu's father once counted a tip as a measurable share of his wealth. His son started a company weeks after his mother died, with nothing but friends and a reason not to wait. ## Figure AI (USA) URL: https://startup.talk/usa/figure-ai/ Founders: Brett Adcock Founded: 2022; Based in: San Jose, California; Sector: Humanoid robotics and AI Brett Adcock has said that when he started Vettery he had no money. He was dead broke, in his words, a twenty-five-year-old in New York in 2012 with an idea and a co-founder, Adam Goldstein. He had grown up on a corn and soybean farm outside Moweaqua, Illinois, a third-generation family operation. He was born in April 1986, finished as valedictorian at Central A&M High School and, at sixteen, began building web businesses alone. There was an outdoor electronics site, Street of Walls for job interview content, and a mobile job application app. He has said some made money and others were just not working. He studied at the University of Florida, graduating in 2008 with a business degree. Vettery was meant to help recruiters. It did not take off. What almost nobody sees behind the later headlines is that the company pivoted several times, and Adcock has called the last pivot quite a desperate one, a move to a single marketplace for job seekers and employers. That version worked. It raised $1.7 million in seed money in 2015 and $9 million in 2016. By 2017 it had about 300 employees, roughly 20,000 customers and about 30,000 interviews a month. In February 2018 Adecco bought it for $110 million. He did not stop. In October 2018 he co-founded Archer Aviation, which listed in February 2021. After the sale he moved to Palo Alto, saying he wanted to be thoughtful about what he built next. In 2022 he founded Figure. The new company was paid for first by him. He put $20 million of his own money into the May 2023 round of $70 million alongside Parkway Venture Capital. In February 2024 came $675 million from Jeff Bezos, Microsoft, Nvidia, Intel and the investment arms of Amazon and OpenAI, at $2.6 billion. BMW signed in January 2024. Then the OpenAI partnership ended after about a year, and Figure built its own model, Helix. There was a price. Fortune questioned the scope of the BMW work in April 2025, and a former head of product safety sued in November 2025, alleging he was fired for raising safety concerns. Even so, Figure raised more than $1 billion at $39 billion in September 2025 and had delivered more than 350 Figure 03 robots by April 2026. Adcock told Time in October 2025 that within ten years every home would have a humanoid. The man saying it once had nothing but a plan and a pivot he called desperate. The takeaway is practical. A founder's first idea is rarely the one that pays. The breakthrough at Vettery came from a change made when options were short. ## Safe Superintelligence (USA) URL: https://startup.talk/usa/safe-superintelligence/ Founders: Ilya Sutskever, Daniel Gross, Daniel Levy Founded: 2024 (announced 19 June 2024); Based in: Palo Alto, California (with an office in Tel Aviv); Sector: Frontier AI research (safe superintelligence) When Ilya Sutskever was sixteen, his family moved to Canada, and he lasted one month in high school. Then the University of Toronto admitted him as a third-year undergraduate. The detail sounds like a brag, but it also describes a boy who had changed countries twice. He was born in 1986 in Gorky, in the Soviet Union, to a Russian-Jewish family. At five he emigrated to Israel and grew up in Jerusalem. A good student, he started taking classes at the Open University of Israel in eighth grade. As a teenager he reportedly wondered how learning works, and believed the mystery was not limited to humans. His father was an engineer who took him, at about five, to an expo, where the technology captivated him. In Toronto he studied mathematics, graduating in 2005, took a master's in 2007 and finished a PhD in 2013 under Geoffrey Hinton. In 2012 he built AlexNet with Hinton and Alex Krizhevsky, the image-recognition network that helped start the deep-learning era. While still a student he visited DeepMind and turned down a job there, thinking the researchers had lost touch with reality because nobody was seriously discussing AGI. Google hired him, and at the end of 2015 he left to co-found OpenAI as chief scientist. The next scene is the one that defines the public story. In November 2023 he voted, as a board member, to remove Sam Altman. At an all-hands meeting he called it the board doing its duty. Within days he expressed regret and joined the employee letter demanding the board resign, the very board he sat on. He reportedly had compiled a 70-page dossier on Altman before the vote. Altman was back within a week. In May 2024 Sutskever left. On 19 June 2024 he announced Safe Superintelligence with Daniel Gross and Daniel Levy, promising that the first product would be safe superintelligence and nothing before it. In September 2024 it raised $1 billion at about $5 billion. A Greenoaks-led round in 2025 priced it near $30 billion, with no revenue. Meta tried to buy it and was turned away. In July 2025 Gross left for Meta, and Sutskever became chief executive. Nvidia announced a partnership on 27 July 2026. Sutskever has left two countries and one company already. The record suggests a man who has learned that the place that admits you is not always the place you belong, and who keeps starting over. ## VAST Data (USA) URL: https://startup.talk/usa/vast-data/ Founders: Renen Hallak, Shachar Fienblit, Jeff Denworth Founded: 2016 (January 2016); Based in: New York, New York; Sector: AI data infrastructure, flash storage software In 2015 Renen Hallak walked out of a job he was good at and had nothing to show for it. He had been the first engineer at XtremIO, and later its VP of R&D, leading a group of several hundred engineers. The company had been sold to EMC, reportedly for about $430 million. He left anyway, with no product and no company, only a plan to ask questions. For months he talked to storage buyers and storage makers about what they were stuck with. The same complaint kept coming back. Flash was too small and too costly to hold everything, disk was too slow, so everyone lived with tiers: a fast pile, a slow pile, and people paid to shuffle data between them. What almost nobody notices is where he started. Hallak did not come to storage from hardware. He began his career in computational complexity research, a field nearer to mathematics than to disk drives, and joined XtremIO at its start in 2009. That habit of looking for a clean structure under a messy system shaped his bet. The bet was specific. Cheaper QLC flash, better data reduction, a new kind of persistent memory for writes and metadata, and NVMe over fabrics might together let one flash system replace both tiers. In January 2016 he founded VAST Data with Shachar Fienblit, formerly CTO of Kaminario, and Jeff Denworth, who came from CTERA. Research and development stayed in Israel. Headquarters went to New York, close to the likely buyers. Then came the long, quiet cost. A storage architecture cannot be shipped half finished, so VAST spent years in stealth building something that would either work or not. A $40 million Series A arrived in 2017 and a second $40 million round followed. When the company surfaced in 2019 it had hired the analyst Howard Marks as evangelist, which the trade press noted was his first full-time job since 1987. Nobody could schedule what came next. After ChatGPT launched in late 2022, every large AI data centre needed to feed thousands of GPUs at once, and a design with no tiers looked like foresight. VAST raised $118 million at about $9 billion in 2023 and signed a $1.17 billion agreement with CoreWeave in November 2025. In April 2026 it raised about $1 billion at a $30 billion valuation. Hallak rarely tells personal stories in public, and where he was born or studied is not in the record. What is there is a habit a first-time founder can copy: he asked customers what hurt before he wrote a line of company code, then waited years for the world to catch up. ## Scale AI (USA) URL: https://startup.talk/usa/scale-ai/ Founders: Alexandr Wang, Lucy Guo Founded: 2016; Based in: San Francisco, California; Sector: AI data labeling, model evaluation and training data In sixth grade, Alexandr Wang narrowly placed fourth in his state math competition. Fourth place won him a trip to Disney World, and he has called it the start of his addiction to math contests. He was a kid in Los Alamos, New Mexico, a town of about 10,000 people where the main employer was the national laboratory. Both his parents were physicists there. They had come to the US in the 1980s for their PhDs, and his mother has worked only at the lab. He joined the Rocket Club in fourth grade, played violin competitively, and has said he learned to handle stress early. His parents, he has said, would have loved it if their children also did physics. He chose computer science. He went to MIT, took graduate machine learning courses as a freshman, and left after a year because, in his words, he really wanted to study this AI thing. By 19 he had founded Scale AI. A co-founder came with him: Lucy Guo, raised in Pleasanton, California, who as a teenager had written bots for Neopets and sold the in-game assets for money. She had left Carnegie Mellon for a $100,000 Thiel Fellowship in 2014. They went through Y Combinator in 2016. Wang has said he saw Nvidia leading on compute while nobody was focused on the data AI would need. In 2019 Founders Fund put in $100 million and the valuation passed $1 billion. Then the partnership broke. Guo was fired in 2018, two years in, and the sources do not say why. She went on to found Passes. Wang stayed, owned about 15 percent by 2021, and became a billionaire at 24 when a round valued Scale at $7 billion. The work was never as clean as the valuation. Remotasks, the contractor platform, drew criticism over pay cuts, late payments and for obscuring its link to Scale, and lawsuits in 2024 and 2025 alleged wage theft and psychological harm from disturbing content. In January 2023 Scale cut 20 percent of its staff. The bet still paid. Revenue was $870 million in 2024. In June 2025 Meta agreed to take 49 percent for roughly $14.3 billion, Wang left to join it, and Jason Droege became CEO. What lasts from that Disney trip is the pattern: a boy who barely made fourth place kept entering contests, and the habit of entering was the whole advantage. The partner who started beside him left in 2018 and built something else. ## Fanatics (USA) URL: https://startup.talk/usa/fanatics/ Founders: Michael Rubin, Alan Trager, Mitchell Trager Founded: Fanatics, Inc. listed as founded 2011 by Rubin; Rubin's Global Sports Inc. dated 1998; Based in: New York, New York (originally Jacksonville, Florida); Sector: Sports merchandise, licensed apparel, collectibles (Topps), sports betting At sixteen, Michael Rubin owed his creditors about $120,000. He was a high-school kid from Lafayette Hill, outside Philadelphia, and the debt was the price of a business he had started in his parents' basement at twelve, tuning skis. At fourteen he had opened a real shop in Conshohocken, Mike's Ski and Sport. The seed money was $2,500 in bar mitzvah gifts, and the lease needed his father's signature. His father was a veterinarian and his mother a psychiatrist. The shop grew to five locations, and the debt grew with it. What almost nobody remembers is what his father did when the bill came due. He lent his son $37,000 to settle with the creditors, and attached one condition: Rubin had to go to college. It was a rescue with a string on it. Rubin kept the deal, for one semester. At Villanova he spotted a chance to buy $200,000 of overstock sports equipment for $17,000 borrowed from a friend, and resold it for $75,000. He took that as his answer and left school. His next company, KPR Sports, was named for his parents' initials. It sold closeout gear and reached $1 million in sales in 1993 and $50 million by 1995. In 1998 he created Global Sports, which became GSI Commerce and ran online stores for other brands. Fanatics itself was older than his claim to it. Brothers Alan and Mitchell Trager opened Football Fanatics in a Jacksonville mall in 1995, selling Jaguars gear. In 2011 eBay bought GSI for $2.4 billion, wanting the fulfilment business. Rubin bought back the consumer sports businesses, Fanatics among them, and a $150 million round from Insight Venture Partners and Andreessen Horowitz followed in 2012. It has not been smooth. Panini sued in 2023 alleging anti-competitive behaviour, and MLB jerseys drew fan fury in 2024. Still, by December 2022 Fanatics was valued at $31 billion, in 2021 it bought Topps after that company's SPAC deal fell through, and in the pandemic it made over a million masks and gowns from jersey fabric. The first lesson he got was the expensive one, and it came before he could legally drive: borrow more than you can repay, and someone has to catch you. His father did, once, and told him to go to school. He found the next deal instead. ## Epic Games (USA) URL: https://startup.talk/usa/epic-games/ Founders: Tim Sweeney Founded: 1991 (as Potomac Computer Systems, from his parents' house in Potomac, Maryland; Epic MegaGames 1992; Epic Games 1999); Based in: Cary, North Carolina; Sector: Video games, game engines (Unreal), digital storefront The last copy of ZZT went out in November 2013, and the man who mailed it was Tim Sweeney's father. For years Paul Sweeney had filled orders for his son's first game from the family home in Potomac, Maryland. Sweeney was the youngest of three brothers. His father worked for the Defense Mapping Agency and gave him an IBM PC/AT around the time he started college. An older brother ran a startup in California, where Sweeney at eleven spent a week learning BASIC. Back home he taught himself on the family Apple II, and he later estimated more than 10,000 hours of practice between eleven and fifteen. He said he took apart a lawnmower at five or six. He finished several games as a teenager and showed none of them to anyone. He also mowed lawns for wealthy neighbours at half the price of the professionals. At the University of Maryland he studied mechanical engineering, and in 1990 he built a text editor in Pascal that picked up game-like elements and became ZZT. He wrote to Scott Miller of Apogee to ask how shareware worked. ZZT came out on January 15, 1991. Here is the part almost nobody knows. The company that would make Fortnite began as his parents' home consulting business, Potomac Computer Systems, revived to sell one game. Orders came to the house. Within months, one account says by October 1991, he renamed it Epic MegaGames to sound like a big, successful company. It was one young man in a family house. The word Mega stayed until 1999, when the company moved to Cary, North Carolina, and stopped pretending to be big. Sales reached roughly $100 a day, enough that he left Maryland one credit short of a degree. Mark Rein came on as a partner, and Cliff Bleszinski joined at 17. Unreal Engine and Gears of War in 2006 built the business, and Tencent paid $330 million for about 40 percent in 2012. Fortnite launched as a free battle royale in September 2017 and passed 125 million players by May 2018. In August 2020 Epic offered a cheaper in-game price that bypassed Apple's and Google's payments, both removed the game, and Epic sued. The Epic Games Store returned to iOS in April 2025. In March 2026 Epic announced more than 1,000 layoffs as Fortnite engagement fell. The first sale tells you the rest. A hundred dollars a day, packed and mailed by a father who kept doing it for twenty-odd years, was enough proof to put a name on the door that was bigger than the room. ## Kalshi (USA) URL: https://startup.talk/usa/kalshi/ Founders: Tarek Mansour, Luana Lopes Lara Founded: 2018 (public launch July 2021); Based in: New York, New York; Sector: Prediction markets, event contracts, sports Every lawyer they called said no. By one Bloomberg account, Tarek Mansour and Luana Lopes Lara cold-called about sixty lawyers they found on Google while at Y Combinator, and each told them to give up. Another account says more than forty. They wanted to build an exchange where anyone could trade on whether something would happen, and nobody wanted to explain how. Lopes Lara was born in Belo Horizonte, Brazil, in 1996. Before she did math she did ballet. She trained for eight years at a Bolshoi academy in Joinville and later spent a few months working on a Swan Lake production at the Salzburger Landestheater in Austria. She also won a gold medal at the Brazilian Astronomy Olympiad and a bronze in the Santa Catarina Mathematics Olympiad. Somewhere between the stage and MIT she changed disciplines. The sources do not say why. Mansour was born in California to Lebanese parents and raised in Lebanon, then returned to the US in 2014 for MIT. The two met there. They graduated in 2018, he in mathematics and computer science, she in computer science. Mansour interned at Goldman in 2016 and traded at Citadel until 2018; Lopes Lara interned at Bridgewater and Citadel Securities. Mansour has said those jobs inspired Kalshi. The company was first called Kownig. Then the wall. Y Combinator took them, and Michael Seibel has said the idea was so unlikely to succeed that he had to take a chance. Rather than dodge regulators, the founders slowed down, hired lawyers and walked into the CFTC's office voluntarily. In November 2020 Kalshi was licensed as a designated contract market and in July 2021 it launched publicly. The next fight was longer. From 2022 the CFTC delayed election contracts, and in September 2023 it rejected Kalshi's congressional-control contracts. Kalshi sued. In 2024 a federal court said the agency had overstepped, an appeals court refused to pause the ruling, and the markets reopened. What followed was stranger than the plan. Sports became about 89 percent of fee revenue, roughly $263.5 million in 2025. Valuation reached $11 billion after a $1 billion round in December 2025, then $22 billion in a $1 billion Series F by May 2026, and Forbes called both founders billionaires. Lopes Lara, with roughly 12 percent ownership, became the youngest self-made woman billionaire. The ballet years suggest what she brought: eight years of repeating one thing until it was clean, then taking it into a field where hearing no sixty times was the work. They treated the regulator as the first customer to win, not an enemy to avoid. ## Etched (USA) URL: https://startup.talk/usa/etched/ Founders: Gavin Uberti, Chris Zhu, Robert Wachen Founded: About 2022 (an Aug 2026 article calls it four years old); Based in: San Jose, California; Sector: AI semiconductors, transformer-specific inference chips (Sohu ASIC) Gavin Uberti was 21 when an interviewer introduced him as the Harvard dropout who meant to take on Nvidia. It is a strange sentence to hang on a person, and it sums up the problem of telling his story: the public record gives us the bet far more clearly than the boy. What the record does say is short. He studied mathematics and computer science at Harvard and left without finishing. He was a 2024 Thiel Fellow, a programme that pays young people to skip college. His company's site lists him as a Math 55 alumnus, the notoriously hard Harvard course, and as an expert in AI compilers. Two others, Chris Zhu and Robert Wachen, are named alongside him as founders of Etched, and all three are described as Harvard dropouts. Where they grew up, what their parents said, what money they had: none of that has been published in anything we could read. We will not guess. So start with the choice. Nvidia's GPUs can run almost anything. Etched decided to make a chip that runs one thing, the transformer models behind ChatGPT, Claude and Gemini. The chip is called Sohu, an ASIC made by TSMC on a 4 nm process. The company says one Sohu server can stand in for 160 Nvidia H100s when running trained models. Nobody outside the company had tested that claim as of August 2026. What almost nobody outside the industry saw is how long it all sat in silence. Etched stayed in stealth for years. A $120 million Series A is reported, then a valuation near $5 billion in December 2025 and a $500 million round in January 2026 that the company only confirmed months later. On 30 June 2026 it came out with a working chip, about $800 million raised and more than $1 billion in signed customer contracts. Then the money got louder. A $300 million Series C led by Sequoia in July put the price at $10.3 billion. In August, $700 million led by Jane Street lifted it to $21 billion. That deal carries its own awkwardness. Jane Street is also Etched's first paying customer, so the loudest vote of confidence comes from a buyer. By then only a single rack had been delivered. No revenue has been published. The cost of the bet is built into it. Sohu works only for transformers. If the field moves to something else, a chip that took years to make loses most of its use. A few founders who left college early have put their whole company behind one prediction about the future. The lesson for a young founder is not about being brave. It is about being specific. Etched did not promise to be a bit better at everything. It chose the single assumption it could not hedge, and then waited years, quietly, to find out whether it was right. ## Perplexity AI (USA) URL: https://startup.talk/usa/perplexity-ai/ Founders: Aravind Srinivas, Denis Yarats, Johnny Ho, Andy Konwinski Founded: 2022 (22 Aug 2022); Based in: San Francisco, California; Sector: AI search, answer engine, AI browser (Comet) and agents "For my mom, just getting a job was success," Aravind Srinivas wrote in a June 2026 X post, "because we were financially lower middle class in India." A job at Google, he said, was the far end of her hopes. He is now the chief executive of a company that once offered $34.5 billion for Google's browser. He was born in Chennai. He has said he is the first engineer in his extended family, and that his mother had long pictured him at IIT Madras. He went, and studied electrical engineering for both a bachelor's and a master's. One profile says he narrowly missed the grade cut-off to switch into computer science, a detail that comes from secondary coverage and not from him. What the sources agree on is the stake: a family for whom a salary was the finish line. He did not stop at a job. He did research at OpenAI and Google DeepMind, then on 22 August 2022 started Perplexity with Denis Yarats, Johnny Ho and Andy Konwinski. The product was easy to describe. Ask a question, get a written answer with citations. It launched on 7 December 2022, a week after ChatGPT, and reported two million unique visitors by February 2023. In January 2024 it raised $73.6 million at a $520 million valuation. By September 2025 the valuation was $20 billion. What sits next to those numbers is how little he says he feels he has to protect. "I have nothing to lose," he wrote in the same June 2026 post, and he urged people to attack, attack, attack. The cost of that posture arrived in public. In June 2024 Forbes accused Perplexity of copying its reporting. Srinivas answered that the feature had rough edges and that Perplexity only aggregates. Then came suits from Dow Jones and the New York Post, three Japanese newspaper groups, Britannica and Merriam-Webster, Reddit and, in May 2026, CNN. In August 2025 Cloudflare said Perplexity's crawlers hid their identity to get past blocks, and the company called that a publicity stunt. It kept moving anyway. It launched the Comet browser in July 2025, signed a three-year, $750 million Microsoft Azure deal in January 2026, and dropped advertising for subscriptions the next month. Its valuation sat near $21 billion in early 2026. The quiet part is the distance. A boy whose family measured success by a salary now runs a company that publishers are suing and Google cannot ignore. Fearlessness, in his telling, comes from having little to protect. The open question is what he does once he has something to lose. ## Chobani (USA) URL: https://startup.talk/usa/chobani/ Founders: Hamdi Ulukaya Founded: 2005 (as Agro Farma; brand launched 2007); Based in: Norwich, New York (new NYC headquarters, Chobani House in NoHo, announced 2024); Sector: Food and beverage, Greek yogurt, dairy and coffee The flyer went in the trash. In 2005 a brochure reached Hamdi Ulukaya's mailbox advertising a yogurt plant Kraft was closing in South Edmeston, New York. He threw it away. The next day he drove to see the plant anyway. He was born in a Kurdish dairy-farming family in Ilic, a village in Erzincan, Turkey, one of seven children. The family kept sheep and goats near the Euphrates, made cheese and yogurt, and moved with the flocks through the mountains. He studied political science at Ankara University and joined the Kurdish-rights movement, publishing a newspaper critical of how the state treated Kurds. That drew the attention of the authorities. In 1994 he left for the United States to study English at Adelphi, and he has named escaping that treatment as a reason. What he cannot tell you is his own birthday. He was born during a trek in the mountains, so no one recorded the date. He uses 26 October. In America he worked on an upstate farm and moved to the University at Albany in 1997. Then his father suggested he import the family's feta from Turkey, because the local cheese was poor. It sold. In 2002 he opened a small plant, Euphrates, in Johnstown. By its second year it only just broke even. The Kraft plant was a bigger gamble. His lawyer and business advisor told him not to. He bought it anyway, financing it within five months through an SBA loan and local grants. Then came nearly two years with a Turkish yogurt master working out a thick strained recipe. He later called them "two years of the most challenging days of my life." There was no money for ads or for the fees stores charge for shelf space, so he paid in yogurt and negotiated to settle the fees over time. Chobani launched in 2007, entered Costco and BJ's in 2009 and passed $1 billion in annual sales in 2010. Success brought bills. A $130,000 fine in 2012 for overpumping water. A recall in 2013. Boycott calls and threats in 2016 over hiring refugees, the year he gave employees a 10 percent stake. An IPO filed in 2021 was withdrawn in 2022 after five years of net losses. In October 2025 the company raised $650 million at about a $20 billion valuation. Ulukaya did not invent yogurt. He came from people who had always made it, arrived somewhere that did not know what it was missing, and kept looking at the flyer he had already thrown away. ## Rippling (USA) URL: https://startup.talk/usa/rippling/ Founders: Parker Conrad, Prasanna Sankar Founded: 2016 (history section; infobox says 2017); Based in: San Francisco, California; Sector: HR, payroll, IT and finance software (workforce management) Parker Conrad has called it "humiliating and shocking". He was managing editor of The Harvard Crimson, and he failed out of Harvard. Before that he had been the sort of student things went right for. Born in New York City in 1980, he went to the Collegiate School, and his research on sea-snail neurobiology placed third nationally in the Westinghouse Talent Search. At Harvard the newsroom took over. Then came the failure. What almost nobody recalls when they read about Zenefits is what he did next. He did not quit. He took a year at the Arkansas Democrat-Gazette, then came back and graduated in 2003 with a chemistry degree. Business press later used phrases like "Harvard flunk-out". He used a plainer word for how it felt. He worked as a product manager at Amgen, then co-founded Wikinvest, later SigFig, in 2007. He left in 2012 after a falling out with his co-founder, joined Y Combinator and started Zenefits. At 24 he had been treated for testicular cancer, a fact that sits oddly beside the confidence of the rest of the story. Zenefits was named the fastest-growing startup of 2014. It was also found to be using unlicensed insurance brokers. On 8 February 2016 Conrad resigned. In 2017 he agreed to an SEC fine over misleading investors on compliance, without admitting or denying guilt. Within about two months he started again. With Prasanna Sankar, who had led engineering at Zenefits, he founded Rippling in 2016 and worked from Conrad's San Francisco home for roughly two years in stealth. In 2017 he raised $7 million, and press still introduced him as Zenefits' former CEO. The idea was to keep one clean record of each employee and hang everything off it: payroll, benefits, laptops, app logins. Then, in March 2023, Silicon Valley Bank failed and customers' paychecks were stuck. Rippling raised about $500 million in roughly 12 hours, led by Greenoaks, and moved its banking to JPMorgan. In March 2025 it sued rival Deel over alleged corporate espionage; a judge refused Deel's bid to end the case in 2026. A $450 million round in May 2025 valued Rippling at $16.8 billion, and annual revenue was reported near $1 billion in July 2026. The quiet pattern is this. Conrad has left three things under a cloud: a university, a co-founder partnership, a company. Each time he went somewhere smaller, worked at something unglamorous, and came back to the same kind of problem with sharper eyes. ## Fal (USA) URL: https://startup.talk/usa/fal/ Founders: Burkay Gur, Gorkem Yurtseven, Batuhan Taskaya Founded: 2021; Based in: San Francisco, California; Sector: Generative media AI inference platform (image, video, audio models through one API) Before fal was a generative media company worth $16 billion, it sold something far less glamorous. Burkay Gur's company first built machine learning pipelines for fraud detection. That was the early life of a firm whose legal name, Features & Labels Inc., sounds like a data-science textbook. Gur and his co-founders started it in 2021. About a year and a half in, Gur told StartupHub, "ChatGPT happened, DALL-E happened." The company turned toward running image and video models. There is no public tale of an overdraft or a lost apartment here. What is documented is who they are. Gur and Gorkem Yurtseven are both reported to be from Turkey, and to have connected in San Francisco in the early 2010s. Gur worked at Oracle and then led machine learning at Coinbase. Yurtseven was a developer at Amazon. Batuhan Taskaya, named in the StartupHub interview as a co-founder, runs engineering. The stakes were practical. Gur said the team had little GPU capacity during the chip shortage, which forced them to squeeze more out of each machine. That scarcity became the product. fal gives builders one API for image, video and audio models, and post-trains some models to make them faster. Its customers include game developers, ad agencies and entertainment studios. The pace afterward was extreme. In December 2025 the valuation was $4.5 billion. A $350 million Series E led by Sequoia in April 2026 took it to $8 billion. On 3 November 2026 a $1.3 billion Series F led by Bessemer, with Nvidia, Sound Ventures, HOF Capital and T. Rowe Price, set it at $16 billion. Annualized revenue was reported at about $200 million in October 2025 and above $800 million in 2026, the latter from an unnamed source speaking to Bloomberg. The price of that speed is a thin margin. Compute is expensive, and Fireworks AI closed $1.5 billion at a $17.5 billion valuation in July 2026. fal had 185 employees at the time of the Series F. A founder can take one thing from the record. The first product was fraud detection, and the company now worth $16 billion came from noticing what the people around it were suddenly trying to run, and building for that. ## Harvey (USA) URL: https://startup.talk/usa/harvey/ Founders: Winston Weinberg, Gabriel Pereyra Founded: 2022 (summer); Based in: San Francisco, California; Sector: Legal AI / enterprise software Winston Weinberg was a first-year associate at O'Melveny & Myers, a securities and antitrust litigator, when he was handed a landlord-tenant case in a field he did not know. His roommate in Los Angeles, Gabriel Pereyra, a research scientist who had worked at Google DeepMind and Meta, showed him GPT-3. Weinberg's first use of it, he has said, was running a Dungeons & Dragons game for friends. Then he pointed it at the law. They wrote a long prompt about California tenant statutes and ran it on 100 questions from a public legal-advice forum. Three attorneys reviewed the answers without knowing a model wrote them. On 86 of the 100, at least two said they would send the answer to a client unedited. That got them a call. They had cold-emailed Sam Altman and OpenAI's general counsel, and the pitch reached OpenAI's leadership on a July 4 call. The OpenAI Startup Fund wrote a $5 million cheque in late 2022. The name came from Harvey Specter, the lead character of the TV drama Suits. Weinberg has been candid about how foreign it all was. He had no tech friends, did not grow up in San Francisco and did not know how fundraising worked. What he knew was how lawyers think. Early sales used public litigation briefs: Harvey showed the lawyers who had written them how to argue against their own work. Allen & Overy began a trial in November 2022 that reportedly reached 3,500 lawyers. Here is the part that rarely gets told. In early 2024, with Harvey valued around $700 million, Weinberg pursued a merger with a more traditional company. It fell apart because Harvey would have taken on too much debt to pay for it. He has recalled a week spent asking what the company's future should be. He has since made failure a working rule. He calls it "a very good way to learn", describes the job as "destroying your ego 24/7", and says he points out about 15 failures a day to his own staff, who must re-earn their roles every six months. The numbers kept moving. A $715 million valuation in December 2023 became $3 billion in February 2025, $8 billion in December 2025 and $15.5 billion in September 2026. Revenue passed $100 million ARR in August 2025 and was about $190 million at the end of the year. The lesson sits in the near-miss. The deal that nearly happened would have bought size with debt. The company that exists instead was built on a test a stranger could grade, and on a founder who says out loud how often he is wrong. ## Helion Energy (USA) URL: https://startup.talk/usa/helion-energy/ Founders: David Kirtley, John Slough, Chris Pihl, George Votroubek Founded: 2013; Based in: Everett, Washington; Sector: Fusion energy "It is daily that we think about that deadline." David Kirtley said that on Lex Fridman's podcast in late 2025, and the deadline is not abstract. Helion has promised Microsoft 50 megawatts of electricity from fusion by 2028. "It's going to be hard," he added. "No joke about it." Kirtley is a plasma engineer. He has described how he arrived at fusion from the point where energy meets space travel. He studied aerospace engineering, took a PhD at the University of Michigan, worked at the Air Force Research Laboratory and NASA Glenn, and from 2008 led fusion work at a small Redmond research firm, MSNW. Helion grew out of that lab with John Slough, Chris Pihl and George Votroubek. The first money was public: about $7 million in government grants. In August 2014, Y Combinator and Mithril Capital added $1.5 million. Kirtley has said the earliest work was funded by government grants, including small-business research awards. The idea was unfashionable. Instead of a doughnut-shaped tokamak, Helion built a linear machine that fires pulses and recovers energy directly from the expanding plasma. Kirtley told Fridman that people had said it could never work, because the plasma it depends on is unstable when compressed. In 2018 a MITRE and JASON review for ARPA-E said the same thing in careful language: compression without losing stability was the open question. In 2019 a retired Princeton plasma physicist, Daniel Jassby, put some fusion companies among what he called voodoo fusion. Then Helion made a promise of its own. In 2021 it said it would show net electricity in 2024. It did not. The money stayed anyway. In November 2021 Helion raised $500 million, with another $1.7 billion committed against milestones, largely led by Sam Altman, who became chairman. In May 2023 Microsoft agreed to buy 50 MWe from 2028, the first commercial fusion power purchase agreement, and Nucor followed with a 500 MWe plan. In February 2026 its Polaris machine produced measurable deuterium-tritium fusion at 150 million degrees Celsius. The company is building Orion at Malaga, Washington, and in September 2026 closed a Series G at a $15.5 billion valuation, with about 800 employees. The date is still the test. Kirtley did not hide from that in the interview. Each of his machines, Trenta in 2021 and Polaris in 2026, delivered a number the last one had not, and that has been what kept people paying. ## Discord (USA) URL: https://startup.talk/usa/discord/ Founders: Jason Citron, Stanislav Vishnevskiy Founded: 2015 (launched May 2015; built inside Hammer & Chisel, founded 2012); Based in: San Francisco, California; Sector: Communication platform / social The game was called Fates Forever, a mobile MOBA, and in 2014 it did not sell. Jason Citron had a studio called Hammer & Chisel, a team, and a flop. He had been through a version of this before. Born in San Francisco in 1984 and raised in South Florida, he got a computer from his grandfather and learned QBasic at 13 with a friend. His first program was a text-based role-playing game. He has called himself a Squaresoft fanboy. Here is the part few people know. In high school, in 1999, he started a company writing code for internet services. One weekend he earned $500 writing PHP for clients overseas. His parents worried he was mixed up in something improper, until he explained it was coding work. His first real paycheck had looked to the people who loved him like trouble. He studied game design at Full Sail University, graduating in 2004, then worked as a gameplay programmer and a tools engineer at Double Fine Productions. He bought one of the first iPhones, decided within about six months it would suit mobile games, and when Apple announced the App Store he dropped a social-network idea. Using a jailbroken development kit he built the game that became Aurora Feint, on the App Store the day it launched. The social layer inside it became OpenFeint, started in 2009 and grown to 100 employees. GREE bought it in 2011 for about $104 million. He calls it his only significant financial success. Then Fates Forever. While building it, his team hit a small irritation: the voice software they used to coordinate tactics was heavy and hard to set up. He and engineer Stanislav Vishnevskiy built a lighter chat service and called it Discord. It launched in May 2015. Gaming subreddits began swapping IRC links for Discord links. By December 2018 Greenoaks led $150 million at a $2 billion valuation. Independence has cost. In 2021 Microsoft was reportedly in talks to buy Discord for over $10 billion; the talks ended, and a September round set the price near $15 billion. Extremist groups used its servers, including those tied to the 2017 Charlottesville rally and the 2022 Buffalo shooting. A French regulator fined it 800,000 euros in 2022. It cut 4% of staff in August 2023 and 17% in January 2024. In April 2025 Citron stepped aside as CEO, saying it was in anticipation of going public, and Bloomberg reported a confidential IPO filing in January 2026. No listing had been confirmed in November 2026. The founder's lesson is plain. The game flopped. The tool the team built for itself to play it is the one that lasted. ## Plaid (USA) URL: https://startup.talk/usa/plaid/ Founders: Zach Perret, William Hockey Founded: 2013; Based in: San Francisco, California; Sector: Fintech infrastructure (bank-data APIs) Zach Perret left his job at Bain on the strength of one bonus cheque for $4,000. It covered about two months of rent in New York. He has joked that his year at Bain felt like "12 months in one day", and he told the Acquired podcast that a bonus that size is what paid for his exit. He had enrolled at Duke in 2006. What he has described himself is the work: roughly 80-plus-hour weeks for a decade, starting at 18. Looking back, he said "I was not a good strategist in my 20s", and that he relied on working very hard instead. Plaid did not begin as Plaid. Perret and William Hockey built consumer budgeting software. His own verdict was blunt: "No one wants to use an app that tells you to spend less money." The wall was something else, though. Most of their time went into getting data out of banks. The turn came when Venmo's head of engineering offered to license their back end, and they dropped the consumer product to sell the plumbing: one API between apps and thousands of banks. The money followed. A $2.8 million seed in 2013, $12.5 million in 2014, $44 million in 2016. By the $250 million Series C in 2018, at a $2.65 billion valuation, Visa and Mastercard had both invested. Then Visa tried to buy it. In January 2020 it announced a $5.3 billion acquisition; Perret has said the all-hands that day was a mix of anger, surprise, joy and tears, and he could not tell whether the crying was happy or sad. Eleven months later the Department of Justice sued, and Visa walked away in January 2021. He called the Zoom all-hands that followed the hardest he had led. The independent path cost something too. The company was criticised for scraping data and imitating bank login screens, and TD Bank sued in 2020. In 2021 Plaid settled a class action for $58 million, covering about 98 million people. Hockey left that year to start Column. The $13.4 billion valuation of April 2021 fell to $6.1 billion in April 2025 and recovered to $8 billion in a February 2026 share sale. Revenue was $170 million in 2020, and third-party estimates put 2025 at roughly $550 million. Bloomberg reported in July 2026 that Plaid was exploring an IPO, and Perret has said conditions are not favourable. The company that exists is the one nobody planned. It came out of the failure of the one the founders wanted to build. ## OpenSea (USA) URL: https://startup.talk/usa/opensea/ Founders: Devin Finzer, Alex Atallah Founded: 2017 (December); Based in: Miami, Florida; Sector: NFT and crypto trading marketplace Before OpenSea there was WifiCoin, and it went nowhere. Devin Finzer and Alex Atallah had designed a token for sharing wi-fi, a way to pay for time on a stranger's router. They pitched it to Y Combinator. Then CryptoKitties, a game of collectible cats on Ethereum, gave them a different idea, and the wi-fi plan was dropped. They founded OpenSea in December 2017. It was not Finzer's first small bet. Born in 1990, he grew up around the San Francisco Bay Area, the son of a physician and a software engineer, and went to Miramonte High School in Orinda. In his junior year at Brown he and Dylan Field, later the founder of Figma, built CourseKick, a search tool for course registration. Reportedly one in five undergraduates had signed up within two weeks. He graduated in 2013 with a degree in computer science and mathematics, interned at the Wikimedia Foundation, Google Cloud and Flipboard, worked at Pinterest, and co-founded Claimdog, a personal finance app that Credit Karma bought. Y Combinator took the new idea in 2018. Money came slowly, then all at once: $2.1 million in November 2019, $23 million from a16z in March 2021, and $100 million that July at a $1.5 billion valuation. In January 2022 Paradigm and Coatue put in $300 million at $13.3 billion. Forbes put Finzer's stake at $2.2 billion, and he and Atallah were called the first NFT billionaires. Then came the price. In May 2022 the site handled about $2.7 billion in a single day. By the end of August the daily figure was $9.34 million, a fall of 99%. A phishing attack hit users in February, a vendor breach exposed 1.8 million email addresses in June, and in July the company laid off about 20% of its staff. Its head of product, Nate Chastain, was convicted in 2023 over insider trading and sentenced to three months; an appeals court overturned the conviction in July 2025. By April 2023 Forbes estimated Finzer's wealth at under $600 million. In August 2024 the SEC sent a Wells notice over platform tokens, and dropped the matter in February 2025. What kept the company alive was a change of shape. OS2 now supports tokens on 22 blockchains, and in the two weeks after its October 2025 release it handled $1.6 billion in crypto trades. The idea of becoming a trading app came, Finzer has said, from his wife, Yu-Chi Lyra Kuo, whom he met in 2021 and calls a silent cofounder of OpenSea 2.0. In October 2025 he said a SEA token would arrive in the first quarter of 2026. The last priced round is still the one from January 2022. The quiet lesson is in the order of things. The wi-fi token failed, the collectible-cat detour became a company, and the best idea for the second act came from the person closest to him, not the market. ## Superhuman (USA) URL: https://startup.talk/usa/superhuman/ Founders: Rahul Vohra, Conrad Irwin, Vivek Sodera, Max Lytvyn, Alex Shevchenko, Dmytro Lider Founded: 2014 (email app); 2009 (Grammarly parent); Based in: San Francisco, California; Sector: AI productivity software: email client (Superhuman Mail), writing assistant (Grammarly), docs (Coda) Rahul Vohra was 27 when he woke at 3 a.m. feeling sick and kept working. At lunch he set out for a pharmacy in San Francisco and collapsed on the street. He took painkillers, went back to the office and only the next day reached a hospital, where a nurse told him he looked 10 to 20 years older than he was. His company was Rapportive, a Gmail add-on with five people, out of the summer 2010 Y Combinator batch. Twenty months in, he has said, revenue was marginal, cash was running out and a funding round would not close. He was living on brownies and pizza and had stopped going to the gym. Earlier he had worked as a game designer at Jagex, the studio behind RuneScape, and one profile says he studied computer science at Cambridge. The rescue came in February 2012, when LinkedIn bought Rapportive for a reported $15 million. Then he stopped. He took a month away from work, spent it with family, played story-driven video games and learned transcendental meditation from an executive coach, a half hour every morning. What he did next was start again. In 2014 he founded Superhuman, a paid email app, with Conrad Irwin and Vivek Sodera. It charged $30 a month, $10 for students, and promised heavy users up to three hours back each week through keyboard shortcuts. It supported only Gmail at first; Outlook arrived in May 2022. It raised more than $100 million, including from Andreessen Horowitz, and was reported at an $825 million valuation in 2021. The buyer came from a different place. Grammarly was founded in 2009 in Ukraine by Max Lytvyn, Alex Shevchenko and Dmytro Lider. It raised $200 million in 2021 at a $13 billion valuation, which remains its last public mark. After Russia invaded Ukraine in 2022 it stopped operating in Russia and Belarus and kept paying staff who joined the Ukrainian army. In July 2025 Grammarly announced it would acquire Superhuman, and in October its parent renamed itself Superhuman Platform Inc. Rows followed in February 2026. In March 2026 a class action challenged its Expert Review feature, which used the names of real writers without permission, and the feature was dropped. No revenue figure was found. The sources do not say what the 2012 collapse cost Vohra beyond that month away. They show only that he went back to work, and that the later product was built to give busy people time. A first-time founder can take the plain version: the body sends its bill early, and it arrives before the exit. ## Shield AI (USA) URL: https://startup.talk/usa/shield-ai/ Founders: Brandon Tseng, Ryan Tseng, Andrew Reiter Founded: 2015; Based in: San Diego, California; Sector: Defense technology: AI pilots (Hivemind) and drones (V-BAT, Nova, X-BAT) Brandon Tseng has said he cried in the shower for about half an hour. He was a Naval Academy graduate who had wanted to be a SEAL since he was ten, when he watched Under Siege and The Rock. Then came the news that he had not been given a SEAL slot and was being sent to surface warfare instead. He told the Shawn Ryan Show in October 2025 it was the most soul-crushing day of his life up to that point. He called the surface-ship job a spiritual torment. He was born in Houston and grew up mostly in Seattle from age ten, then finished high school in Florida as his father's corporate jobs moved the family. He was the third of four kids; his father, an electrical engineer, would ask why he had not scored 100% on a test. A grandfather had graduated from the Naval Academy. Tseng did get to the teams eventually, and served seven years in the Navy, with deployments to Afghanistan. He has also said he failed the tread-water test at SEAL training four times, awkward for a former water polo player. The idea for Shield AI is traced to Afghanistan. As David Ignatius wrote in The Washington Post, Tseng's unit took casualties in Uruzgan province after a hostile building was poorly scouted. Send something in first, and make it smart enough to work without a pilot. His older brother Ryan started the company, and Brandon has said Ryan had earlier sold a wireless power company to Qualcomm. In 2015, with Andrew Reiter, they launched in San Diego on about $100,000 from friends and family. A contract from the Defense Innovation Unit came in 2016. Nova, a lidar quadcopter that flies without GPS, went to the Middle East in 2018. Then the founders faced a limit. Brandon has said the quadcopter alone could not support a large business, so the company widened out into Hivemind, its autonomy software, and the V-BAT drone it gained through Martin UAV. A $7.2 million Air Force contract in 2021 came as its valuation passed $1 billion. The costs were real. In April 2024 a V-BAT partially amputated three fingers of a Navy service member, who fully recovered; flights were restricted for months. China put the company on its Unreliable Entities list in April 2025. In March 2025 Gary Steele became chief executive, and Ryan, who had started it all, moved to chief strategic officer. Brandon stayed as president. The money kept climbing: $165 million at $2.3 billion in 2022, $200 million at $2.7 billion in 2023, and $1.5 billion at $12.7 billion in March 2026. Revenue has not been published. Tseng's own account is of a man who was told no at a point he cared about most, cried, and went on to build a company out of the work he did get. ## Devoted Health (USA) URL: https://startup.talk/usa/devoted-health/ Founders: Todd Park, Ed Park Founded: 2017; Based in: Waltham, Massachusetts; Sector: Healthtech / health insurance: Medicare Advantage plans with care delivery After athenahealth went public in 2007, Todd and Ed Park's father told his sons something they had waited a long time to hear. It was all right that neither of them had become what he and their mother wanted: one a doctor, the other a lawyer. The parents had come from Korea in 1969, as the brothers have described it on the a16z podcast, chasing the American dream. Their father studied chemical engineering at the University of Utah, and Todd was born in Salt Lake City in 1973. The father's first real job was at Dow Chemical, where he became one of the company's most decorated engineers, with more patents than almost anyone there. The family landed in Granville, Ohio, among cornfields and apple trees. Todd was a Presidential Scholar in 1990 and went to Harvard, graduating magna cum laude in economics. Ed studied computer science. A doctor and a lawyer, it was not. What almost nobody outside the family knew was where the first money came from. Their father had put away $50,000 for graduate school of his own. When his sons asked for it to start a company, he said yes. On the podcast the brothers recalled his one wish: that they would "fail fast because you're my only two kids." They did not fail fast. At 24, in 1997, Todd co-founded athenahealth with Jonathan Bush. In 2008 he co-founded Castlight Health, and later he helped lead the first HealthCare.gov, reportedly built in 90 days. In 2017 the brothers started Devoted Health. Medicare Advantage lets private plans run seniors' coverage, and Devoted tries to run the plan and the care together. Ed is generally listed as its chief executive, with Todd as executive chairman. By January 2026 it reported 466,000 members in 29 states, up 121% in a year, and a Series F at about $16.1 billion. The price of that speed is murky. One estimate puts 2024 revenue at $3.27 billion, while another says 2025 revenue fell 5% to $2.1 billion. The two do not sit easily together, and no profit figure has been published. A father's savings account is not a term sheet, and he knew what he might be giving up. The thing a founder can take from it is that someone once decided the bet was worth more than the safety, and the founders have been paying that back ever since. ## Faire (USA) URL: https://startup.talk/usa/faire/ Founders: Max Rhodes, Daniele Perito, Marcelo Cortes, Jeff Kolovson Founded: 2017; started as Indigo Fair; Based in: San Francisco, California; Sector: B2B wholesale marketplace connecting independent retailers with brands In Faire's first year, Max Rhodes worked 12-hour days, seven days a week. Later it was six. By the end of that year, he has said, he was not in a great place, and he took time off to recover. He told the story to Sequoia Capital in 2019. The man who wore himself out was not an obvious founder. He studied history at Yale. He worked at the consulting firm Bain, then spent five years at Square, where he became a product lead on Square Cash and Caviar and, he says, learned how to build both a product and a startup. He has said soccer helped him get both jobs, and that having no MBA pushed him to make up for it by reading, six or seven books at once, mostly about how to do his job. Where he grew up did not turn up in what we read. At Square he met Daniele Perito and Marcelo Cortes, who started the company with him; some accounts add Jeff Kolovson. It began as Indigo Fair, aimed at independent shops that buy from small brands through trade shows, catalogs and phone calls. Their first pitch did not land. They believed machine learning could predict what a store would sell. Retailers, Rhodes says, were viscerally offended by the idea that some computer could shop for their store better than they could. So Faire changed what it promised, and sold a better, more enjoyable way to shop. He also admits that his own first attempt at the website was still hideous. He wishes he were a designer. The price of those first months was personal. He and his wife took only a few days off, over the holidays in New Orleans, during that first year. What almost nobody mentions is how fast the retailers' resistance turned. By late 2019 Business Insider reported a $1 billion valuation. By 2022 the figure was about $12.4 billion. Then, in a tender offer in late 2025, the company was valued at about $5.2 billion, less than half the peak. For a founder the lesson is in the first pitch. The thing Rhodes was proudest of, a smarter way to buy, was the thing buyers rejected. The company he built went on only after he heard what they said and changed course, even as it cost him a year of seven-day weeks. ## NinjaOne (USA) URL: https://startup.talk/usa/ninjaone/ Founders: Sal Sferlazza, Chris Matarese Founded: 2013; Based in: Austin, Texas; Sector: IT management / endpoint management and remote monitoring software for IT teams and MSPs Before Sal Sferlazza sold software that keeps office computers running, he and Chris Matarese tried to make a video game. In 2001 they founded Realm Interactive and set out to build Exarch, a follow-up to the old space-trading game Trade Wars. Sferlazza hired the comic artist Joe Madureira to draw it. In 2004 the studio was sold to NCSoft, and the game was later released as Dungeon Runners. It is an odd start for a man who now runs a company in IT management. The rest of the early record is about work and exits. Sferlazza graduated from SUNY Albany in 1996. While still in school he did IT consulting for his uncle in New York City, and after graduating he worked as a software consultant at Accenture. After the game came a chain of start-ups in the plumbing of corporate IT. In 2005 he co-founded Lasso Logic, which built off-site backup and was sold to SonicWall for about $20 million. Next came PacketTrap, a network management company later sold to Quest Software, and then Anchor Networks, a cloud managed service provider sold to eFolder in 2013. That year he and Matarese started NinjaRMM in San Francisco. The sources do not describe what it cost the two men to start again, or how they felt about it. The product was monitoring and patching software so an IT team can see and fix every computer from one screen. In 2020 Summit Partners invested $30 million for a minority stake, and the company said it remained independent and founder-led. In 2021 it moved to Austin and became NinjaOne. Then it accelerated. In February 2024 it raised $231 million at a $1.9 billion valuation, led by ICONIQ. A January 2025 agreement to buy Dropsuite for $252 million closed that June at $270 million. A $5 billion valuation followed in 2025, and later a $400 million round at $12.3 billion. A co-founder told CRN the company had passed $600 million in annual recurring revenue from more than 40,000 customers. Channel Dive ranked it third in its market, with 14.6% of 2025 revenue. It still has hard days. In May 2026 NinjaOne Backup was disrupted by a Windows update, the kind of failure that reaches many customers at once. Profit has not been published. The lesson is in the pattern. A game studio, a backup firm, a network tool, a cloud provider: each was a try that ended in a sale, and the company that is worth the most now was the fifth. Some founders get there by sticking with one idea. These two got there by starting again. ## Cyera (USA) URL: https://startup.talk/usa/cyera/ Founders: Yotam Segev, Tamar Bar-Ilan, Yonatan Itay Founded: 2021; Based in: New York City, New York (R&D center in Tel Aviv, Israel); Sector: Cybersecurity, data security posture management (DSPM) and AI data security Less than a year after founding Cyera, Yotam Segev stood in front of a Fortune 500 security team in the Carolinas and showed them a demo. He remembers the chief information security officer's excited reaction as one of his favorite moments, he told Pulse 2.0. It was the first sign that a company built from nothing could say something a large buyer wanted to hear. The idea started in the Israeli military. Segev attended the IDF's Talpiot leadership academy, where he met Tamar Bar-Ilan more than a decade ago, and served in Unit 8200, where he saw security teams struggle to know where their sensitive data lived. He later led teams that deployed and ran cybersecurity technology in the field. They founded Cyera in 2021 with Yonatan Itay, who ran engineering. It came out of stealth in March 2022 with a $60 million Series A. Segev has pointed to the economic climate as a hard thing for an early company creating a new category. Companies had spent a decade putting data into cloud warehouses and SaaS apps, and few could answer a basic question: where is the sensitive stuff? Cyera's software finds it, labels it and flags what is exposed. The money came in large steps. Accel led a round in 2023. Coatue led $300 million in April 2024, and Accel and Sapphire put in another $300 million that November. In June 2025 came $540 million at a $6 billion valuation, and Blackstone led $400 million in January 2026. In June 2026 a $600 million Series G, led by Evolution Equity Partners with Cyberstarts and Temasek, priced it at $12 billion. A September extension from Goldman Sachs Alternatives followed. Total funding exceeds $2.7 billion. It also started buying: Trail Security for about $162 million in October 2024, Shape AI and Otterize in 2025, and in July 2026 Oasis Security for about $1 billion, completed in September and renamed Cyera Identity. The skeptic's questions are fair. Revenue is not disclosed in anything read here, and bigger platform vendors sell overlapping tools. And a $1 billion acquisition is hard to digest at five years old. What stands out is a founder who still singles out one customer's excitement in a Carolina meeting room. A first-time founder can take that: before the valuation, someone has to lean forward. ## Thinking Machines Lab (USA) URL: https://startup.talk/usa/thinking-machines-lab/ Founders: Mira Murati, John Schulman, Barret Zoph Founded: 2025 (February); Based in: San Francisco, California (moved to 2300 Harrison Street, Aug 2026); Sector: Artificial intelligence, frontier models and fine-tuning tools At 16, Mira Murati left Albania for an island off the coast of British Columbia. She had won a Davis United World Colleges scholarship to Pearson College, a school that gathers teenagers from around the world, and she went. She was born in Vlore in December 1988, in the last years of the country's communist regime. What almost nobody pictures when they hear her name is how her house was run. Her parents were both high school literature teachers, and they encouraged her studies. Her own taste ran to maths and science. She has said she had an organic interest in them, and she competed in Olympiads through school. After Pearson came Colby College, where she took a BA in mathematics in 2011, then a mechanical engineering degree at Dartmouth in 2012. A summer at Goldman Sachs in Tokyo, a spell at Zodiac Aerospace, and in 2013 a job as a product manager on Tesla's Model X. Leap Motion followed from 2016 to 2018. Then OpenAI, where by 2022 she was chief technology officer, overseeing ChatGPT, DALL-E, Codex and Sora. In November 2023 she held the chief executive job for about three days after the board removed Sam Altman. Altman returned and she went back to her old role. In September 2024 she left, writing that there was never an ideal time to step away from a place one cherishes, yet this moment felt right. She founded Thinking Machines Lab in February 2025. Within five months it had raised about $2 billion at a $12 billion valuation, and Albania's government added $10 million, which forced the country to amend its 2025 budget. Reports said Meta tried to hire away her researchers with packages of up to $1.5 billion; the same reports said none left. Then they did. In January 2026, cofounders Barret Zoph and Luke Metz went back to OpenAI and Andrew Tulloch left for Meta. A reported $50 billion funding ask faded. In July 2026 the lab shipped Inkling, a 975-billion-parameter model under an Apache license, though it was partly built on Chinese open-weight models. The sources are thin on how any of this felt, and it would be wrong to guess. What they do show is a girl from a family of literature teachers who kept choosing to leave a place she was good at. A founder can learn that the people you hire can leave in the same way, and the only lasting answer is the work they came to do. ## OpenEvidence (USA) URL: https://startup.talk/usa/openevidence/ Founders: Daniel Nadler, Zachary (Zack) Ziegler Founded: 2022; Based in: Miami, Florida; Sector: Healthcare AI, medical search and clinical decision support for doctors Daniel Nadler's father spent his working life listening for trouble. He was an engineer who used sound to find microscopic cracks in bridges and submarines, flaws no eye could see until something failed. His parents had come from Poland and Romania, and Daniel was born in Toronto in 1983. He went to Harvard to study mathematics and classics, and he sat in poetry classes taught by the Pulitzer Prize winner Jorie Graham. It was an unusual double life for a future founder: equations by day, ancient verse after. In 2013, still a doctoral student, he co-founded Kensho with Peter Kruskall, building machine learning for Wall Street. He was also finishing a Harvard PhD on how credit derivatives are priced, and he spent time as a visiting scholar at the Federal Reserve. In 2016 Farrar, Straus and Giroux published his first poetry collection, Lacunae: 100 Imagined Ancient Love Poems, which the Washington Post called a striking debut. What the profiles rarely connect is the detail about his father. The sources I could read say nothing about what Nadler felt about it, and I will not invent it. But a man who spent his life finding hidden faults before they became disasters raised a son who sold his first company, in 2018, to S&P Global for $550 million by one account and about $700 million by NBC's. Then he went looking for another place where something was quietly failing. Doctors, he saw, were drowning in studies they had no time to read. In 2021 he founded OpenEvidence with Zachary Ziegler. A physician asks a clinical question in plain English and gets an answer with citations. It is free to verified doctors and paid for by advertising, and it signed content deals with the AMA, the New England Journal of Medicine and the JAMA Network. The growth was steep. A $75 million round at $1 billion in February 2025. $3.5 billion by July, when 430,000 US physicians, about 40% of the country's doctors, were registered. $12 billion in January 2026 and a reported $15 billion in September. Forbes put his net worth at $7.6 billion in 2026. Not everything is settled. In June 2025 the company accused a competitor of hacking its prompts. Revenue is undisclosed, and an advertising-funded tool that advises doctors invites questions about influence. A founder can take one thing from Nadler: find the person who has the daily pain and the least time, and charge somebody else. ## Biosplice Therapeutics (USA) URL: https://startup.talk/usa/biosplice-therapeutics/ Founders: Osman Kibar Founded: 2008; Based in: San Diego, California; Sector: Biotech, small-molecule drugs for osteoarthritis and cancer (Wnt pathway, alternative splicing) Osman Kibar says he still counts all 70 of his primary school classmates in Izmir as friends. Gazi Pasa school put about 70 children in one class, and decades later many of the senior people around him are friends from his next school, Robert College in Istanbul. His grandfather, also named Osman Kibar, had been mayor of Izmir. His father was an economist. When it came time to choose a university abroad, Kibar says he looked for a climate like Izmir's and picked California: mathematics and economics at Pomona, electrical engineering at Caltech, then a doctorate in biophotonics at UC San Diego. What almost nobody mentions is what he did while he was still a student. He told the Turkish paper Sozcu in 2019 that he founded two companies during the PhD. One, a cancer-diagnostics firm, was Genoptix. Novartis agreed to buy it for $476 million on 24 January 2011. The other made antennas for cell phones, and it was sold in pieces, partly to a military technology firm and partly to Intel. In between he worked in New York at the hedge fund sponsor Pequot Capital, and in 2006 he won the first poker tournament he ever entered. He had money and could have stopped. In August 2011 he went back to San Diego and started again, first as Wintherix, with $3.5 million from his friend Cevdet Samikoglu. It became Samumed, and later Biosplice. The name came from a Japanese Zen term for meditation in activity, plus the word for medicine. Its bet was the Wnt signaling pathway, a cell-communication system tied to tissue growth. The money got large. A $438 million round in August 2018, then $120 million in April 2021 led by Eventide at a reported $12 billion valuation. Forbes put his net worth at $2.9 billion in November 2020. He has said his own spending is small. At one summit he arrived with a single pair of trousers and a T-shirt borrowed from his brother-in-law, and he stops work at 6 PM for two hours with his children, no phone. The cost is time. For years the company had no approved medicine. A New Drug Application for lorecivivint, a knee osteoarthritis treatment, was announced in January 2026, and no decision had been reported by early November 2026. Kibar has said Silicon Valley founders tend to credit themselves for wins and blame others for losses, which stops them learning. The takeaway is plain: a valuation is a number on a date, and the useful habit is keeping an honest account of what went wrong. ## GoodLeap (USA) URL: https://startup.talk/usa/goodleap/ Founders: Hayes Barnard, Jason Walker, Matt Dawson Founded: 2003 (as Paramount Equity Mortgage; renamed Loanpal 2017, GoodLeap June 2021); Based in: Roseville, California; Sector: Fintech, financing for residential solar, batteries and home improvement In his freshman year at Central Missouri State, Hayes Barnard got hurt, and the football scholarship that was supposed to carry him went with the injury. He transferred to the University of Missouri and finished a degree in management and marketing. Football had been the way out. He grew up in Creve Coeur, a St. Louis suburb, with a single mother. His father left when he was three. What he rarely led with was school. In a video interview summarized online, Barnard says he flunked first grade and spent much of his childhood in the resource room, with dysgraphia and a form of dyslexia that were poorly understood and easy to mistake for low intelligence. The University of Missouri, in a 2024 profile, confirms he struggled with a learning disability in elementary school and says the early obstacles fed his drive to improve his family's circumstances. The same summary says a gym teacher, Ron Edwards, became an early mentor. In 1995 he moved to San Francisco in the middle of the tech boom. His first jobs were working trade-show booths. He later became a sales executive at Oracle. In September 2003 he started Paramount Equity Mortgage, one of the first to offer home loans online. It nearly ended in 2008. Washington state regulators sought to revoke the company's license and fine it $500,000 for alleged deceptive lending. It settled in 2009, paying $253,366 and providing restitution to customers. The 2008 crisis also forced layoffs, which the summary puts in the hundreds. He started Paramount Solar in 2008, Guthy-Renker invested in 2011, and in 2013 SolarCity bought it for about $120 million. In 2016 he left SolarCity to lead Loanpal, later GoodLeap. Installers could sell a system, but the customer needed financing fast. By 2019 it had about 30% of new US residential solar loans. Two $800 million rounds in 2021 followed, at a reported $12 billion valuation. The cost came as lawsuits. By 2024, dozens of cases and arbitrations from customers said loan terms were deceptive or loans unauthorized, and Minnesota's Attorney General sued on 8 March 2024. By 2025 it had financed over $30 billion for more than a million homeowners. The summary also says he reunited with his father after decades apart. For a founder the lesson is quiet: what looks like a handicap in a classroom can become a habit of working twice as hard when it gets tough, a point Barnard himself puts as don't quit. ## Whatnot (USA) URL: https://startup.talk/usa/whatnot/ Founders: Grant LaFontaine, Logan Head Founded: 2019; Based in: Marina del Rey, California (remote-first, with hubs in the U.S., U.K., Ireland, Poland and Germany); Sector: Live-shopping marketplace (collectibles, trading cards, fashion and 80+ newer categories) The first card Grant LaFontaine remembers selling was a holographic Chansey. He grew up in Maine, trading Pokemon cards with friends, and by fourth grade he had opened an eBay store and was shipping cards across the country. He has told interviewers he started selling online at seven. Either way, the habit was set long before anyone called it a business. He went on to Facebook, where he worked on the Oculus app store, and to a startup of his own, Kit.com. With Logan Head, another founder, he started Whatnot in 2019, a marketplace for Funko Pops. It did not work. LaFontaine has called the first product something really bad that could not make money. Some investors passed, and it took about ten months to find something real. Then the pitching began in earnest. What almost nobody saw was the spreadsheet. The two founders kept a log of every investor who turned them down, with the feedback in a row of its own. By the time it reached 100 rows, LaFontaine told Fortune in August 2026, they were too demoralized to keep counting. They had pitched hundreds. Nobody had told them yes in the way they needed. So in 2020 they tried something else: live video. Head had never built video and was skeptical it would work. He shut himself in his room, coding long days, about six weeks by the founders' 2022 account and four or five by their 2026 one. LaFontaine called it mentally exhausting. When Head finished, he was nervous the first livestream auction would flop. LaFontaine went live himself. In two and a half hours, late in the evening, he sold about $5,000 of Funko Pops. The first outside seller, a husband-and-wife team called Funko Whisper, later streamed their own wedding. The numbers that followed are large. Sellers moved about $3 billion of goods in live sales in 2024 and over $8 billion in 2025. Whatnot raised $225 million at $11.5 billion in October 2025, then $545 million at $20 billion in August 2026. Bloomberg put each founder's stake above $2.8 billion. There is a bill attached. The Wall Street Journal reported on buyers who ran up debts in rapid auctions, and in March 2026 arbitration claims accused the company of running an illegal gambling scheme through randomized breaks, which it rejects. It has added optional spending limits. The boy with the Chansey became the man who sold the first $5,000 himself. When the spreadsheet stopped being bearable, he did not make it longer. He changed the product, and then he went on camera and was the first customer. ## Bilt Rewards (USA) URL: https://startup.talk/usa/bilt-rewards/ Founders: Ankur Jain Founded: ; Based in: New York, New York (Meatpacking District); Sector: Fintech / loyalty rewards (points on rent, mortgage and housing payments; credit cards) The shoes cost $60. Ankur Jain was nine or ten, and he wanted them. His father, Naveen Jain, who grew up in a very poor Indian village, would not buy them. Ankur has recalled his father asking how he could possibly spend sixty dollars on a pair of basketball shoes. So Ankur did something else. He built a website, then pitched a shoe company to send him free pairs in exchange for ads. He came home with the shoes. His parents did not punish him. They congratulated him. Years later, in a Fortune interview, he said his father had threatened to make him pay for private school himself if he could earn his own money. Naveen, for his part, said the aim was for his children to earn privileges and to spend each summer learning something new. He was born around 1990 in Bellevue, Washington, and raised in Redmond. His parents, Anu and Naveen, were both technology entrepreneurs. He joined his father's road shows at six and started his first company at eleven. At Wharton he started the Kairos Society, an incubator for young entrepreneurs, in 2008. In 2012 he founded Humin, an address book that sorted contacts by context. It raised $15 million and was sold to Tinder in 2016. Jain stayed on as vice president of product and left in May 2017 to return to Kairos. What almost nobody connects to the shoes is Bilt. In 2017, Jain has said, he looked at San Francisco startups funding electric scooters while residents struggled with rent, health care and student loans. He called Wharton a factory for bankers and consultants, and argued the best companies start with a founder's own frustrating experience. Rent, about 40 percent of many people's income, built no credit when paid on time. He called that one of the most infuriating experiences for a customer. A profile summary says he and Starwood Capital founder Barry Sternlicht talked it through in 2018. Bilt launched in 2021, and its first card, with Wells Fargo, came in October 2022. The money rose fast: $200 million at $3.1 billion in January 2024, then $250 million at $10.75 billion in July 2025. About 5.5 million members had signed up by 2026. Forbes put Jain's net worth at $3.4 billion in August 2025. Then Wells Fargo left. Bilt relaunched its cards in January 2026 with new issuers, and two days after launch Jain revised the housing rewards because members were confused. When Jain was a boy, a no from his father became a pitch to a shoe company. The rent pitch is the same move at scale: take what people already pay for, and get someone else to fund the reward. ## Colossal Biosciences (USA) URL: https://startup.talk/usa/colossal-biosciences/ Founders: Ben Lamm, George Church Founded: 2021 (launched Sep 13 2021); Based in: Dallas, Texas; Sector: Biotech / genetic engineering (de-extinction and conservation) Ben Lamm left Austin at five. His parents had divorced, and he was raised mostly by his mother and his grandparents, whom he called Gigi and Poppy, near Dallas. On their farm he caught frogs. He has said, on the Armchair Expert podcast, that he feels blessed by his childhood. He also described a quieter fact, relayed through his brother: their father's addiction meant weekends and summers with grandparents who had the time and the resources. College was handled by someone else. An uncle on Baylor's board offered to pay if Ben chose finance and accounting. Lamm joked on the same podcast that the uncle never realized he was massively unemployable. He was not wrong about that. As a senior at Baylor he co-founded an e-learning company, Simply Interactive, with a cognitive psychologist, and taught himself Flash to build it. He has been frank about the early output: we made a lot of bad games, he said, and one good one. He sold the company in 2010. The pattern held. Chaotic Moon Studios, co-founded in 2010, went to Accenture in 2015. A gaming company went to Zynga in 2016, a bot company to LivePerson in 2018, and Hypergiant to Trive Capital in 2023. What almost nobody notes is that the idea was not his. In a 2008 interview the Harvard geneticist George Church first voiced interest in a mammoth-elephant hybrid. In 2019 Lamm read press coverage and contacted him. After the first call he left inspired, he has said, and stayed up that night thinking about it. He called Church back the next day. Colossal launched on September 13, 2021, with $15 million in seed money led by Thomas Tull. At launch Lamm said the company would bring back the woolly mammoth within a decade; the target is now a calf by 2028. A $150 million round in 2023 valued it above $1 billion, and $200 million in January 2025 valued it at $10.2 billion. In July 2026 Axios reported talks at $20 to $30 billion. The price was credibility. In April 2025 the company announced three gene-edited pups as dire wolves. The IUCN's canid specialist group said they were not dire wolves, and chief science officer Beth Shapiro described them as grey wolves with 20 edits. New Scientist reported in July 2025 that critics had been hit by AI-generated posts and bogus copyright strikes; Colossal denied any role. Lamm grew up with grandparents who had the time. He now sells people a promise that will take years, and asks them to wait. ## Alchemy (USA) URL: https://startup.talk/usa/alchemy/ Founders: Nikil Viswanathan, Joe Lau Founded: 2017; Based in: San Francisco, California; Sector: Blockchain / Web3 developer infrastructure (APIs, nodes, tools) Nikil Viswanathan has said that Alchemy was actually his team's sixteenth pivot. By his own telling, they had spent about nine years on startups with little to show before 2021. One of them nearly made it. Down To Lunch, a social app he co-founded around 2016, began as a Sunday afternoon project. The first version was built in a single night, with deliberate technical debt so it could ship at once. It solved a personal itch: meeting friends spontaneously. Social media reviews spread it. Then, in under 48 hours, the app reportedly lost 90 percent of its growth. The team also dressed up as Berkeley students to get feedback and recoded around what they heard. He grew up in a small Texas town few people have heard of, he has said in a Stanford talk. Other profiles say Lubbock and Chicago for birth, which should be treated cautiously. He studied computer science at Stanford, reportedly for both degrees, and there met Joe Lau, with whom he taught an online database course. Alchemy's own company page says Lau holds Stanford bachelor's and master's degrees and worked at Pinterest, Google and LinkedIn. By 2017 the two had a new angle. Ethereum had shown what Web3 could be, but building on it was painful, Greylock says. A developer had to run servers just to read the chain. Alchemy offered to run them. The pitch became the AWS for blockchains. What almost nobody saw coming was the speed. After years of near-misses, the 2021 boom sent builders to Alchemy. $80 million at a $505 million valuation in April 2021, $250 million at $3.5 billion in October, then $200 million at $10.2 billion in February 2022. A Stanford summary of his 2023 talk says it went from public launch to a $10 billion valuation in 16 months. Adobe and OpenSea were customers. The cost was a market cycle. The $10.2 billion valuation is a high-water mark, now about 33 months old. In 2026 the company hired an AWS veteran, Bill Platt, as COO, and headcount stood at 384. The sources are thin on the personal side: no interview read here describes a family reaction, debt or a night of doubt. What they do record is a count. Sixteen versions, one that took off and then slid in two days, and a partner from Stanford who stayed. For a first-time founder, the number to notice is not ten billion. It is sixteen, and the fortune arrived to people who were still there. ## WHOOP (USA) URL: https://startup.talk/usa/whoop/ Founders: Will Ahmed, John Capodilupo, Aurelian Nicolae Founded: 2012; Based in: Boston, Massachusetts; Sector: Wearable health and performance tech (subscription fitness tracker, recovery and sleep analytics, blood testing) Will Ahmed was watching a basketball game when he saw LeBron James wearing a WHOOP strap. He was, as the Boston Globe later put it, smarting from hundreds of investor rejections. The company was showing no meaningful revenue, the first strap cost $500, and over the next eighteen months not many sold. It had started with a fall. As captain of the Harvard squash team he kept getting fitter, and then, all of a sudden, he fell off a cliff. He has described this himself: he overtrained, and he did not know why. He was born on Long Island, an only child. His father came from Egypt and worked in financial services; his mother is American. Ahmed went to St. Paul's School in New Hampshire, then Harvard, where he studied government. Nothing about it pointed to engineering. He has said he really did not know he wanted to be an entrepreneur. The question that would not leave him was how to avoid being run down in a season. He read hundreds of medical papers. He was nervous about measuring the human body, he has said, so he did extensive research first. In 2012 he founded WHOOP at Harvard Innovation Labs with other Harvard students. What almost nobody remembers is who paid for the start. Ahmed raised about $300,000 from family and friends, and roughly $700,000 more from angels. Fundraising was nearly constant in the first twelve months, and you get rejected, he has said, a lot more than you ever have. Many investors urged him to drop the hardware and go software only. He refused. The team made a modular charger so the band could be worn while charging. The first device did not ship until 2015. He chose his first buyers carefully: elite teams and athletes, to create a halo. In 2020 a golfer, Nick Watney, noticed a spike in respiratory rate and tested positive for COVID-19 without symptoms. $100 million at $1.2 billion came in 2020, and $200 million at $3.6 billion in 2021. The price of loyalty came later. In May 2025, upgrade fees angered members and the company reversed course within days. The FDA fought it that year over its blood pressure feature. In March 2026 it raised $575 million at $10.1 billion, reported 2.5 million members and a $1.1 billion bookings run rate exiting 2025, and Ahmed said the round was expected to be the last private one. The cliff he fell off at Harvard became the product. The money from his family and friends carried a company through years of no. The lesson is plain: when investors tell you to drop the hard part, the hard part may be the company. ## Gusto (USA) URL: https://startup.talk/usa/gusto/ Founders: Joshua Reeves, Edward Kim, Tomer London Founded: 2011 (as ZenPayroll; public launch Dec 11, 2012); Based in: San Francisco, California; Sector: Payroll, benefits and HR software for small businesses Somewhere over the country, Edward Kim watched his mother run her business's payroll on a plane. She was using software he had helped build. He has called it a mind-blown moment. To understand why it hit him, go back to Los Angeles. His mother, Sheena, a South Korean immigrant, ran the front desk of his father's small medical practice. When the family started selling an improved diabetic lancet online, Edward, still a boy, wrote a program with a tool called Endicia that weighed the packages and printed the postage. 'He made me a software with Endicia,' she later told Fortune. Hold on to that picture, because his partners lived the same one. Tomer London grew up in Israel and worked in his father's clothing store, where he wrote inventory software and then carried it door to door to shops in Haifa. His mother, Irit, says he sold plenty of software before Gusto. Josh Reeves grew up in Marin County as an Eagle Scout; his mother, Eliana, came from Bolivia, and her favorite story is of him hauling wooden steps up a Muir Woods hill with park rangers. The steps are still there. So the three men who started ZenPayroll in 2011 were each the child of a small business. They came through Y Combinator's Winter 2012 batch and launched on December 11, 2012, with a $6.1 million seed that included Box's Aaron Levie and David Sacks. What stands out is what they had watched: parents with no spare hours for paperwork. What almost nobody notices is how the three families ended up. In 2014 the mothers met for the first time in Haifa, at Tomer's wedding, and spent three days together on a bus tour of Israel. They have since traveled together to Tenerife and Disneyland. The company grew the way those households would have wanted: simply. By August 2013 it said it moved more than $100 million a year. In September 2015 it dropped the Zen name for Gusto and added health benefits, and that December it hit a $1 billion valuation. A June 2025 employee tender valued it at about $9.3 billion, below the $10 billion-plus figure quoted from 2021. In 2026 Gusto passed $1 billion in trailing revenue, which Reeves stressed was cash collected, and it has been cash-flow positive for years. The cost is harder to find in the record than the reward; no one has said what the early years took from the founders. What is documented is the thing they built from. A founder can take one idea from this: the customer was never an abstraction to these three. They had sat at that dinner table, and they built the product their mothers would approve of. ## Mercor (USA) URL: https://startup.talk/usa/mercor/ Founders: Brendan Foody, Adarsh Hiremath, Surya Midha Founded: 2023; Based in: San Francisco, California; Sector: AI training data: expert contractor marketplace for AI labs The first customer paid $500 a week for one developer. Of that, the engineer took about 70 percent, and Mercor kept the rest. That was the whole business. Brendan Foody had decided something before he had a company. 'I knew I wanted to drop out before finals my sophomore year,' he told Fortune about leaving Georgetown in the spring of 2023. 'I just didn't go to finals.' He has said he has not taken a day off in three years, and that he thinks about work even at the dinner table with his parents. Where he was born, and what his parents do, are not in anything I could verify, so the want behind the hurry stays unnamed. His partners were old friends. Adarsh Hiremath and Surya Midha were his debate teammates at Bellarmine College Preparatory in the Bay Area. All three became Thiel Fellows, and reports say Hiremath left Harvard and Midha left Georgetown. They met again at a hackathon in Sao Paulo in early 2023, where they built the model: match companies with overseas engineers for a cut of each deal. The idea was narrow, aimed at freelance programmers in India and an AI interview that vetted them. The stakes were small and the margins thin. Then the market shifted under them. AI labs needed experts, including lawyers, doctors and bankers, to write and grade the examples that train chatbots. Mercor pointed its vetting machine at those people. By October 2025 it said it managed about 30,000 contractors, with OpenAI and Anthropic among its customers. What almost nobody remembers is how small the start was. Five hundred dollars a week is rent for a shared apartment, not a unicorn. The jump from there to a $10 billion valuation, raised as a $350 million round in October 2025, took about two and a half years. The founders were described as the youngest self-made billionaires. In 2026 the CEO said annualized revenue passed $2 billion, a self-reported figure, with most of it paid out to contractors. The price arrived in March 2026. A supply-chain attack through the LiteLLM package reportedly put up to 4 terabytes of data at risk, including contractor personal information. Class actions followed and Meta paused work. Contractors told The Verge, New York Magazine and Wired of stress and shrinking pay. Reports in July said talks were underway at about a $20 billion valuation. A first-time founder can take this from it: Mercor did not need to invent a market. It needed one customer willing to pay, and the sense to notice when a bigger one was knocking. ## Notion Labs (USA) URL: https://startup.talk/usa/notion-labs/ Founders: Ivan Zhao, Simon Last, Akshay Kothari, Chris Prucha, Jessica Lam, Toby Schachman Founded: 2013; Based in: San Francisco, California; Sector: Productivity and collaboration software, AI workspace In the summer of 2015, Ivan Zhao and Simon Last were in a rented house in Kyoto, coding 18 hours a day. They had closed their San Francisco office to stretch the money they had left. The first Notion product had not worked. Zhao has said he built it for what he wanted rather than for users, and had learned that people do not want to build apps. So he and Last scrapped it, let their small team go, and started again. Zhao says in Notion's own retrospective that the company moved to Kyoto after it nearly failed in 2015 and rebuilt the application. Zhao did not start out in Silicon Valley. He has described growing up in a small town in China before moving to Canada. One profile says his mother moved the family to Vancouver while he was in high school, for his education; another names Urumqi as the town. Neither is confirmed in a primary source we could read. He studied cognitive science and fine arts at the University of British Columbia, and has said he did not study computer science. He is a designer by instinct, drawn to making tools. Notion Labs was founded in 2013 in San Francisco, with Zhao, Akshay Kothari, Chris Prucha, Jessica Lam, Simon Last and Toby Schachman listed as founders. It had a seed of about two million dollars and, as the record notes, early money trouble. What the outside world never saw was how small it was. Notion 2.0 shipped in March 2018 with fewer than ten employees. The rebuild worked. Notion hit a million users in September 2019. In January 2020 Index Ventures led $50 million at a $2 billion valuation, and by October 2021, with 20 million users, Coatue and Sequoia invested $275 million at $10 billion. Success had its own near miss. During Covid the product's single database nearly ran out of space, with only weeks to spare, Zhao said on Lenny's Podcast. The design that let Notion move fast had left it with one point of failure. In January 2026 about $270 million of employee shares were sold in a tender at an $11 billion valuation. Third-party estimates put annual revenue at about $500 million in 2025. Not everything stayed: Notion Mail launched in April 2025 and was set to shut down on September 22, 2026. Zhao's advice to founders, on the same podcast, is short: reset and start again if you need to. The house in Kyoto is the proof he meant it. ## Ripple (USA) URL: https://startup.talk/usa/ripple/ Founders: Chris Larsen, Jed McCaleb, Arthur Britto, Brad Garlinghouse Founded: 2012 (OpenCoin; renamed Ripple Labs Sept 26, 2013); Based in: San Francisco, California; Sector: Crypto and cross-border payments, XRP Ledger, stablecoin (RLUSD) Chris Larsen was the first in his family to go to college. His father was an aircraft mechanic, his mother a freelance illustrator, and the boy born in San Francisco in 1960 first planned to study aeronautics at San Jose State. Then he found business, moved to San Francisco State, and picked accounting because he thought it was practical. He graduated in 1984 and took a job auditing Chevron operations in Brazil, Ecuador and Indonesia. Later he went to Stanford's business school, where, he has said, entrepreneur Jim Collins urged students to 'cut the lifeboats.' Larsen did. After his MBA in 1991 he co-founded the online lender E-Loan, which went public in 1999, then Prosper, a peer-to-peer lender. What almost nobody remembers is how personal that second company was. By 2008 he had funded 450 loans on Prosper with his own money, backing the idea that strangers could lend to strangers. For a mechanic's son who had picked accounting for safety, it was a strange place to put his cash. Ripple's own start was more tangled. In 2011 David Schwartz, Jed McCaleb and Arthur Britto began work on a ledger, and McCaleb opened a BitcoinTalk thread called 'Bitcoin without mining.' In September 2012 McCaleb, Britto and Larsen founded OpenCoin, which became Ripple Labs on September 26, 2013. One early decision shaped the next decade: 100 billion XRP tokens were created at inception, and the founders gave 80 billion to the company. That supply paid the bills. It also drew the regulator. The trouble came in stages. In 2015 Ripple paid a $700,000 FinCEN penalty as an unregistered money services business. In December 2020 the SEC sued over XRP sales. A July 2023 ruling found institutional sales were unregistered securities offerings but exchange sales were not, a $125 million penalty followed in August 2024, and both sides dropped appeals on August 7, 2025. Forbes had called the company a 'crypto zombie' in 2024. After that it moved fast, buying Rail for $200 million, Hidden Road for $1.25 billion and GTreasury for about $1 billion. In November 2025 it raised $500 million from Fortress, Citadel Securities and others at a $40 billion valuation, its first outside raise since 2019. Larsen's wealth followed XRP up and down. Forbes put it at $59 billion in January 2018; by 2020 he was 319th on the Forbes 400, and he and Ripple gave $1 million each to five San Francisco food banks. Larsen picked the practical major and then spent his career betting on the impractical one. What a founder takes from it: the choice that looks safe at twenty can still teach you how to stomach risk at fifty. ## Digital Currency Group (USA) URL: https://startup.talk/usa/digital-currency-group/ Founders: Barry Silbert Founded: 2015; Based in: Stamford, Connecticut (moved from Manhattan, Nov 2021); Sector: Crypto investment and holding company (Grayscale, Foundry, Luno, Yuma) As a kid in Gaithersburg, Maryland, Barry Silbert traded baseball cards. A Stanford case study on his company says that hobby taught him more about illiquid assets and inefficient markets than any job he held in banking. He was born there in 1976 and told a House committee in 2011 that he was born and raised in the town. He earned a business degree at Emory's Goizueta school in 1998, then spent about five and a half years as an investment banker. By his own account it was a good experience and not enough. 'I really wanted to go try to build something,' he told Emory Business. His parents and childhood do not appear in the interviews and profiles I could read, so the personal layer is thin; the documented through-line is the card trader who grew into a market maker. In 2004 he founded SecondMarket, a marketplace for shares in private companies. It was valued at $200 million in 2011 and sold to Nasdaq in 2015. He ran it for nearly eleven years, and Fortune named him to its 40 Under 40 in 2013. What almost nobody connects is that he began buying bitcoin in 2012, years before most of his peers took it seriously. Digital Currency Group followed in 2015. It grew by owning pieces of the industry: Grayscale's bitcoin trust, the lender Genesis, the mining firm Foundry, the news site CoinDesk. By November 2021 it had made more than 200 investments, moved to Stamford, Connecticut, and a $700 million secondary sale valued it at $10 billion, with Silbert holding almost 40 percent. Then the structure he built turned on him. Genesis had lent to Three Arrows Capital and Babel Finance, and when they fell, it took heavy losses. DCG took on debts to keep it afloat. On November 16, 2022, Genesis halted withdrawals days after FTX collapsed, and about 10 percent of DCG staff were already gone. In January 2023 Genesis filed for Chapter 11, and in October the New York Attorney General sued DCG, Gemini and Genesis, alleging fraud on more than 230,000 investors. DCG denied the claims. It survived by selling and repairing: CoinDesk went to Bullish for $125 million, Genesis completed a restructuring in August 2024 with about $4 billion returned to creditors, and Grayscale's bitcoin fund became a spot ETF in January 2024. In November 2024 Silbert launched Yuma, an AI company on the Bittensor blockchain. No valuation newer than the 2021 figure was found. A card trader learns early that anything is worth what someone will pay. It is a useful instinct when you are the one holding the lender, the fund and the miner. ## Talkdesk (USA) URL: https://startup.talk/usa/talkdesk/ Founders: Tiago Paiva, Cristina Fonseca Founded: 2011; Based in: San Francisco, California (large engineering base in Portugal); Sector: Cloud contact-center software (CCaaS), AI customer experience Tiago Paiva wanted a laptop. In 2011 a hackathon hosted by Twilio offered a MacBook Air as first prize, and he has said he wanted it and everything it represented. He was a computer science graduate of the University of Lisbon, in a country where the startup scene was still taking shape. He and Cristina Fonseca, a fellow Lisbon student, built a browser-based call center over a weekend. It won. A judge, Paul Singh, invited Paiva to 500 Startups, the accelerator, and he stayed in the United States. Before that batch's Demo Day in November 2011, an investor named Alex Khein put money in. Within weeks Talkdesk had $350,000. What almost nobody counts is how long that cheque had to last. The company did not raise again until 2014, and by then it held about the same cash it had started with. In the meantime the team stretched one meal into three, and Paiva went without a salary for stretches. Fonseca went back to Portugal to build the engineering team while he stayed in San Francisco. For the first two years he was the only person in marketing, sales, customer success and support, while also leading product. And he had one problem. He was reluctant to talk to customers by phone, and the phone was the entire job. He got past it. Early customers were treated personally, and they stayed. They put up with product changes and sent friends. Talkdesk reached $1 million in annual recurring revenue on that first angel money, with no marketing spend. The turn came from engineers in Portugal who answered phones and email around the clock, which let small customers become large ones. In July 2020 the company was valued at $3 billion. In August 2021 a $230 million round put it above $10 billion, with nearly 2,000 employees, most of them in Portugal. The price was uneven. Fonseca left in 2016 and stayed a shareholder. The 2021 number is now more than five years old; no newer priced round has been found, and a secondary-market estimate from May 2026 sits far lower, though that is an estimate, not a deal. Paiva won the laptop. Every new Talkdesk employee still gets a MacBook Air on their first day. A first-time founder can take the plainest lesson from it: an honest want is a better beginning than a plan, and the thing you dread, here a phone call, may be the actual job. ## Vercel (USA) URL: https://startup.talk/usa/vercel/ Founders: Guillermo Rauch Founded: 2015 (as ZEIT; renamed Vercel April 2020); Based in: San Francisco, California; Sector: Developer cloud / frontend platform (Next.js, v0, AI SDK) His father came home with a magazine about free software and a dare. Replace Windows with Linux, he told Guillermo Rauch and his brother. Rauch has said he grew up in a pretty poor area of Argentina, in a suburb outside Buenos Aires where few families owned a computer. Linux was, in his words on the First Round Review podcast, really hard to use. He started with Red Hat, moved to Debian, and found that almost nothing was written in Spanish. So he taught himself English along with the code, because that was where the answers lived. What almost nobody connects to Vercel is where his first friends were. An IRC chat attached to his internet provider gave him a room full of people who knew more than he did, and he learned, he has said, that he could learn together with others. A stranger told him that code could earn money. He says he started his first business at eleven. Open source did the rest. His work on a UI library, MooTools, built a name, and his teenage demos reached the front page of Digg. He has called himself a five-minute internet celebrity. Facebook reached out while he was still in Argentina. "Open source really was my path out of Argentina in the end," he told the podcast. He dropped out of high school and found his way to the Bay Area at around eighteen. He built Cloudup, a file-sharing product, and sold it to Automattic before it had even opened to the public. He has admitted he was a little sad, as if he were letting his baby go. Then, in 2015, came ZEIT, renamed Vercel in 2020. The itch was a memory: his first attempt to put a simple website online had taken weeks, and he kept asking why the cloud had been made so hard. Next.js was the answer, and investors agreed. $250 million at $3.25 billion in May 2024, then $300 million at $9.3 billion in September 2025. Being visible has a price. A selfie with Benjamin Netanyahu in September 2025 led some developers to boycott the company. In April 2026 Vercel disclosed that attackers had reached some internal systems through a third-party AI tool. Rauch still describes the work the way the boy with the magazine would have: take something hard and make it feel simple. What a first-time founder can take from him is quieter than any growth tactic. He had no money and no school to lean on, and he used the one door that was open, a connection to strangers who were willing to teach. ## Tanium (USA) URL: https://startup.talk/usa/tanium/ Founders: Orion Hindawi, David Hindawi Founded: 2007 (July 16 2007); Based in: Kirkland, Washington (operations in Emeryville, California); Sector: Cybersecurity / endpoint management (Autonomous Endpoint Management) Orion Hindawi left the University of California, Berkeley before he graduated, Forbes has reported, to work with his father. Berkeley was his hometown and the school where David Hindawi had earned a PhD. In 2007 the two of them started Tanium. To see why that mattered, go back to Baghdad. David Hindawi was born there in 1944 to an Iraqi-Jewish family. In 1951, when he was about seven, the family moved to Israel. He served in the Israeli Air Force's operations research department, then came to the United States in 1970 and earned his doctorate at Berkeley. Orion was born in Berkeley in 1980. They built things together twice. David started a software company in 1984, sold it in 1995, and founded BigFix in 1997. IBM bought BigFix in 2010 for $400 million. Tanium's idea was that machines could pass questions along a chain instead of sending everything to one data center. It raised $52 million from Andreessen Horowitz in 2015 and $120 million that September. In February 2016 Orion became CEO and his father became executive chairman. A year later the praise turned. What almost nobody remembers is how that story was told. In April 2017 Bloomberg ran a piece headlined 'Tanium's Family Empire Is in Crisis.' Business Insider's summary says it alleged Orion kept a list of staff nearing option vesting and fired some of them, which employees called Orion's List, and that he publicly insulted colleagues. Tanium said it was not a company in crisis and denied the claims. A week later the Wall Street Journal reported that Tanium's software had been demoed on a hospital's live network. The hospital said it had not authorized it; Tanium said it had not protected customer data well enough in demos, though it said no patient data was revealed. The company grew. TPG put in $175 million in 2018, and revenue reached $440.2 million in 2019. In 2023 Orion stepped aside for an outside CEO, Dan Streetman. In August 2026 he returned and replaced Streetman after less than four years. Secondary-market estimates this year put the value near $3.9 to $4 billion; those are estimates, not rounds. The sources do not include a first-hand account from either Hindawi of what the 2017 reporting or the 2023 exit cost them. What is on the record is the pattern. A boy from Berkeley left school to join his father's company, led it, lost the room, left, and came back. The family that left Baghdad seventy-five years ago is still deciding who holds the keys. ## Replit (USA) URL: https://startup.talk/usa/replit/ Founders: Amjad Masad, Faris Masad, Haya Odeh Founded: 2016; Based in: Foster City, California; Sector: AI software creation / cloud development environment (Replit Agent, vibe coding) Amjad Masad did not own a laptop. As a student in Amman he walked to an internet cafe, borrowed a computer, and reinstalled his tools every time he wanted to write code. He told HubSpot's startup magazine that this was the exact frustration that planted the idea of coding in a browser. In the early 2000s he led Counter-Strike teams good enough to reach an international tournament in Turkey, then moved from playing to building game maps and scripts. He studied computer science at Princess Sumaya University for Technology and kept missing class, which under the rules meant automatic failure. What almost nobody knows is what he did about it. He hacked the university's system to change his grades. "I just wanted to see if I could do it and how it worked," he has said. The registration database crashed after one student was marked both banned and passed. He did not get expelled. He was put to work hardening the same systems over the summer, and at his final defense he decrypted a dean's password on screen. He graduated in 2010. With little venture money in Jordan, he moved to New York to join Codecademy on an O-1 visa, the route for people of extraordinary ability, after his open-source in-browser coding tool went viral on Hacker News in 2011. Facebook followed, then Replit in 2016, with his brother Faris and his wife, Haya Odeh. Most investors passed. The company raised about $600,000, barely enough to build the product, before Y Combinator took it in 2018. The low point was not about money. In 2019 he paused the company to be with his mother, who was dying of cancer. He has said Jordanian doctors would not let him into the ICU until he begged and yelled. He told the San Francisco Standard he was so angry at modern medicine. She died. The company he returned to nearly stalled. Growth slowed in 2023, the education product was cut, and 30 people were laid off in 2024. Then the AI agent took off. Replit raised $250 million at $3 billion in September 2025 and $400 million at $9 billion in March 2026. In July 2025 that agent deleted a customer's database during a code freeze. After October 2023 his public statements on Gaza cost him friendships and, reportedly, investors. He did not step back. The student who borrowed machines built a tool so no one else would have to. His own history suggests the lesson: the system you break to prove you can may be the one you are later asked to fix. ## Nuro (USA) URL: https://startup.talk/usa/nuro/ Founders: Dave Ferguson, Jiajun Zhu Founded: 2016; Based in: Mountain View, California; Sector: Autonomous driving (Nuro Driver, licensing; earlier autonomous delivery vehicles) Dave Ferguson was not a computer kid. As a boy he preferred climbing trees, he has said in a profile for Kea New Zealand. His father was a diplomat, so the family lived in the United States for a time, then returned to New Zealand so he could finish high school there. After school he applied to the University of Otago, and in a large first-year class one teacher made the difference. Willem Labuschagne pointed him toward computer science. Ferguson remembers that Labuschagne learned everyone's name in the first year, which he found remarkable. He credits the man with shaping both his research and his approach to life. When Labuschagne died, Ferguson set up a scholarship in his memory. The next stop was Carnegie Mellon's Robotics Institute, which he calls the Disney World of robotics. In his first year there he worked on a robot to map abandoned coal mines, inspired by a mining accident. He later wrote an algorithm that NASA uses for long-range autonomy on its Mars rovers. After Intel, in 2011 he joined Google's self-driving car project as a principal machine learning engineer, working beside Jiajun Zhu, a Fudan and University of Virginia graduate who led perception and simulation. What almost nobody saw coming was how small the first idea was. In 2016 both men left Google, convinced robotics would reshape daily life within 10 to 15 years. They started with a pared-down vehicle that carried groceries rather than people. Money followed fast. $92 million in January 2018, $940 million from SoftBank in 2019 at $2.7 billion, and $600 million at $8.6 billion in November 2021. Kroger, Domino's and CVS signed on, and in 2020 Nuro received the first federal autonomous exemption. Then the cost arrived. In November 2022 Nuro cut about 20 percent of its staff, roughly 300 people. In May 2023 it paused commercial expansion and cut about 30 percent, around 340. In September 2024 it dropped its own vehicle business and began licensing its driving software. In August 2025 it raised $203 million at $6 billion with Uber and Nvidia, below its 2021 peak, and in October 2025 Ferguson became co-CEO alongside Zhu. The company built to carry groceries now sells the driver. The boy who liked trees ended up writing code that runs on Mars, and paying forward to a teacher who learned a stranger's name. A first-time founder can see the debt in that: someone at the start said computer science, and he is still repaying them. ## Flexport (USA) URL: https://startup.talk/usa/flexport/ Founders: Ryan Petersen, David Petersen Founded: 2013 (Customs licence granted Mar 31 2013; joined Y Combinator 2014); Based in: San Francisco, California; Sector: Logistics / freight forwarding software and supply chain platform In the year before business school, Ryan Petersen made about $17,000. He told a Y Combinator interviewer in 2018 that he picked Columbia partly as a backup plan. He left in 2008 owing about $140,000. Born around 1980 in Bethesda, Maryland, he had a government economist for a father and a biochemist mother who ran her own business. At 17 he and his brother David were already buying goods from China to sell on eBay. After an economics degree at UC Berkeley he moved to China in 2005 and spent about two years running a supply chain for imported motorbikes and electric scooters. He has called it a couple of guys trying to earn a living, never a startup. Then came the debt. To pay it down he stitched together flexible work: a Columbia case study on a Nigerian tech company, teaching the GMAT, a small SEO consulting firm. He knew, he said, he would be bad at ordinary MBA jobs. He now tells people to avoid business school if it means borrowing that much. What almost nobody connects to Flexport is how long he waited. The idea began as a side project with no revenue. He needed a licence from US Customs and Border Protection, and it took about three years to arrive: March 31, 2013. Before it came, companies including Foxconn and Saudi Aramco signed up through a marketing page for a service that did not yet exist. By his count it took about five years from first idea to first dollar, four of them as a solo founder. He recalled selling his Amazon stock to buy a house and then watching the stock keep climbing, calling it the dumbest thing he had ever done. Debt payments, rent and food, he said, are what kept him experimenting. Flexport joined Y Combinator in 2014. By early 2022 it raised $935 million at an $8 billion valuation, and revenue reached $2.1 billion in 2024. The price came later. In 2022 he handed the CEO job to former Amazon executive Dave Clark. Clark resigned abruptly in September 2023, and Petersen took the job back and withdrew dozens of job offers. In early 2025 he warned that 145 percent tariffs on Chinese imports could bankrupt many small American importers, the kind of customer he once was. A first-time founder can take one thing from the licence. He did not wait for permission to learn whether anyone wanted the product, and behind those three years of waiting sat a debt, a rent bill and a landing page. ## Zipline (USA) URL: https://startup.talk/usa/zipline/ Founders: Keller Rinaudo Cliffton, Keenan Wyrobek, Ryan Oksenhorn, William Hetzler Founded: 2014 (Half Moon Bay, California; grew out of Romotive, founded 2011); Based in: South San Francisco, California; Sector: Autonomous drone delivery / medical and retail logistics Keller Rinaudo Cliffton was shocked that Harvard let him in. The essay that got him there was about his grandfather, who was dying of cancer. He grew up in Arizona, short and teased for it, at a high school he has called a pretty depressing place, full of gangs and drug dealers. He started a debate magazine and worked in a restaurant stocking a refrigerator. As a teenager, he has said, he had no idea what he wanted to do. At Harvard he built the university's first climbing wall over a summer without waiting for approval. After graduating he left consulting to climb professionally, reached national prominence, and was happy living simply, out of a Honda Fit. He came back to business with Romotive in 2011, a toy that turned an iPhone into a small robot. It sold on Kickstarter and Sequoia backed it. Then he grew uneasy. Home robots, he feared, were gadgets people used for a few months and put in a drawer. What almost nobody knows is what the pivot cost. In 2014 the company laid off half to two-thirds of its staff, a pretty scary time, he told Sequoia. The toy business was shut before anyone knew what would replace it. Alfred Lin, the Sequoia partner, called it a hail mary. Many investors had largely written the company off, Rinaudo Cliffton has said, and others thought a logistics company run by people with no healthcare or aviation experience was extremely stupid. He described those months as chewing glass. A 2014 trip to Tanzania, and his wife, an epidemiologist, pointed him at the real problem: medicine that never reached clinics. Zipline began delivering blood in Rwanda in October 2016. The first year of commercial operation, he said, was extraordinarily painful, with all-nighters and breakdowns, and it took about nine months to get the first hospital working. A few people refused to quit when many had left. The bill still arrives. In November 2025 Zipline suspended three Ghana hubs over roughly $15 million in unpaid government bills. Yet in January 2026 it raised more than $600 million at $7.6 billion and passed two million deliveries. His father once told him to work out what impact he wanted to have before he died. The teenager with no plan learned that you can build the thing people call stupid, and that the layoffs come first. ## AlphaSense (USA) URL: https://startup.talk/usa/alphasense/ Founders: Jack (Jaakko) Kokko, Raj Neervannan Founded: 2011; Based in: New York, New York; Sector: AI market intelligence / enterprise search for financial and corporate research Sweating, barely awake, and afraid of what he had missed. That is how Jack Kokko has described sitting in client boardrooms as a young technology analyst, after working through big piles of paper and sticky notes. He called it a constant nagging fear: that somewhere in the filings nobody could search was the one thing that mattered. He grew up in Finland, and long before banking there were small jobs. He delivered newspapers and cleaned furnaces, and says they taught him to do his best at whatever he was doing. A teaser for a 2025 interview adds that as a boy he experimented with electronics. Profiles say he studied engineering and finance in Finland, went into telecom banking in London, then talked Morgan Stanley into sending him to Menlo Park, where the fear took hold. Asking for help, he has said, does not come easily to him. So he set out to fix the problem himself. At Wharton he met Raj Neervannan, an engineer, and the two worked on a search engine that read financial documents the way an analyst needed them read. What almost nobody sees is how long nothing happened. It took two or three years of development before the beta. Then Kokko and the team gave the product to a few hedge funds and worked with them for close to a year before anyone paid. He was building for people he knew, in a market where Bloomberg already sat on every desk. And then came the call. Getting the first client was terrifying, he told an interviewer in 2016, because he had to make it himself. He is not a salesperson and had never been a cold caller. It had to be done, he said. It got done. By late 2016 AlphaSense had over 500 clients and had just raised $33 million. The price of that slow start was years of building without revenue, and a phone call he dreaded. The company's recurring revenue passed $500 million in October 2025 and $600 million in early 2026. In June 2026 it raised $350 million at a $7.5 billion valuation, nearly double a mark of about $4 billion set after it bought the expert-call firm Tegus. The product began as an answer to one analyst's private fear. The sales call he feared most came only after the tool was good enough to carry it. ## Carta (USA) URL: https://startup.talk/usa/carta/ Founders: Henry Ward, Manu Kumar Founded: 2012 (as eShares, renamed Carta Nov 2017); Based in: San Francisco, California; Sector: Fintech / cap table, equity and valuation software; fund administration Henry Ward could not raise a seed round. In 2011 he had started Secondsight, a company that aimed to bring algorithmic fixed-income trading to retail investors, and the investors he met kept saying no. He has said little in public about where he came from. He studied at the University of Michigan, planning to major in math, and has a degree in mathematics and computer science and a master's in market finance from EDHEC. Before Secondsight, he worked on software for fixed-income traders. His own account is that he is in this for the work, not the money: he has said he would do it for free. That belief was tested by the failed raise. A startup that cannot find seed money cannot grow, and Secondsight did not get there. What almost nobody notices is how Carta started: with a pitch that went nowhere for the person who had come to give it. In that search Ward met Manu Kumar, an investor. Kumar did not fund Secondsight. He suggested something duller, a company to manage stock certificates. In July 2012 the two founded eShares. The pitch was plain: replace paper certificates and Excel spreadsheets with one online registry. The product grew outward from there, into 409A valuations, compliance and eventually fund administration. In 2017 it took the name Carta. In 2021 an auction set a $6.9 billion pre-money value and a $500 million round led by Silver Lake valued it at $7.4 billion. Ward has said it took ten years to reach roughly $500 million in annual recurring revenue. The idea that began after a rejection took a decade to look obvious. The bill for the boom was public. In 2020 a former marketing executive sued, alleging gender discrimination; the claim settled in 2023. A former CTO alleged in 2022 that Ward dismissed reports of abusive behaviour; Carta denied wrongdoing, fired him and sued him, and the allegations are unproven. In January 2023 Ward cut about ten percent of staff, roughly 200 people, telling them that if customers suffer, so does Carta. In 2024 Carta shut CartaX and left secondary trading after a customer data controversy. The $7.4 billion mark is now five years old, and secondary-market estimates ran far lower. Ward's own pitch was turned down, but the investor who heard it left with a different idea. The company that came from the no was not the one he went in to sell. ## Automation Anywhere (USA) URL: https://startup.talk/usa/automation-anywhere/ Founders: Mihir Shukla, Neeti Mehta Shukla, Ankur Kothari, Rushabh Parmani Founded: 2003 (as Tethys Solutions, San Jose; rebranded Automation Anywhere, Inc. in 2010); Based in: San Jose, California; Sector: Enterprise software / robotic process automation and agentic process automation Mihir Shukla stepped out for a coffee. It was 2003, the day he and his wife, Neeti Mehta Shukla, launched their software company from the guest bedroom of their home, and the coffee was meant as the celebration. When he came back, Neeti has said, they already had a customer. The customer was in Australia. Neeti knew Australia well. She took her Bachelor of Commerce and MBA at Bond University there, and she has said she has been coding since the eighth grade. The company began six years after she finished that degree, in a spare room, with her husband and two co-founders, Ankur Kothari and Rushabh Parmani. They called it Tethys Solutions. Mihir came to it from search and supply-chain software, with earlier jobs at Infoseek, the early search engine, and E2Open. The Shuklas have told the story of the company far more often than the story of themselves. The problem they picked was unglamorous. Neeti told Bond University that automation was very difficult for most companies, and that they wanted to hand it to the end user. The product was a software robot that repeated the clicks of an office worker, so a company could automate an old system without rebuilding it. What almost nobody sees is how quiet the early years were. The company kept the name Tethys until 2010, when it became Automation Anywhere. Fifteen years after the bedroom, in 2018, it reportedly raised $550 million from General Atlantic, Goldman Sachs, NEA and SoftBank, and in late 2019 another $290 million led by Salesforce Ventures at about $6.8 billion. That number has not moved. Almost seven years on, no newer round has been found and the company has not listed. The market it helped create filled with rivals such as UiPath and the big cloud vendors, and in 2024 the company recast itself around AI agents. In October 2026 it announced plans to buy conversational AI company Boost.ai, and it says it grew at a double-digit rate in the second quarter of its 2027 fiscal year. The price of a patient company is a valuation frozen in an earlier era, and the Shuklas carry it in public. But the first customer is worth remembering. It arrived before the coffee was finished, from the far side of the world, because the product was something a stranger could use from anywhere. Neeti Mehta Shukla is now the company's Chief Impact Officer, still working on what began in that room. ## DataRobot (USA) URL: https://startup.talk/usa/datarobot/ Founders: Jeremy Achin, Tom de Godoy Founded: 2012; Based in: Boston, Massachusetts; Sector: Enterprise AI and automated machine learning software In March 2021, nine years after he started DataRobot, Jeremy Achin was pushed out as its chief executive by the board. CNBC later reported that the split was partly over his resistance to taking the company public. To understand what he lost, go back to a contest. Around 2011, Achin and Tom de Godoy, who had met as students at the University of Massachusetts about two decades earlier, entered a Kaggle data science competition as a team. They had two weeks to build a solution. They placed sixth. De Godoy later told Vulcan Post: 'It was during the competition that we had the idea for DataRobot.' They kept entering, and won a streak of top-three finishes. Both once ranked as high as 20th on Kaggle. Both were also working data scientists at the insurer Travelers. Achin, who studied math, physics, computer science and statistics at UMass Lowell, is described in one profile as director of research and modeling there. What they saw from inside was a gap. Businesses piled up data, yet few could use it for prediction, because the work needed highly trained specialists. Their plan was to automate the specialist. Getting anyone to listen was harder. The founders have said that pitching investors was a struggle: around 2012, many people still thought of AI as science fiction, and they felt their vision was well ahead of its time. A $3.3 million seed round led by Atlas Venture came in April 2013. The product tried many modelling approaches on its own and handed back the best. It grew. The company bought a rival, Nutonian, in 2017, and by 2022 it had raised about $1.1 billion, reported more than a million AI projects, and signed a five-year, $249 million contract with the U.S. Department of Defense. Then came the board's decision. Dan Wright had taken over as an operating chief in 2020; Achin left in March 2021. Four months later, a $300 million Series G valued DataRobot at $6.3 billion, with Tiger Global and Altimeter back in the round. The other founder, CTO de Godoy, left as well. That figure is now more than five years old. Two people proved an idea in a two-week contest and then built a company on it. Years later they could not keep it. The people who start a machine and the people who decide when it goes public can want different things, and the argument tends to surface just when the company looks strongest. ## The Boring Company (USA) URL: https://startup.talk/usa/the-boring-company/ Founders: Elon Musk, Steve Davis Founded: 2016 (announced Dec 2016; registered Jan 2017); Based in: Bastrop, Texas; Sector: Tunneling and underground transport infrastructure On 17 December 2016, stuck in Los Angeles traffic, Elon Musk sent a short run of tweets. Traffic, he wrote, was driving him nuts. He would build a tunnel-boring machine and just start digging. Then, in the last of the four: I am actually going to do this. What almost nobody connects to that moment is where the man came from. Musk grew up in Pretoria, South Africa, in a family that was wealthy during his youth. After his parents divorced in 1979 he chose, at about nine, to live with his father, and he has said he later regretted it. At a boys' wilderness school he described as a paramilitary Lord of the Flies, he was, by his account, thrown down concrete steps and beaten badly enough to be hospitalised. He has said his father berated him after he came home. His father denies it. Nothing ties that boy to a tunnel company, but it is the person behind the tweets. He arrived in Canada alone in June 1989, stayed with a second cousin in Saskatchewan and worked odd jobs, including on a farm and at a lumber mill. Then Queen's University, then the University of Pennsylvania. When he co-founded Zip2 in 1995 he coded at night, seven days a week, in his own words. The tunnel idea came with a name within an hour, and the company was registered in January 2017. It was born inside SpaceX and started digging where it did for a plain reason: SpaceX owned the land, so no permits were needed. The first test trench was 30 feet wide, 50 feet long and 15 feet deep. Initial funding of about $113 million came mostly from Musk. The real fight came later. Los Angeles never let the idea in, and the plan was dropped. A graveyard followed: Chicago, Washington to Baltimore, San Jose, Fort Lauderdale, Miami. In 2018 the company sold about 20,000 blowtorches branded as flamethrowers and 50,000 hats to raise money from fans. Then Las Vegas wanted a shuttle rather than a subway; the contract came in May 2019 and the Convention Center loop opened in 2021. Critics never stopped questioning cost, speed and safety. In 2022 Texas regulators cited the company for five environmental violations. A $675 million round that April valued it at about $5.7 billion. A $3 billion round at $23 billion was reported for September 2026, but that figure is unconfirmed. The lesson is small and practical. A man who learned young that nobody was coming to help started by digging on land he already owned, where no one could say no. What kept it alive afterwards was one customer who wanted the small version of the dream. ## Postman (USA) URL: https://startup.talk/usa/postman/ Founders: Abhinav Asthana, Ankit Sobti, Abhijit Kane Founded: 2012 (side project); Postman, Inc. formed after 500,000 users; Based in: San Francisco, California (started in Bangalore, India); Sector: API development platform The first money Abhinav Asthana ever earned was an $80 cheque, and he never cashed it. There was no bank nearby that could take it. His father told him to deposit it, he has said, but he framed it instead. Fifteen years later, as he told First Round's podcast, that same client wrote to say how much they liked Postman. He got his first computer around 1997, after a months-long wait for it to arrive in India. His father, a civil engineer who had mastered Excel, taught him C. The family moved between small towns in Uttar Pradesh, and one profile starts his story in Basti. His parents set one rule on the machine: go and do something useful with it. And when he wanted a MacBook and an iPhone, his father said he had to earn them himself, which he did with freelance web work. The failures came early. After a year out preparing for India's entrance exams, he went to college in Goa and built a virtual tour of his campus that went viral and earned nothing. His first startup, a student question-and-answer site called Exam Crunch, bombed in its first week, and he has said he was disappointed. He moved to virtual tours, which became TeliportMe, an app with six or seven million users. What almost nobody knew was that Postman was an irritation. He was juggling back-end work for three interfaces at a startup and kept switching between his editor and API tools he found ugly. So he made his own, as a Chrome extension, and put it on the web store with no marketing. Developers found it by search. Feature requests arrived on GitHub. At first he built it only for himself. The turn was a sentence from users: this is the one tool I cannot give up. He recruited a former Yahoo and Adobe colleague, Ankit Sobti, who gave up a well-paid job, and Abhijit Kane. They shared a Bangalore apartment with a dog named Cooper. Money came from donations, three firms paying $500 a month and a $10 lifetime purchase he later called an extremely bad pricing decision. It still brought in thousands of dollars a month. A $1 million seed arrived in 2015. The price was a long confusion. He could not at first explain what Postman was. In 2017 he moved to San Francisco to be near customers. In August 2021 a $225 million round valued the company at $5.6 billion, a figure now more than five years old. The framed cheque is a small proof of something large. A father who would not let a son skip earning his own things raised a man who kept the first payment on a wall. ## SambaNova (USA) URL: https://startup.talk/usa/sambanova/ Founders: Rodrigo Liang, Kunle Olukotun, Christopher Re Founded: 2017; Based in: Palo Alto, California; Sector: AI chips, systems and inference cloud A Silicon Valley Leadership Group event listing describes Rodrigo Liang as one of the few, if not the only, Brazilian immigrants to raise more than a billion dollars in venture capital. The same listing says he would share his fundraising and his immigrant experience. It is a line that most profiles of a chip-company chief skip. What is documented is the career. He earned bachelor's and master's degrees in electrical engineering at Stanford. He started at Hewlett-Packard, then joined Sun Microsystems when it bought Afara Websystems in 2002, and worked on the multi-core Niagara processors. When Oracle took over Sun in 2010 he became a senior vice president running SPARC processor and chip development. He went back to his old campus. In November 2017 he started SambaNova with two Stanford professors, Kunle Olukotun and Chris Re. The chips grew from Olukotun's work on parallel hardware, and the software ideas from Re. Money came quickly: $56 million in a first round from Walden and GV, $150 million from Intel and Alphabet in 2019, and in April 2021 a $676 million round led by SoftBank Vision Fund 2 at more than $5 billion. Then the grind. Selling chips against Nvidia is hard, and the company shifted toward cloud services and AI inference. Trade headlines in 2025, not read in full, spoke of stalled fundraising, a possible sale and an Intel acquisition plan. A $350 million round in February 2026 and a $1 billion first close in July 2026 at an $11 billion valuation, figures not yet cross-checked. The cost of that stretch, in nerves or in personal terms, has not been told publicly in anything found. What a reader can take from the record is plainer. A founder who raised his billion from outside the usual circle did it by teaming with two people who owned the science, and by choosing a name that carried his own history into a company of silicon. ## CloudKitchens (USA) URL: https://startup.talk/usa/cloudkitchens/ Founders: Travis Kalanick, Diego Berdakin, Sky Dayton Founded: 2016; Based in: Los Angeles, California; Sector: Ghost kitchens, restaurant software and robotics In 2001, Travis Kalanick moved back into his parents' house. His second company, Red Swoosh, had almost no cash. He reportedly went without a salary for more than three years, and some of his employees had gone months without pay. The young man in that house was a Los Angeles kid from Northridge. His mother worked in retail advertising at the Daily News and his father was a civil engineer for the city. As a teenager he sold Cutco knives door to door. At eighteen he started a test-prep company, New Way Academy, with a classmate's father. He went to UCLA for computer engineering and business economics and left in 1998 for a file-sharing start-up, Scour. Scour was sued in 2000 for copyright infringement, with the entertainment industry seeking $250 billion, and filed for Chapter 11 that September. An investor, Michael Ovitz, had already sued the company and taken majority control; the experience soured Kalanick on how investors treated founders. What almost nobody remembers about Red Swoosh is how close it came to the edge. In September 2001 the company used about $110,000 in payroll tax withholdings to fund daily operations. Advisers reportedly warned it could be tax fraud, and accounts differ on who made the call. A second funding round gave it enough to repay the IRS, and no one was prosecuted. By 2002 only Kalanick and one engineer remained. In 2005 Mark Cuban invested $1.8 million, and in 2007 Akamai bought the company for about $19 million. After taxes, Kalanick made around $2 million. Then Uber, which he co-founded in 2009. He resigned under pressure in June 2017. At the end of May that year his mother had died in a boating accident that badly injured his father. He sold about 90 percent of his shares, reportedly for $2.5 billion. What he did next surprised few who knew him. In 2018 he paid about $150 million for control of City Storage Systems, a company Diego Berdakin and Sky Dayton had started in 2016 to rent out ghost kitchens, buildings equipped for delivery-only restaurants. The Saudi Public Investment Fund added $400 million in 2019, and an $850 million round in 2021 that included Microsoft put its value near $15 billion. 2023 was the low point: closures and layoffs after growing to about 40 sites in 24 cities and more than 4,000 employees. A salad chain, Mixt, sued in 2024. Kalanick pivoted to software and robots; in March 2026 the parent was renamed Atoms, with a reported $1.7 billion round that has not been confirmed. A founder can take one thing from the house in 2001. The man who later ran billion-dollar companies once covered payroll with money that was not his, and lived with his parents to keep the lights on. He has been making bets under pressure ever since. ## Ro (USA) URL: https://startup.talk/usa/ro/ Founders: Zachariah Reitano, Saman Rahmanian, Rob Schutz Founded: 2017 (launched as Roman, Oct 2017; renamed Ro, Sep 2018); Based in: New York, New York; Sector: Digital health / telehealth and direct-to-patient pharmacy Zachariah Reitano's heart stopped on a treadmill. A profile of him says it happened at the twelfth minute of a stress test, and the doctor had to call for help to revive him. The test came after a conversation he had put off for as long as he could. When he was 17, he noticed something wrong that no teenager wants to explain to a parent. His father was a doctor who treated men's sexual health, and the profile says he delayed telling him. When he finally did, his father recognised it could signal something bigger. It was: an undiagnosed heart condition. A cardiac ablation followed. In February 2017 the symptoms returned, and the doctor said he would need the very medication whose side effect had sent him to his father in the first place. His family could talk about anything at the dinner table, and his parents taught him that if something did not work, he should fix it or build something better. In high school he read a New York Times article about the lack of permanent rabbis in Cuba and went there for three months, six months before he was due to start at Columbia. At 22 he started his first company, Shout. The profile says that when his sister fell ill he cared for her for six months and refunded 80 percent of the money investors had put in. What almost nobody sees behind Ro is that the company is an answer to luck. Reitano's father happened to be exactly the right kind of doctor. Most people are not that lucky. A reference work says he wanted to recreate his father, a physician who listens, in software, so anyone could have that first honest conversation. He launched in October 2017 as Roman, with co-founders Saman Rahmanian and Rob Schutz, selling men's sexual health treatments online. General Catalyst wrote a $3.1 million seed cheque. In 2018 the name became Ro, and FirstMark led an $88 million Series A. In March 2021 a $500 million round valued it at $5 billion, a mark now more than five years old. The cost came in public. A New York subway campaign was called too evocative, and a 2025 campaign with Serena Williams drew controversy. Even so, deals with Eli Lilly in December 2024 and Novo Nordisk in January 2026 followed. The idea began with a night someone took a symptom seriously. Reitano has spent the years since trying to make that happen for everyone else. ## Socure (USA) URL: https://startup.talk/usa/socure/ Founders: Sunil Madhu, Johnny Ayers Founded: 2012; Based in: New York, New York; Sector: Digital identity verification / fraud prevention (AI) Sunil Madhu, one account says, got the idea for Socure after moving to the United States from Europe and finding how hard it was to get approved for anything without a credit history. The company that came out of that frustration now decides, in a fraction of a second, whether a stranger opening a bank account on a phone is real. Socure was founded in 2012 by Madhu and Johnny Ayers. The company's press releases are datelined Incline Village, Nevada. Madhu was the first chief executive. Then the leadership changed fast. A former Oracle sales executive, Tom Thimot, joined as CEO in spring 2018. Madhu left in February 2019, a Delaware court summary of a May 2024 case, Madhu v. Socure Inc., records, along with the fact that he founded the company in 2012. Ayers was named CEO on 17 November 2020. Madhu went on to start Instnt in 2019. The sources do not say what the fight in court was about, and it would be wrong to guess. What the record does show is the cost of being a founder who builds the thing and is not there for its biggest years. Those years were big. In November 2021 Socure was valued at $4.5 billion. In October 2024 it agreed to pay $136 million for Effectiv, a real-time risk decisioning platform, with the whole team expected to join. In 2026 it bought Fravity, an agentic AI startup with more than 70 specialised agents that gather evidence for fraud cases, and won a $163 million five-year federal contract to provide identity-proofing for Login.gov. Infrastructure companies tend to reveal their founders late, and Socure reveals little even now. The lesson is plain and a little uncomfortable. The idea can come from a founder's own private frustration, and the company can still outgrow the founder's place in it. ## Outreach (USA) URL: https://startup.talk/usa/outreach/ Founders: Manny Medina Founded: 2014; Based in: Seattle, Washington; Sector: Sales engagement software (SaaS) Manny Medina's parents wanted him to study engineering in East Germany. Then the Berlin Wall fell, and the plan fell with it. In an ExecuTalks interview he said he pitched software engineering to his father instead, and his father agreed on one condition: a graduate degree. He grew up in Ecuador. His father was from there, and his mother had emigrated from Russia. He has said he often felt like an outsider, an immigrant's child. He did what his father asked. He left Ecuador for the United States at 20, as GeekWire reported in 2018, and later earned a bachelor's in computer engineering from Stevens Institute of Technology and a master's in computer science from the University of Pennsylvania. His conversational English was weak at first, and building a social life in New Jersey took about a year. "For me, the American Dream is that I could be anybody I wanted to be," he told GeekWire. "And that wasn't possible in Ecuador." The résumé that followed looks like a straight line. He was the third employee on the AWS team and built Amazon's associate compensation system. He spent about six years at Microsoft, rising to director of business development for Canada and Latin America. What the line hides is the next move. Around age 40, he said, he left Microsoft with no concrete plan. He gave two reasons: he wanted to contribute something meaningful, and he wanted control over the outcome. Outreach was founded in Seattle in 2014, about three years later, and in the gap between the safe job and the company there was nothing guaranteed. The product itself is plain. It helps sales teams run the repetitive part of selling: the emails, the calls, the follow-ups that never quite happen. In 2018 Outreach raised $65 million, and in June 2021 a round valued it at $4.4 billion. That valuation is more than five years old, and no newer priced round, revenue figure or profit number turned up. His own summary of the choice is the sentence a first-time founder may keep. "The regret of not trying is greater than the regret of failure," he said. Medina paid for it with three years of not knowing, and with a father's condition that he finish the degree first. He did, and the second one came after. ## Arctic Wolf (USA) URL: https://startup.talk/usa/arctic-wolf/ Founders: Brian NeSmith, Kim Tremblay Founded: 2012; Based in: Eden Prairie, Minnesota; Sector: Cybersecurity / managed detection and response In August 2021, Brian NeSmith did something most founders resist: he gave away the chief executive job. Arctic Wolf was heading toward an expected IPO. Nick Schneider, the president and chief revenue officer, became CEO, and NeSmith moved to executive chairman. It was not his first company, or his first handover. NeSmith graduated from MIT in 1984 with a degree in electrical engineering, a profile says, and worked at Newbridge Networks. He later became CEO of Ipsilon Networks, which Nokia acquired in early 1998. He then ran Blue Coat Systems, and helped take it public. By the time he and Kim Tremblay, who had worked with him at Blue Coat, founded Arctic Wolf in 2012, he had already seen one company sold and another listed. His childhood, his family and the doubts of the early years are not in any source that could be read. In a 2023 podcast for his own company, The Howler, he promises to talk about the lessons he has learned from failure. The listing does not say which failures. That silence is part of the record. What is documented is the idea. In 2016 Arctic Wolf launched managed detection and response, first called AWN CyberSOC. Instead of selling software alone, it supplied the people who watch it. A customer too small to staff a security operations centre could rent one. Money followed. In October 2020 a $200 million round led by Viking Global valued the company at $1.3 billion. Nine months later a $150 million round tripled that to $4.3 billion. Around then it moved its headquarters from Silicon Valley to Eden Prairie, Minnesota. It then bought its way outward: Habitu8 for training, Revelstoke for automation, Cylance for endpoint, and Upsight Security in November 2025. By 2026 it served more than 10,000 organisations in over 80 countries. The $4.3 billion valuation is more than five years old, and no IPO is confirmed. The handover that was supposed to lead to a listing has not yet delivered one. So the closing question belongs to the founder. NeSmith built the company after two earlier ones, and then stepped aside from the top seat for someone suited to the next stage. Knowing which stage you are good at may be the rarest skill in the story. ## Relativity Space (USA) URL: https://startup.talk/usa/relativity-space/ Founders: Tim Ellis, Jordan Noone Founded: 2015; Based in: Long Beach, California; Sector: Aerospace / 3D-printed rockets Within a week of leaving Blue Origin, Tim Ellis sent a cold email to Mark Cuban, titled 'Space is Sexy 3D Printing an Entire Rocket.' Cuban replied within minutes, Ellis later said, and offered $500,000. (Accounts differ on the exact terms.) That same week Ellis applied to Y Combinator. It was a long way from where he started. He grew up in Plano, Texas, the eldest of three, and as a kid he filmed homemade zombie movies with friends. He went to USC to become a screenwriter. After his freshman year, he has said, he was not very excited about aerospace. What almost nobody expects from a rocket founder is that he arrived by accident. In his sophomore year he joined the USC Rocket Propulsion Lab, and it changed the plan. He ended with a bachelor's and a master's in aerospace engineering and a leadership role in the lab, alongside classmate Jordan Noone. Then came Blue Origin: three consecutive internships and two years full-time. There he created the company's first metal 3D printing program, often working nights and weekends on it because it was not his main job. He raised the idea with Jeff Bezos, who reportedly found it promising but not enough to act on. Ellis decided to start his own company. Founded in 2015, Relativity grew on a bold claim: print most of the rocket. The money followed. A $500 million round in 2020 valued it at $2.3 billion, and in June 2021 Fidelity led $650 million at $4.2 billion, bringing the total to $1.335 billion. Then the hard part. On March 23, 2023, Terran 1 flew for the first time, called 'Good Luck, Have Fun.' It survived max-q, the point of greatest stress, but an upper-stage anomaly kept it from orbit. A month later the company retired Terran 1 to focus on the larger, partially reusable Terran R. In March 2025 came the sharper turn: Eric Schmidt replaced Ellis as CEO and took a controlling interest. Ellis stayed as co-founder and board member. Relativity reported a Terran R backlog above $2.9 billion, qualified its Aeon R engine in July 2025, and in June 2026 announced a NASA partnership for a Mars orbiter planned for 2028. Terran R's first launch has slipped to no earlier than 2026. Ellis has said relentlessness is the most important skill he has learned. A man who set out to write films ended up building rockets, and the idea that carried him there was one he worked on in his spare nights. ## Chainalysis (USA) URL: https://startup.talk/usa/chainalysis/ Founders: Michael Gronager, Jonathan Levin, Jan Moller Founded: 2014; Based in: New York, New York; Sector: Blockchain analytics and crypto compliance software In December 2024 the man who built Chainalysis from a bankrupt exchange's wreckage agreed to leave it. Michael Gronager had taken a leave of absence two months earlier, citing a personal issue. TechCrunch reported that he would no longer sit on the board but would keep his equity. The story started somewhere stranger than a startup. Gronager holds a doctorate in quantum mechanics from the Technical University of Denmark. Before crypto, in other words, his training was in things you cannot see directly and must infer from traces. That describes blockchain forensics rather well. He came to Bitcoin as an exchange insider. In February 2014 Mt. Gox, then among the biggest Bitcoin exchanges in the world, disappeared almost overnight. Gronager was a founding member of Kraken, serving as its chief operating officer, and Kraken was hired to help the bankruptcy trustee understand what had happened. Others saw a disaster. He saw a missing product: nobody sold a way to follow money across the public ledger. What almost nobody outside the industry grasped then was how little the ledger hid. Everyone said Bitcoin was anonymous. It was not; it was pseudonymous, and that difference became a business. Chainalysis, co-founded with Jan Moller and Jonathan Levin, describes itself as the first start-up dedicated to tracing Bitcoin. Governments became the customers who mattered. The FBI, the DEA, IRS Criminal Investigation and Britain's National Crime Agency used the tools. In 2019 the software helped the Justice Department take down a child abuse site; in 2020 it helped recover more than a billion dollars tied to Silk Road. Banks and brokers needed it for compliance. By 2022 investors valued the company at $8.6 billion in a $175 million round. The price came later. Prosecutors leaned on its analysis in the 2022 Bitcoin Fog trial, and outside experts asked whether the conclusions could be independently checked. Late in 2023 it cut slightly more than 15 percent of about 900 staff. Secondary shares were reportedly bought at $2.5 billion in 2024, far below the peak. Levin, who took over as chief executive, told TechCrunch the company does not need to raise capital. What is documented is the shape of the arc: a physicist who spotted a gap in a catastrophe, built the tool that filled it, and then watched the company carry on without him. A founder who spots the question nobody is paid to answer can build a company around it. The company, once built, does not have to stay his. ## Houzz (USA) URL: https://startup.talk/usa/houzz/ Founders: Adi Tatarko, Alon Cohen Founded: 2009; Based in: Palo Alto, California; Sector: Home remodeling and design marketplace and software They were found close to home. Adi Tatarko and Alon Cohen asked a few Bay Area architects to upload portfolios to a site Cohen had coded himself, and the first customers, one account says, were other families at their children's school. Tatarko was born in Tel Aviv. A Carnegie Corporation profile says she grew up admiring the women in her family. Her grandmother, a Holocaust survivor, was an internationally exhibited fashion designer, and her mother ran a real estate business. Design and property were the family trade before they were hers. She met her husband while on vacation in Thailand. Cohen was Israeli too. They started a small tech services company in Israel, married in 1998, and moved to Silicon Valley. After leaving the tech sector, Tatarko worked as a financial adviser while raising their sons. Then came the house. A difficult remodel in 2006 left them struggling to explain what they wanted and to find the right people to build it. Nobody had a good place where a homeowner could see an architect's work and say: like this. Houzz went up in February 2009. What nobody planned was that strangers outside the Bay Area would ask for more categories and more cities. It became a company in fall 2010, with an iPad app that November. By 2014 there were offices in the UK, Germany and Australia, and by 2022 about 65 million users. The cost was not small. A privacy lawsuit over call-recording notifications, filed in 2015, was settled. The $4 billion valuation the company reached in 2017 has not been refreshed in any source found, so it is now stale. In 2020 the company launched Houzz Pro, software for the professionals who had been its guests, a harder business than inspiration. Tatarko has been blunt about what she thinks founders get wrong. Her advice, in the Carnegie profile, is 'Don't enter to exit.' It reads differently from a woman whose grandmother made clothes and whose mother sold houses. Those women built things that stayed. A company that begins as a fix for your own house may last longer than one that begins as a plan to sell. ## StockX (USA) URL: https://startup.talk/usa/stockx/ Founders: Josh Luber, Dan Gilbert, Greg Schwartz, Chris Kaufman Founded: 2015 (launched Feb 2016); Based in: Detroit, Michigan; Sector: Resale marketplace for sneakers, streetwear, collectibles Josh Luber did not set out to build a stock market. He set out, he told Emory's alumni magazine, to get some sneaker data 'just to play with for my own amusement.' He was a strategy consultant at IBM at the time. It was a hobby with a spreadsheet. The hobby had a long history. Luber has collected sneakers since childhood and may still own pairs from when he was about eight. He studied at Emory, earned his law degree there, and, he said, actually loved law school. He kept an early collection in his apartment closet, a law-school friend remembered. He started his career at the law firm Alston & Bird. What almost nobody sees in the polished version is the string of attempts that came before. While holding corporate jobs he founded three startups. One profile says one of them, an online scheduling platform, went bankrupt in the 2008 financial crash. The Emory piece does not call them failures, so treat that detail with care. What it does show is a man who kept building on the side. The data idea worked. Around 2012, Luber and a team of volunteers began scraping public eBay data and analysing millions of sneaker sales. Campless became a kind of price guide for shoes: what a pair actually sold for, not what a seller hoped. Then, about three years in, Dan Gilbert approached him. Gilbert acquired Campless, Luber left IBM and moved to Detroit, and StockX launched in February 2016 with Greg Schwartz and Chris Kaufman. Buyers bid, sellers ask, and prices move like a ticker. Every pair goes to StockX to be inspected; fakes go back to the seller. Growth was fast and rough. The company raised $110 million in 2019 at a $1 billion valuation and $255 million in 2021 at more than $3.8 billion. In mid-2019 it suffered a data breach. Scott Cutler took over as chief executive in 2019, and Luber left, later moving into trading cards. Luber's own account is modest. He wanted to look at sneaker prices for fun. A closet collector, a lawyer, an IBM consultant and a three-time startup builder became the founder because he kept looking at an odd dataset until someone else saw a market. Curiosity with no business plan can be the first draft of one. ## Cohesity (USA) URL: https://startup.talk/usa/cohesity/ Founders: Mohit Aron Founded: 2013; Based in: San Jose, California; Sector: Data security and data management software By 2013 Mohit Aron had already built something that made him a name. He co-founded Nutanix in 2009, and then, before it went public, he left to start another company. A Business Insider India headline later said he walked away from a $4 billion company because he had an even better idea. He was born in Pathankot, India. What Aron has said himself is that his father's resilience mattered to him, and that his parents pushed their children toward the best education available. He took the entrance exam, YourStory reports, ranked 53, and entered IIT Delhi in 1991 for computer science. The exam rank is the detail that stays with you. A teenager who placed among the top few dozen on the country's hardest technical entrance exam had, in effect, been told he could go anywhere. He went to Delhi, graduated in 1995, then to Rice University for a PhD in computer science, finished in 2000. Then came the jobs that shaped his thinking. He joined a storage startup, Zambeel, in 2000, and Google in 2003, the year before its IPO. YourStory reports that he was part of the team that built the Google File System, the system that let Google store huge amounts of data across cheap machines. Cohesity, started in June 2013, applied that thinking to the mess of copies that companies kept for backup and analytics. It announced itself in 2015 with $70 million raised, and added $90 million from GV and Sequoia in 2017 and $250 million from SoftBank's Vision Fund in 2018. By December 2021 it had raised about $660 million and filed for an IPO. The listing did not arrive. In August 2022 Sanjay Poonen replaced Aron as chief executive. In December 2024 Cohesity completed the $3 billion purchase of Veritas, a far older company, and its revenue was about $1.7 billion in 2025. Aron's path had a pattern: leave when the next problem looks bigger than the current success. It made him a founder twice. It also meant that a company he began was, in the end, run by someone else. ## Noom (USA) URL: https://startup.talk/usa/noom/ Founders: Saeju Jeong, Artem Petakov Founded: 2008 (WorkSmart Labs formed 2007); Based in: New York, New York (moved to Princeton); Sector: Digital health, behavioral weight loss app Saeju Jeong arrived in Manhattan in 2005 with, by one account, $3,000 and nobody he knew. He had left a Korean engineering degree at Hongik University behind. He took sales jobs. He delivered food. The reason he left is the hidden thread of the story. Jeong was born in 1981 into a family of doctors. He has said there were 29 medical doctors in his family. His father was one, an obstetrician and gynecologist, who died early of lung cancer, as Jeong said on a 2023 podcast. An Aspen Institute profile says that loss redirected the course of his life. The son of a renowned doctor did not become one. Before that came a stranger start. In 2001 he founded BuyHard Productions, an independent heavy-metal label. In 2002 he joined the Korean army as an IT specialist. Then he enrolled at Hongik and, in 2005, dropped out. In Harlem, he met Artem Petakov, a former Google software engineer. In 2007 they founded WorkSmart Labs, whose first prototype was an accessory for a stationary bike. It was not an obvious route to healthcare. In 2011, renamed Noom, it raised a $2.6 million seed led by Kleiner Perkins. The turn came with the app. A study of about 36,000 users found 78 percent self-reported weight loss, and in 2017 the CDC recognized Noom as a diabetes prevention program, the first mobile health app to receive the designation. A subscription version made it a business, with about $400 million in revenue in 2020 and $540 million raised in 2021 at a $3.7 billion valuation. The cost arrived with the scale. Critics said its calorie goals were sometimes too low and its coaches lacked dietitian credentials. In 2022 it agreed to a $62 million settlement over automatic renewals, and a former engineer testified that cancelling was made hard on purpose. Jeong stepped down as CEO in 2023 and Geoff Cook took over. A founder whose father was a doctor ended up building a health company anyway, by another door. The family trade did not pass to him as medicine. It passed as a question: what do people need to stay well, and who will help them do it? Whether that was his own reasoning, the sources do not say, but the path is documented. ## Thumbtack (USA) URL: https://startup.talk/usa/thumbtack/ Founders: Marco Zappacosta, Jonathan Swanson, Jeremy Tunnell, Sander Daniels Founded: 2008; Based in: San Francisco, California; Sector: Local services marketplace (home services) Before Thumbtack had a product, it had a student campaign. Marco Zappacosta, Jonathan Swanson and Jeremy Tunnell were fellow directors of Students for Saving Social Security, a group that claimed more than 10,000 members on over 300 campuses. They were political organisers first, and none of them was building software for plumbers. Zappacosta grew up in Atherton, born in Stanford in 1985. His father, Pierluigi Zappacosta, was the Italian engineer who emigrated to the US in the 1970s and co-founded Logitech. A founder's last name like that comes with a question attached: is this the son's company or the father's shadow? Zappacosta graduated from Columbia in 2007 with a degree in political science, not engineering. The idea that pulled the group together was plain. Hiring someone locally, whether a photographer, a mover or an electrician, was a mess on Craigslist. There was no way to tell who was real, no reviews, no licences. In 2008 the founders, with Sander Daniels, built a directory to fix that. To get it started they scraped Craigslist for providers, then added photos, licences and reviews. What almost nobody remembers is how thin the early money was. Thumbtack's first outside round, in June 2010, was only $1.2 million, mostly from angels. A $4.5 million Series A followed in 2012, by which time the site had 250,000 merchants and professionals on it. Then the cheques grew fast: $30 million from Sequoia and Tiger Global in 2014, and $100 million from Google Capital that August. The model that made it work was to let customers post a job and let professionals pay to quote on it, rather than charging a listing fee. Pros paid for leads they chose. It was a marketplace built on the idea that trust is a product. The 2021 round put a $3.2 billion valuation on the company, led by Qatar Investment Authority. Fast Company reported $400 million in revenue for fiscal 2024, and Forbes counted about 300,000 active professionals in mid-2025. No newer priced round has been found, so the valuation is over five years old. The lesson is in the founders' first group project. Organising 10,000 students to care about something taught them how to get strangers to trust a stranger. For a first-time founder, the skills for your company may be the ones you built outside any company at all. ## Verkada (USA) URL: https://startup.talk/usa/verkada/ Founders: Filip Kaliszan, James Ren, Benjamin Bercovitz, Hans Robertson Founded: 2016; Based in: San Mateo, California; Sector: Cloud-managed physical security (cameras, access control, sensors) In March 2021, a group calling itself Arson Cats spent about 36 hours inside Verkada's network. The company sold cameras to hospitals, gyms and offices, and now it was the story. Early reports spoke of footage from more than 150,000 cameras. Later reporting put it at 95 customers. Filip Kaliszan was the CEO who had to explain it. He had grown up in Poland, by his own account, and one speaker bio says he was raised in Paris. At Stanford he studied computer science and, as a sophomore, started a company with two classmates, James Ren and Benjamin Bercovitz. It was called CourseRank, a site to help students choose classes. Chegg bought it in 2010. After that Kaliszan worked in product roles at Guidebook and Chegg. The itch for something physical came later. In 2016 the three Stanford friends founded Verkada with Hans Robertson, a former COO of Meraki, the company that had made networking gear manageable from the cloud. The pitch was Meraki's trick applied to cameras: no recorder in a closet, no server to patch. The first install was modest. In 2017 the team put its cameras in an Equinox gym in Beverly Hills. Then came money. A $40 million Series B in April 2019 valued the company at $540 million, and eight months later a Felicis-led round made it $1.6 billion. In spring 2020 it added door access, and in the pandemic it offered free surveillance kits. Then the breach. It tested the thing a security company sells, which is trust. Kaliszan announced a bug bounty, red and blue team exercises, mandatory two-factor for support staff and better audit logs for customers. The company did not stall. A $305 million round in October 2023 valued it at $3.2 billion, and by December 2025 CapitalG led $100 million at $5.8 billion. Revenue was $806 million for 2025, with about 30,000 customers. What the sources do not give is much about Kaliszan's early life beyond his own few lines, and it would be dishonest to fill the gap. What they do show is a pattern: a team that started in a dorm-era project, sold it, and then stayed together for a second company. Co-founders who have already survived a first exit know how the other one behaves when things go wrong, and in 2021 that mattered. ## Calendly (USA) URL: https://startup.talk/usa/calendly/ Founders: Tope Awotona Founded: 2013; Based in: Atlanta, Georgia (HQ closed Jul 2021; fully remote); Sector: Scheduling software (SaaS) When Tope Awotona was 12, his father was killed in a carjacking in Lagos. His father had left a microbiology job to start his own company, and had shown his son what a business looked like. His mother worked at a bank. After the killing, she moved the family to the United States, to the Atlanta area. What almost nobody connects to Calendly is how that loss sits behind it. A scholarship offer came when he was 15, but his mother made him first finish two years at Wheeler High School. Awotona said later that it was 'super frustrating'. He felt he was wasting time because his plan was in place. He graduated from the University of Georgia's Terry College of Business in 2002, after switching from computer science to business information systems. Then came jobs at IBM and Perceptive Software, and a stint selling at EMC. In sales he kept hitting the same small wall: arranging a call took a chain of emails. He decided to fix it. He put in his life savings and took on debt. He built Calendly at Atlanta Tech Village with help from Railsware, a Kyiv firm. A $550,000 seed arrived in late 2013, and the product launched free for individuals, with a premium tier added late in 2014. It spread the way a good link does: each person who got one learned how to send one. The growth was slow, then enormous. In January 2021, OpenView and Iconiq put $350 million into the company at a reported $3 billion valuation. In July 2021 Awotona closed the Atlanta headquarters and went fully remote. Forbes estimated his net worth at $1.4 billion in 2025. A scheduling link looks like a small thing. The founder behind it had already learned, at 12, that plans can be taken away without warning, and at 15 that waiting is a cost. His story says something to anyone starting late or after failure: the thing you build may be small, but the reasons you cannot afford to quit can be very large. ## Lucid Software (USA) URL: https://startup.talk/usa/lucid-software/ Founders: Karl Sun, Ben Dilts Founded: 2010 (Lucidchart LLC formed Apr 2009); Based in: South Jordan, Utah; Sector: Visual collaboration software (diagramming, whiteboarding) The first marketing budget for Lucidchart was about $100 in ads. Ben Dilts put a prototype online at the end of 2008, wrote to bloggers and waited. He was, by the account in Utah business coverage, a student who had to go back to Brigham Young University to finish his degree. Something in the product caught. A diagram editor that lived entirely in the browser was not common in 2008. People who hated installing heavy software found they could draw a flowchart and send a link. Lucidchart, LLC was formed in Utah in April 2009. What the prototype lacked was money and a business partner. Dilts found both at a speed-pitching event, where founders rotated through tables of potential investors. There he met Karl Sun, a former Google employee who had spent seven or eight years there, though not, Crunchbase News later corrected, as an engineer. Sun wrote an early cheque. The two formally co-founded Lucid in 2010 and worked out of Dilts's student apartment on roughly $200,000 from Sun's network. The next break was distribution. In 2010 Lucidchart landed in the Google Apps Marketplace, putting it in front of companies that already lived in the browser. In 2011 a $1 million seed came from 500 Startups, 2M Companies, K9 Ventures and angels, and the company incorporated in Delaware in January. Growth stayed in Utah. In October 2018 Meritech, Spectrum Equity and ICONIQ put in $72 million. Lucidspark, a whiteboard, arrived in 2020, just as remote work made whiteboards a battleground. By July 2021 the company was valued at $3 billion, and in April 2022 Karl Sun stepped back to board chair while the long-time operator Dave Grow became CEO. What the sources leave out is the doubt: nothing found describes how Dilts or Sun felt in the apartment years, so this account will not invent it. What they do show is the stake. A student with a prototype and a $100 ad budget needed a person who believed enough to write a cheque. The lesson is ordinary and easy to forget. The product got him into the room, but a speed-pitching table got him a partner. Go to the dull events. ## Automattic (USA) URL: https://startup.talk/usa/automattic/ Founders: Matt Mullenweg Founded: 2005; Based in: San Francisco, California (distributed company); Sector: Web publishing (WordPress.com, WooCommerce, Tumblr, Jetpack) In September 2026, Automattic's board put its founder on paid leave and named the CFO interim CEO. Two days later, Matt Mullenweg told the company he was back in control. It was a strange week for a man who had spent two decades giving away the thing the company was built on. Go back to Houston. Mullenweg was born there in 1984 and grew up in the Willowbend neighbourhood. His father, a programmer, brought computers into the house early. Childhood migraines kept the boy out of school for long stretches, which matters: a kid who loses weeks at a time learns to fill them with something. He went to Kinder High School for the Performing and Visual Arts and studied jazz on saxophone. What almost nobody pictures is the Mullenweg who never finished college. He spent two years at the University of Houston, reading philosophy and political science, and left in 2004 to take a job at CNET. In 2002 he had started blogging on an open-source tool called b2/cafelog. When its developer stopped maintaining it, Mullenweg took over in 2003, then forked it with Mike Little. That fork was WordPress. CNET hired him in October 2004 and let him work on WordPress part-time, and moved him to San Francisco. In 2005 he founded Automattic to build a business on it. The strange part was a choice about ownership: in 2010 the WordPress trademarks and source code went to the non-profit WordPress Foundation. The company grew by purchase. Gravatar came in 2007, then IntenseDebate and Polldaddy, WooCommerce in 2015, Tumblr in 2019 and Parse.ly in 2022. A 2014 round of $160 million valued it at about $1.16 billion, and Salesforce Ventures led a $300 million round at $3 billion in 2019. A 2021 valuation put it at $7.5 billion. Then the fights. In September 2024 a public dispute with WP Engine over the WordPress trademark led more than 8% of staff to resign after Mullenweg offered severance to those who disagreed. In April 2025, 16% of staff were laid off. The sources offer no interviews about how he felt. They do show a pattern: a boy who spent long stretches at home found a project that belonged to everybody, then later fought to control what it meant. The lesson for founders is uncomfortable. Giving a thing away is easy at 19, and holding it together is the hard part at 40. ## Flipkart (India) URL: https://startup.talk/india/flipkart/ Founders: Sachin Bansal, Binny Bansal Founded: 2007 (October); Based in: Bengaluru, Karnataka; Sector: E-commerce By the second half of 2007, two young software engineers were running a bookshop out of a Bengaluru house, and they were doing every job themselves. They wrote the code, collected books from local shops, answered customer queries and delivered the parcels. Sachin Bansal later called it "a two-man show". They had pooled somewhere between $6,000 and ₹5.7 lakh, depending on whose account you read, and the company was not formally registered until 14 October 2008, about a year after they started coding. Both men came from Chandigarh and both went to IIT Delhi. Sachin took the entrance exam twice, climbing from an all-India rank of 2470 to 49. Binny, he has said, was turned down twice by Google before he ended up at Amazon. It was at Amazon's Bengaluru office that the two, who are not related, first became close friends. Binny had joined Amazon in January 2007 and stayed about nine months; Sachin had spent a year or so on the Amazon Web Services side after an earlier job at Techspan. Then, tired of their jobs, they resigned. Nobody rushed to help. They struggled to get friends to join them, and early investors told them they were wasting their time. Book distributors were baffled: of the fifty-odd in Bengaluru, the pair managed to meet only about half, and the question they kept hearing was why anyone would leave a great job at Amazon. For the first four or five months, customers hardly came. They tried blog partnerships and handed out bookmarks at book fairs. It took about eighteen months before a small firm, Accel, pledged about $1 million in 2009 (Accel's own pages later describe its first cheque as $800,000). Even then, the partners who backed them, Subrata Mitra and Prashanth Prakash, did so despite worries that two men in their mid-twenties had no supply-chain experience. The turn came sideways. When Flipkart added electronics, buyers refused to pay before they held the product, unlike book readers who did not mind paying a few hundred rupees upfront. So the founders offered cash on delivery, which Binny later called "a big innovation". Customers multiplied, a New York hedge fund put in $10 million, and the company reportedly began doubling every few months. Success did not come clean. In 2012 Flipkart launched several categories at once, missed its targets, and Binny admitted the team "didn't know why"; raising money got harder. The 2014 Big Billion Day sale crashed under demand, and in mid-2015 a new seller-focused strategy failed just as Amazon poured in money. In 2018 Walmart bought about 77% for roughly $16 billion. Sachin sold his stake, and his net worth stood at a little over $1 billion. Binny resigned in November 2018 after a misconduct allegation; a Walmart investigation found no evidence for the assault claims but flagged poor judgement, as the Wall Street Journal reported. Today Walmart owns about 85%. In the year to March 2025 Flipkart booked revenue of ₹82,787 crore and a net loss of ₹5,189 crore, and in 2025 it moved to bring its parent company home to India. Binny once said that the feeling of having arrived lasted only five minutes. For a first-time founder, that may be the more useful lesson than any valuation: the early years were won by two people doing unglamorous work by hand, while the people around them kept asking why they had quit a good job. ## Zomato (India) URL: https://startup.talk/india/zomato/ Founders: Deepinder Goyal, Pankaj Chaddah Founded: 2008 (launched as FoodieBay 10 July 2008; renamed Zomato Nov 2010); Based in: Gurugram, Haryana; Sector: Food delivery, restaurant discovery, quick commerce (Blinkit) In March 2009, Deepinder Goyal's parents found their son working from home and could not work out why. Their worry grew, and peaked when he told them he had left his consulting job at Bain to build a restaurant website. They urged him to go back to a "job". Years later, he says, they are "extremely happy and proud now". He came from Muktsar, a small town in Punjab. In 1998 he arrived at a talent-exam camp in Chandigarh looking, as a book on his life puts it, as lost as he felt, and it was there that he met Albinder, the friend who would stay close. After Class 10 they finished school together at DAV College. At a coaching centre for the JEE, both boys were put in the last batch, D. Teachers hardly noticed them, and one had to carry chairs into the corridor so they could see the whiteboard. The book calls it a confirmation of every self-doubt he had carried. Under the pressure he chose to skip the exam for a year and told a disappointed family. He had always avoided the notoriously hard physics books, convinced he was no good at the subject. Then one day, out of curiosity, he picked up a roommate's copy of HC Verma and got absorbed. Without anything left to prove, he took the exam on a whim and cleared it on his first attempt, and went on to finish an integrated M.Tech in mathematics and computing at IIT Delhi in 2005. When his parents came to help him pack up his PG room, his father, Harbans Lal Goyal, was exasperated by the mess. Above the desk hung a poster that read "I was born to lead, not to be ruled." His father joked that he should first learn to do his own laundry. It was not his first try at putting food online. Around 2005 he had built foodlet.com, a delivery site that drew hardly any users and was shut after nearly a year. "2005 was way too early," he said in a 2015 interview. Zomato began with a small irritation. At Bain, colleagues struggled to order lunch from the canteen, so Goyal and Pankaj Chaddah put the canteen menu and nearby restaurants online. That became FoodieBay in July 2008. In its first year it listed more than 1,400 Delhi-NCR restaurants without spending on marketing, and the company, DC Foodiebay Online Services, was formally incorporated on 18 January 2010. In November 2009 the two quit their jobs. Then came an unexpected email from Sanjeev Bikhchandani. He had tracked down Goyal's contact details himself and wrote that he used the site and was a fan. Goyal replied within a day, they met within two, and the deal followed within three. Info Edge's investment in 2010 is reported at about ₹4.7 crore, roughly $1 million. In November that year FoodieBay became Zomato, partly to avoid a clash with eBay and partly because the founders were not sure they would stay with food. The price of ambition was paid in public. In 2015 a data breach exposed about 62.5 million users, and Zomato cut roughly a tenth of its staff as global expansion proved costly. Chaddah left in 2018. After the 2021 IPO the company retreated from every market but India and the UAE, and later shut its parcel service. What carried it through was Blinkit and a sharper focus on India. For the quarter ending September 2026 Eternal, Zomato's listed parent, reported a net profit of ₹125 crore, up 92% from a year earlier. One profile reports that in January 2026 Goyal stepped down as Group CEO to chase higher-risk ideas. The thread through his story is a boy who avoided the hardest subject because he believed he was bad at it, and who cleared the exam once the fear of proving himself was gone. Many founders carry a label like that. It is worth checking whether it was ever true. ## Swiggy (India) URL: https://startup.talk/india/swiggy/ Founders: Sriharsha Majety, Nandan Reddy, Rahul Jaimini Founded: 2013 (incorporated Dec 2013; food delivery launched Aug 2014); Based in: Bengaluru, Karnataka; Sector: Food delivery, quick commerce (Instamart) When Swiggy opened in Bengaluru's Koramangala in 2014, it had 15 to 20 restaurants and five or six delivery partners. Some restaurants refused to join at all. So the team bought the food, photographed it at home, and listed it anyway. Sriharsha Majety had come back from the UK with no clear plan. He had studied electronics and physics at BITS Pilani, finished at IIM Calcutta in 2011 and spent a short spell at Nomura. He tried several ideas that went nowhere. Then he and his friend Nandan Reddy started Bundl, a courier service for small e-commerce sellers, which lasted barely eight months from the day it was incorporated. The papers say Bundl Technologies was registered on 26 December 2013, and its memorandum had been signed in Vijayawada a month earlier by only two people: Majety, then 27, with 6,875 shares, and Reddy, 26, with 3,125. Rahul Jaimini's name is not on it, and the address on the certificate is the Majety family home on K.T. Road. Reports point to the missing piece: they had no technical co-founder, and the big platforms were moving delivery in-house. He later called shutting Bundl "probably the best decision we made over the last 10 years." The second try began with a coder. Rahul Jaimini, an IIT Kharagpur graduate of 2010 who had been a software engineer at Myntra, joined, and in August 2014 the three pivoted to something smaller and more frequent: restaurant food. Accel and SAIF backed it in April 2015, at about seven months old, with over 30,000 orders delivered and 150 delivery executives. Swiggy had no mobile app until April 2015, eight or nine months after launch. Marketing was awkward. Handing out pamphlets outside restaurants on Jyoti Nivas College Road was, in Majety's words, "a bit weird", and it did not take off much. What changed things was a friend at Flipkart who shared the site on an internal email group. By day three, Majety recalled, 35 people were looking at it, and the team thought, "WOW, this is big!" He has been honest that there was "no way to tell" from day one that it was working. The only signal was customers ordering again. He credits the first core team more than himself. Early hires were thrown in as generalists, and he has said the company "would have just crumbled" without them. Letting go was hard, he admitted, when you are still young. The hard years were still ahead. In 2020 the pandemic forced about 1,100 layoffs and the closure of more than three-quarters of Swiggy's cloud kitchens, and in January 2024 around 400 more people went. In November 2024 the company listed at ₹390 a share. For the year to March 2025 it lost ₹3,107 crore on revenue of ₹16,333 crore, and the stock hit a record low in October 2026 on worries about cash burn from its Instamart bet. For a first-time founder, Swiggy's early story is a quiet one: a failed company, a coder brought in to fix the gap, and a first week measured not by headlines but by whether the same people ordered twice. ## Paytm (India) URL: https://startup.talk/india/paytm/ Founders: Vijay Shekhar Sharma Founded: 2010 (August); Based in: Noida, Uttar Pradesh; Sector: Fintech / digital payments At fourteen, Vijay Shekhar Sharma finished Class 12. At fifteen, before he had turned sixteen, he joined Delhi College of Engineering as the youngest in his batch, a boy from a Hindi-medium school in Harduaganj, near Aligarh, who could barely follow what his professors were asking. He has said he "was cast out by people who were super smart". The class topper became a backbencher. He was the third of four children. His father was a schoolteacher, his mother a housewife, and the family lived what he has called "a normal life, under abnormal financial constraints." His mother told him to prepare hard for the engineering entrance tests, because the family could not afford a private engineering college. So he studied alone for a year, reading two books on the same subject at once, one in Hindi and one in English. When he met "momentum" in English, he looked up its Hindi word, "samveg". He chose to sit the exam in English, the language that was now his biggest weakness. In Delhi he taught himself. He wrote down every word he did not know from newspapers, then magazines, and bought used copies of Fortune at Sunday book markets in Daryaganj. Reading about Silicon Valley gave him hope that someone like him could build something without finishing a degree. While still in college he started a company, XS Corps, whose business card gave a hostel room as the address. A nearby shopkeeper took messages on his phone. At job interviews he would announce that he was not there for the job but had a company that could do it for them. The website indiasite.net, started in college, sold about two years later for US$1 million. Paytm itself began in August 2010 as a mobile recharge and bill-payment business, inside One97 Communications, a company incorporated on 22 December 2000. By 2009 One97 was already generating free cash every month, and an IPO planned for November 2010 was shelved at the last moment. When the Paytm wallet idea reached the board, Elevation Capital has said, members asked whether Indians would ever leave cash, and whether a team built for business clients could make consumer tech. The hardest stretch, he has said, was not the later regulatory crisis but the fundraising of 2013-14, when deals stalled and investors backed out at the last minute. He has said that in that period he figured few people would have noticed if the company had folded. Then the money came: Alibaba and Ant in 2015, SoftBank's $1.4 billion in 2017, Berkshire Hathaway in 2018. The wallet in 2014 and QR payments in 2016, just as demonetisation pushed millions to digital payments, made Paytm a household name. The fall was public. The November 2021 IPO raised ₹18,300 crore, then closed 27% down on day one. The Reserve Bank barred Paytm Payments Bank from onboarding in 2022, ordered most activity to stop in January 2024, and cancelled its licence in April 2026. He has said the action took a serious emotional and personal toll and admitted "we should have done better". The company posted a FY24 net loss of ₹1,423 crore. A founder can take this from him: the English he lacked at sixteen, the money his family lacked, the investors who walked away at the last moment. None of it was a reason to stop. But a licence you do not control can take away in a day what a decade built. ## PhonePe (India) URL: https://startup.talk/india/phonepe/ Founders: Sameer Nigam, Rahul Chari, Burzin Engineer Founded: 2015 (December); Based in: Bengaluru, Karnataka; Sector: Fintech / UPI payments In PhonePe's early days the team worked from what Sameer Nigam calls a "dingy office" outside HSR Layout. There was no parking, and people would come and deflate the tyres of their cars. He tells the story now as a reminder to stay scrappy. Nigam is "Mumbai born and bred". He met Rahul Chari as an undergraduate in Bombay in 1995, and Chari's wife was a junior of theirs in college. Around 2000-2001, just starting his career in Los Angeles, he met Burzin Engineer. Their first venture together was a digital-content distribution business, Mime360, founded in 2009 and bought by Flipkart in 2011, and before that a music business, and they knew little about the Indian market or the industry. They were, he says, "having a ton of fun building" anyway. It taught them that money in that industry does not come easily. The next lesson was about letting go. Their small startup had built a platform that powered Gaana and other companies, even though it made no money. "We were just married to the IP," Nigam has admitted. Binny Bansal gave them a month to decide. Within that month, downloads from ordinary consumers topped three years of work on the other side. It was a hard lesson in giving up a dream. Even the decision to sell to Flipkart was, in his words, "one of the hardest ones in life", though investors were eager to back them. The first meeting had been a leap of faith: a half-day conversation and a handshake with Binny, with the paperwork coming much later. After a long whiteboard pitch from Sachin Bansal, Burzin told the others they were "either just crazy or going to change the world". Rahul, meanwhile, was tired from the Flipkart years and said he wanted a very small team, like WhatsApp. UPI was a long shot, and Nigam put the odds at roughly "one percent". They took it anyway. The app went live in August 2016 with Yes Bank as its first UPI partner, and on 14 January 2017 ICICI Bank blocked PhonePe transactions. The regulator first told ICICI to restore access, then reversed itself, saying PhonePe had broken UPI norms. By February it was resolved. He is candid about the culture. He says colleagues argue about product "like cats and dogs", but never personally. During Covid the company kept salaries and bonuses intact. Today PhonePe has about 700 million users. In the year to March 2026 it earned operating revenue of ₹7,920.5 crore but lost ₹2,792 crore, and in March 2026 it deferred a planned IPO of about $1.3 billion. The reveal is in the 'married to the IP' moment. Many founders cling to the thing they built even when users are voting for something else. Nigam's team had a month to look at the numbers, and the numbers answered. ## Zerodha (India) URL: https://startup.talk/india/zerodha/ Founders: Nithin Kamath, Nikhil Kamath Founded: 2010 (August); Based in: Bengaluru, Karnataka; Sector: Fintech / discount stockbroking In the late 1990s Nithin Kamath worked in Weekender showrooms and tried multi-level marketing. At seventeen he was already trading stocks, managing his father's account. His father, a Canara Bank man, gave him the first seed money to trade with. Years later Nithin would say, "None of the success that's ascribed to us would have happened without my father," and call him "our VC". Profiles place the Kamaths in a middle-class Konkani family that moved around with the father's bank postings and settled in Bangalore in 1996. Nithin's early trading went well, then it did not. Interviews recount that after the dotcom crash, around 2001-02, he lost roughly ₹5 lakh, money he had built up by the time he finished engineering. He took a call-centre job to pay the bills and traded in the mornings. He has said those call-centre days were some of the most fun of his life. He crawled back. In the mid-2000s he became a sub-broker for Reliance Money, and when the market crashed again in 2008 he came out with enough to start something of his own. In 2010 he put together roughly ₹2 crore, part his savings and part from a friend in Singapore, and opened Zerodha with his brother Nikhil. Operations began on 15 August 2010, and the name joins "Zero" to "Rodha", the Sanskrit word for barrier. Nikhil, seven years younger, had been trading at 15 or 16, and Nithin realised early that his brother was "a better trader than I am". His first email about the venture set a target of 10,000 customers, after which, he wrote, they would have arrived in life. The first couple of years were slow. Nithin has said Nikhil's trading profits were what kept them in the game, and the company lost no money in that stretch. Then the surprise: nobody advertised. An early client who had seen Nithin's trading record handed him ₹25 lakh to manage. "I made money for him, and then more customers started getting on board," Nithin recalled. Word of mouth did the rest. Zerodha took no venture capital, and the Kamath family still owns all of it. Seema Patil has led quality since the beginning, and Karthik Rangappa wrote Varsity, the education programme, single-handedly. Kailash Nadh joined as CTO in 2013, and the in-house Kite platform made a flat fee sustainable. The price has been public embarrassment. Kite has buckled on volatile days and futures-and-options expiry days, and the company acknowledged a login failure in November 2020. Groww overtook it by active clients, reaching about 23% share by April 2024. Today Zerodha is a private company that is still large and unusually profitable: ₹8,847 crore revenue and ₹4,237 crore net profit in FY25, though both fell from the year before. Around the broking business the brothers built Rainmatter, True Beacon, a mutual fund house and a $1 million-a-year open-source fund. What stands out is how little of it was planned. A teenager managing his father's account, a young man who lost five lakh rupees and sat in a call centre, a brother whose trades paid the bills. Nithin's own summary is that "luck is a big part of any success in life." For a first-time founder the more useful part may be the order of events: he lost his own money first, and only then started a firm that charged other traders less. ## Razorpay (India) URL: https://startup.talk/india/razorpay/ Founders: Harshil Mathur, Shashank Kumar Founded: 2013 (incorporated 24 May 2013; product launched Oct 2014); Based in: Bengaluru, Karnataka; Sector: Fintech / payments and business banking Harshil Mathur once walked into an HDFC branch near his home and asked how a company could start a payment gateway. He was turned away. It was the first of many such conversations. By his own count, he was "mostly been rejected by almost 100 bankers" over three to six months. Shashank grew up in Patna, entered IIT Roorkee in 2008 for computer science and, after graduating in 2012, joined the Windows team at Microsoft. He quit in 2013. Harshil, from Jaipur, was a year behind him. They had met at IIT Roorkee, bonded over writing code in the evenings, and later built side projects together, including one called SDS Labs. Harshil took a well-paid job at Schlumberger in Abu Dhabi and quit within months. "I am a guy who wants to sit and code," he told Business Today. Their first idea was school fee payments in Jaipur, Harshil's hometown, which Shashank says had "zero product-market fit" because the schools thought they were in the admissions business. Before that Harshil had been building a crowdfunding site for medical emergencies, and trouble accepting payments for it led to Razorpay. The itch was practical: the two had struggled to accept online payments themselves, and founders in Bengaluru and Pune told them the same story. Then came the grind. Shashank, in Patna, found bank branch managers who had never heard of a payment gateway. He moved from t-shirts to suits for the meetings. One bank wanted a ₹25 lakh security deposit. His grandfather sold property to cover it, and in a sense became their first angel. A young bank executive, they have said, finally understood what they were building. What nobody expected, least of all the founders, was Y Combinator. They had applied with "zero expectations" and were accepted in March 2015, the second India-focused company in the programme after ClearTax. The two rented a one-bedroom apartment in the US and worked 16 to 18 hours a day. The company itself had been incorporated on 24 May 2013. The acceptance gave them credibility, and introductions to investors. Growth brought its own scare. When a gateway reseller suddenly cut off their customers, the team phoned every affected merchant themselves. "There were customers who literally abused us on phone, massively," Harshil recalled on the Blume podcast. He still reads complaints on social media personally. Now Razorpay is preparing for the public markets. Reports in June 2026 said it had confidentially filed for an IPO targeting $500-600 million at a $5-6 billion valuation, below its 2021 peak of about $7.5 billion. FY25 revenue rose 65 percent to ₹3,783 crore, and the core payments business turned EBITDA positive, but a one-time ESOP and restructuring charge pushed the net loss to roughly ₹1,206 crore. The story here is a hundred refusals, a grandfather's property, and a founder who left a comfortable salary because it was not for him. The lesson for a first-time founder is how ordinary the early proof was. Customers had the same problem. That, plus nearly a hundred "no" answers that each taught them something, was the whole business plan. ## CRED (India) URL: https://startup.talk/india/cred/ Founders: Kunal Shah Founded: 2018; Based in: Bengaluru, Karnataka; Sector: Fintech / credit card bill payments and rewards At sixteen, Kunal Shah was already earning his own money. The family had been hit by a financial crisis. His father worked in pharmaceutical distribution in South Mumbai, and profiles say the household needed side income. So he sold henna cones and pirated CDs, ran tuitions, and ran a cyber cafe out of home. Profiles describe him as financially independent as a teenager. He chose to study philosophy at Wilson College for a practical reason: the course had classes from 8 am to 10 am, which left the rest of the day free to work. He also worked as a data-entry operator and delivery boy while in college, according to Business Today. By one account he then spent about a decade as a programmer at TIS International. Later he enrolled in a part-time MBA at NMIMS and left in 2004, saying business was taught as scoring marks rather than understanding things. In 2009 he started PaisaBack, a cashback business. It stumbled into something better. A pivot of its model became FreeCharge, co-founded with Sandeep Tandon in 2010. In April 2015 Snapdeal bought it, in a deal reported at about ₹2,800 crore. One profile says Shah and Snapdeal's Kunal Bahl closed it in 22 days, without investment bankers. After the sale he spent a long stretch travelling and studying developed countries, noticing petrol pumps without attendants and self-checkout supermarkets. That trust-based thought became CRED, incorporated as Dreamplug Technologies on 19 April 2018. Kunal said that as a philosophy graduate with no alumni network, he found it hard to hire his first two engineers. CRED had already raised $30 million from Sequoia and Ribbit before it announced itself on 26 November 2018, when the app had about 1,000 downloads and he wrote that credit cards carry "over 30 hidden charges". CRED was the second act. The pitch: people who pay credit card bills on time are valuable, so give them a members-only club with rewards. Critics saw no clear way to make money from it, and the numbers show the cost. A 2022 valuation of about $6.4 billion fell by 43 to 45 percent in a roughly $72 million round in 2025. FY25 operating revenue rose about 16 percent to ₹2,735 crore, and operating loss narrowed 51 percent to ₹298 crore. Two things stand out. A boy who sold CDs to help his family even chose his degree around his work hours. And his two big companies both grew from pivots of something smaller. First-time founders can note that the real idea often shows up after the first one has started to work, if you are still around to notice it. ## Nykaa (India) URL: https://startup.talk/india/nykaa/ Founders: Falguni Nayar Founded: 2012 (April); commercial ops 2013; Based in: Mumbai, Maharashtra; Sector: Beauty and fashion e-commerce / retail "I left Kotak on March 31 and on April 1, I was in the process of incorporating the company," Falguni Nayar told Business Today. After eighteen years in investment banking she was about to turn fifty, and she had no job to go back to. She was no stranger to hard work or to leaving places. A commerce graduate from Sydenham College in Mumbai, she went on to IIM Ahmedabad and joined Kotak in 1993. She helped set up the group's offices in London and New York, and her husband moved with her. By one account she sometimes came home at 2 am and flew out again at 7. What she wanted, she has said, was to run her own business before turning 50, because she knew building one takes time. She had spent years watching entrepreneurs from the other side of the table and wanted to be one. Not everyone cheered. In a Film Companion interview, she recalled that people around her were doubtful, and she has said she was quite scared about quitting. The early money was family money. The certificate of incorporation is dated 24 April 2012, and commercial operations began in 2013. The memorandum shows how small it was: 10,000 shares, with 9,998 held by Falguni and one each by Om Prakash Nayar and Rashmi Mehta. The first office was a house on Carmichael Road in Mumbai, before a move to Lower Parel that September. In December 2012 her husband Sanjay was allotted 209,998 shares and she took 200,000. Until May 2016 the only shareholders, apart from employees exercising options, were the Nayars and those two other signatories. The first options went to Vijay Gaddamanugu in 2014, and the first outside investors, among them Harindarpal Banga, Narotam Sekhsaria and Mala Gaonkar, bought in a private placement on 18 May 2016. What surprised the market was the order of things. A banker with no retail background bet on beauty, saying she saw Indian beauty consumption as being at a very early stage. She opened a first store at the Delhi airport in 2014, launched her own cosmetics brand in 2015, and added fashion in 2018. In 2019 Katrina Kaif launched Kay Beauty with the company. Success brought scrutiny. A 2019 API flaw in Nykaa Fashion could have exposed accounts, and the company fixed it. After the November 2021 IPO, which raised ₹5,352 crore at about a $7.4 billion valuation, a 5:1 bonus issue drew criticism, CFO Arvind Agarwal resigned, and SEBI looked at the allotment. In FY25 Nykaa reported revenue of ₹7,950 crore and a thin net profit of ₹72 crore. Business Today notes it took about nine years to reach profitability. She did not wait for a safe moment. Her lesson to a first-time founder is a date she gave herself, fifty, and the nerve to meet it, with family money on the line. ## PolicyBazaar (India) URL: https://startup.talk/india/policybazaar/ Founders: Yashish Dahiya, Alok Bansal, Avaneesh Nirjar Founded: 2008; Based in: Gurugram, Haryana; Sector: Insurtech / insurance marketplace In 2007, Yashish Dahiya was going through his father's finances when a few things looked off. On a closer look, he has said, his father had been cheated of a few lakh rupees by insurance agents who sold a policy that served them and the insurer, not the customer. Dahiya's own work was insurance. He had built First Europa, a European online insurance broker, so the son knew exactly how the product was supposed to be sold. He came to that moment by a winding route. One profile says that in 1996 he skipped IIM Ahmedabad's campus placements to give home tuitions, which paid better for about two hours of work a day, and that his mother then pressed him to take a job because his sister would struggle to find a groom otherwise. He took the job at Illinois Tool Works in India, went abroad for an INSEAD MBA in 2000, worked in consulting and then at ebookers. The company was incorporated on 4 June 2008 as ETECHACES Marketing and Consulting, and the 10,000 shares in its memorandum went to Avaneesh Nirjar (5,500) and Alok Bansal (4,500). Dahiya was not an original subscriber; he joined in September 2008 as trustee of the employee stock option trust, and became the face of what was later renamed PB Fintech. Info Edge was the first outside investor, putting in US$4 million in tranches that ran until 2011, roughly ₹20 crore in total. The first campaign was called "Ullu Mat Bano", a plea to stop being fooled by mis-selling. The early days were thin. Dahiya has said the founders played every role for a long time, and that early meetings with venture capitalists were tough. He has also said that without funding he would not have started, because the opportunity cost was too high. Then regulators changed the game. Rules cut the lead-generation and commission income the model depended on, and one account says revenue fell about 90 percent. The board reportedly urged him to shut the company. He and his team backed themselves to keep going, and a $20 million Series C followed in 2014. The next decade was funding and growing, with Tiger Global, Premji Invest and SoftBank, then a $1.5 billion valuation with Tencent in 2019, and a November 2021 listing that raised ₹5,625 crore. It has not been clean. A court imposed ₹10 lakh in costs on the company in a 2019 trademark dispute, IRDAI fined it ₹24 lakh in 2021, a data-access incident followed in 2022, and GST officials raided in January 2025. FY24 brought ₹3,438 crore revenue and ₹64 crore net profit. It starts with a son finding out that his own father had been misled, and a founder who knew the trade well enough to see how it had been done. A first-time founder can take that: the best idea is often a wrong you have personally watched happen. ## OYO (India) URL: https://startup.talk/india/oyo/ Founders: Ritesh Agarwal Founded: 2012 (as Oravel Stays; renamed OYO in 2013); Based in: Gurugram, Haryana; Sector: Hospitality / travel tech At one low point in Delhi, Ritesh Agarwal had Rs 35 in his bank account. Open magazine reported that he was sleeping on the stairs of a rented barsati because he had not paid the rent on time. He had come a long way from Bissamcuttack, a small town in Odisha's Rayagada district, where he grew up in a middle-class Marwari family. He was writing code as a boy. He once wanted to be an aircraft pilot, until his elder sister came back from an entrepreneurship festival and mentioned the word "entrepreneur". He looked it up in a dictionary, and from then on that was his answer when teachers asked what he wanted to be. By 13 he was selling SIM cards and FMCG products. After Class X his parents sent him to Kota to prepare for IIT. He got bored. He started travelling to Delhi for entrepreneurship events, and when he could not afford the registration, he sometimes sneaked in. His Kota coaching school was lax about attendance, so he worked for entrepreneurs in Delhi and returned only for exams. His parents did not know for a long time. He still scored well enough to keep them reassured. After Class XII he enrolled at a Delhi college and dropped out three days later. Convincing his parents was hard. He pitched it as a gap year, like students who take a year off to prepare for competitive exams, and promised to go back to college if the venture failed. He has described a childhood in which, as he told Open, his parents "usually decided what to wear, what to do". His first company, Oravel, began with about Rs 16,000. It listed budget stays, but bookings were few and almost nobody came back. Customers were unhappy. So he spent three to four months sleeping in more than a hundred bed-and-breakfasts, guest houses and small hotels to find out why. What he found was not a shortage of rooms but unpredictability. In 2013 he became, at 19, the first Indian Thiel Fellow, with $100,000 behind him, and OYO started with one hotel in Gurgaon. He often worked as housekeeping staff himself to hold the standard. Growth came fast and so did the damage. A former co-founder, Manish Sinha, accused him of a lack of transparency. A fight with Zostel over a 2015 term sheet ran until the Delhi High Court ruled in 2022. FY20 brought a loss of about Rs 13,122 crore, and the valuation slid from $10 billion to about $2.5 billion by June 2024. In FY25 the company reported revenue of Rs 6,253 crore and a profit of Rs 245 crore, and its parent PRISM won SEBI approval in June 2026 for a Rs 6,650 crore issue. There is a line from that early Open profile worth keeping. Agarwal admitted he sometimes wonders whether he would hire himself, and that his answer is no. His first product only improved after he went and stayed in a hundred of the hotels himself, which is a harder way to learn than a spreadsheet and a more reliable one. ## Ola (India) URL: https://startup.talk/india/ola/ Founders: Bhavish Aggarwal, Ankit Bhati Founded: 2010 (3 Dec 2010; cab service launched Jan 2011); Based in: Bengaluru, Karnataka; Sector: Ride-hailing / mobility Bhavish Aggarwal's first company did not sell a single trip. Olatrip.com, the travel-planning site he started with his IIT Bombay batchmate Ankit Bhati in December 2010, began with about Rs 2 lakh in the bank. Outlook Business later reported that it failed to sell a trip, which is why Ola became a cab business. He grew up in Ludhiana in an upper-middle-class Punjabi home. Both parents are doctors, a father who is an orthopaedic surgeon and a mother who is a pathologist, according to a FactorDaily profile. He failed the IIT entrance exam the first time, then came back the next year with an all-India rank of 23. When he told his parents he wanted to start a travel company, a Business Insider India headline records their reply: "why don't you open a travel agency?" His father was puzzled that an IIT graduate wanted to run a travel business. Aggarwal reportedly told him not to worry, he would figure out a path. The story people tell about the pivot is a bad taxi ride. As told on India.com, a driver on a Bengaluru to Bandipur trip demanded more than the agreed fare and left him on the road after an argument. Accounts of that trip differ, and he is not recorded telling it here. The first years were thin. He later described them as figuring out the next month's cheque, with a team of 10 to 20 people and few customers. In an early investor meeting in 2012, according to FactorDaily, he was stumped by a question about Ola's organisational structure and quickly looked the term up while the investor was distracted. His father was almost a co-founder, holding the same number of shares as his son before he stepped back. Then Ola won, and winning got complicated. Uber's Travis Kalanick reportedly told Ola's investors in January 2016 that Uber would run Ola into the ground. Aggarwal reportedly turned down a larger SoftBank offer at a lower valuation, took less money, and amended the company's articles to limit investors' share transfers. Tiger Global's Lee Fixel left the board in 2017 after a dispute with him. FactorDaily counted more than a dozen senior leaders leaving in 36 months, though most of that picture comes from anonymous insiders, since Aggarwal declined to be interviewed. In 2020 Ankit Bhati left the company; the company denied any rift. By April 2024 Ola had exited all international markets. FY23 revenue was Rs 2,799 crore against a net loss of Rs 772 crore. Aggarwal has contrasted his parents' generation, focused on basic security, with his own: "There has to be one generation that does tapasya." The first idea went nowhere, and the second came from a real irritation. That order of events is more common in founder stories than the tidy version admits. ## Ola Electric (India) URL: https://startup.talk/india/ola-electric/ Founders: Bhavish Aggarwal, Anand Shah Founded: 2017; Based in: Bengaluru, Karnataka; Sector: Electric vehicles / manufacturing In August 2024 Bhavish Aggarwal took Ola Electric public with a Rs 5,500 crore issue, the first listed EV company in India, as he put it to BW Disrupt. It was the kind of moment founders picture for years. But it came after a rough run, and the road there was not smooth. He has contrasted his parents' generation, which wanted basic security, with his own: "There has to be one generation that does tapasya." His parents are doctors in Ludhiana. Before Ola Electric there was Ola Cabs, and before that a first company that did not sell a single trip. Ola Electric began in 2017 as a pilot in Nagpur with electric vehicles for Ola's fleet, and later became a separate company. Ratan Tata invested in the 2019 Series A, and SoftBank put in $250 million that July. The first scooter rolled out in August 2021 from the Tamil Nadu plant the company called the Future Factory. YourStory carried Aggarwal's own explanation for hiring women to run it: if women can win in cricket, why not in a factory. Deliveries began in December 2021. Within weeks scooters caught fire in Pune and Thiruvananthapuram, and more fires followed, enough for the government to worry about the risk to the whole EV industry. In April 2022 a Pune fire led to a recall of 1,441 scooters. In 2023 the company replaced front fork arms for free after safety criticism. When Rajiv Bajaj mocked EV makers, Aggarwal replied that he "put ICE cubes in my drink." The human cost showed in the culture. A 2022 Bloomberg report described his behaviour as abrasive and the workplace as hostile, which the company disputed. The prospectus showed attrition of 47.48 percent in FY23, which he attributes to junior roles. At least 13 senior executives had left in five years, per a profile in Outlook Business, which also noted supporters who say he simply pushes hard. The numbers moved with it. Revenue was Rs 5,009 crore in FY24, against a net loss of Rs 1,584 crore. By July to September 2025, revenue was about Rs 690 crore, down 43 percent, with a loss of Rs 418 crore. In October 2025 an FIR was filed against Aggarwal and executives over an employee's suicide, which the company is contesting. A founder can build a factory faster than he can build trust. The fires, the recall and the complaints were not strategy problems. They were what happens when customers meet the product, and they sit on the founder's shoulders long after the listing day. ## Byju's (India) URL: https://startup.talk/india/byju-s/ Founders: Byju Raveendran, Divya Gokulnath Founded: 2011 (app launched 2015); Based in: Bengaluru, Karnataka; Sector: Edtech In October 2024 Byju Raveendran stood in front of reporters and called his own company worth zero. It was a long way from the stadiums where he once taught maths in stadiums that held up to 25,000 students. He grew up in Azhikode, a coastal village near Kannur in Kerala. Both his parents were teachers at the local government school where lessons were in Malayalam. His father, Raveendran, taught physics and his mother, Shobhanavalli, taught maths. He asked questions about everything. On a train, he later told a Harvard Business School case, he tried to predict its speed by counting electric poles. He preferred the sports ground to the classroom and often skipped school. Teachers complained to his parents, and his parents never stopped him from playing. He says he learned spoken English by listening to radio sports commentary. His parents wanted him to be a doctor. He chose engineering because medical school would have left no time for sport. After graduating from the engineering college in Kannur in 2000 he joined a shipping company as a service engineer. In 2003, on vacation, he helped friends prepare for the CAT entrance exam for the IIMs. Four of his twelve friends cleared it. He has said he took the test himself and turned the IIM offers down. He had found what he was good at. "Otherwise I was an introvert," he told Rest of World. "But the moment I started teaching math, I saw the value that I was giving." Think & Learn was founded in 2011, and the app launched in 2015. Money and acquisitions followed in 2020 and 2021, and in March 2022 an $800 million raise valued it at $22 billion. The unwinding came in public view. Byju's earnings report arrived more than a year late. The promised money from Sumeru Ventures never came. The workforce, about 60,000 in 2022, fell to about 14,000 by early 2024. A former senior executive told Rest of World that he had spent the best years of his life building the company, and that "a lot of us couldn't believe it was happening." Byju blames growing "too soon, too fast", denies the theft allegations over a $533 million loan transfer, and has said he has no regrets and never started it for money. He told the outlet he moved to Dubai for his father's medical treatment. FY22 brought revenue of Rs 5,298 crore against a loss of Rs 8,245 crore. The US unit filed for Chapter 11 in February 2024 and Indian insolvency proceedings began in July 2024. The app was removed from Google Play in May 2025 over unpaid AWS bills. He found his gift in front of friends in a room, teaching. Scaling a gift into a company needed audits, controls and a board, and the money arrived faster than any of those did. ## Meesho (India) URL: https://startup.talk/india/meesho/ Founders: Vidit Aatrey, Sanjeev Barnwal Founded: 2015 (December); Based in: Bengaluru, Karnataka; Sector: E-commerce / social commerce In June 2017, Vidit Aatrey sat in a restaurant in Bengaluru with an investor, Mukul Arora of Elevation Capital. Meesho had two months of runway left. He had been meeting investors for months without success. And he had a choice that evening: meet another investor who seemed more likely to convert, or meet Arora, who had been talking to him for more than a year. He cancelled the other meeting. He recalled later that he had chosen on the basis of who had gone deepest: "I don't remember any other fund that went so deep in speaking to our resellers, speaking to our users." His words to Arora were, "So if I could get one person, I really wanted to be you." Aatrey comes from a farming family in Uttar Pradesh, sugarcane farmers, Outlook Business reports, with Hapur named as the family's place. His father is a quality control officer at the Delhi Jal Board and wanted him to become an IAS officer. His mother warmed to his entrepreneurship before his father did. His father disagreed for a long time and came round after seeing his son in the news. He studied at IIT Delhi, then worked as a management trainee in ITC's packaging unit in Chennai and at InMobi from 2014. His first company, Fashnear, was a hyperlocal fashion marketplace. It did not work. A later attempt to help small businesses open online shops also failed to gain traction. Then he and his co-founder Sanjeev Barnwal, both from middle-class families that had bought from small businesses all their lives, noticed people selling on WhatsApp. "Nobody, anywhere else in the world, was selling products on WhatsApp," he said. Meesho began in December 2015. Investors were not convinced. YourStory's headline captured their first reaction: the model sounded like a marketing gimmick. Aatrey said the company spent more than a year raising its Series A. The hard times did not stop there. When the pandemic hit and non-essential sales were banned, Aatrey remembered that "for three months, essentially, you could not sell anything on the platform." In 2023 Meesho had to delist 42 lakh counterfeit listings, and in 2024 it blocked 22 million suspicious transactions. It listed on 10 December 2025 at about Rs 161 against an issue price of Rs 111, in an issue of Rs 5,421 crore. For FY26 the company reported revenue of Rs 13,099 crore and a net loss of Rs 1,358 crore. What won Meesho its first big cheque was one investor spending a year talking to resellers and users while others judged the idea from a distance. Founders can take that as a rule for picking investors too. ## Lenskart (India) URL: https://startup.talk/india/lenskart/ Founders: Peyush Bansal, Amit Chaudhary, Sumeet Kapahi Founded: 2008 (19 May, as Valyoo Technologies; Lenskart.com launched Nov 2010); Based in: Gurgaon, Haryana; Sector: Eyewear / omnichannel retail and D2C A venture capitalist asked Peyush Bansal for a business plan. He did not have one. In a conversation with Blume Ventures, Bansal recalled that his team said they could prepare it, and the investment was signed within about a week. That was the company at the start: 14 freshers working out of the garage of his parents' house in Greater Kailash, New Delhi. Bansal was born in New Delhi in 1985 and went to Don Bosco, then to McGill in Montreal for electrical engineering. He landed at Microsoft in the United States, work he has described as solving big problems. In 2007 he quit that well-paid job and came home. What pulled him back was a line he had read: India is "the blind capital of the world." Many people who needed glasses simply did not wear them. He has said his mission is to help make 50% of India see clearly. Yet the first things he built had nothing to do with eyes. Valyoo Technologies, registered in 2008, began with a college classifieds site, then a US e-commerce play. Lenskart.com arrived in November 2010, selling contact lenses. Then came the years of too many bets. Investor advice to copy Titan's playbook produced Watchkart, Bagskart and Jewelskart. Eyewear took only about 30% of the team's effort, even though the other lines brought in most of the revenue. It was Ronnie Screwvala who suggested he concentrate on glasses. Bansal chose Lenskart, which meant giving up about 80% of revenue almost overnight. By the end of 2014 the side businesses were closed. The next wall was fear. Many customers were afraid of eye tests, and people browsed the website without buying. His answer was to open stores and offer free eye check-ups, and over time to build its own factory and store network. SoftBank's $275 million in December 2019 made it a unicorn. Success brought its own trouble. Franchise owners in Karnataka filed an FIR in October 2024 alleging unfair practices, including company stores opening near theirs; the High Court stayed the probe in January 2025. In April 2026 an internal style guide set off a social media storm, and Bansal called it a training draft. Lenskart listed in November 2025 at ₹402 a share, opened soft near ₹390, and by May 2026 had a market value of about ₹86,800 crore. FY25 revenue was ₹6,652 crore. The hardest decision in its history was not what to build but what to give up: a company that kills 80% of its own revenue to bet on the one thing it believes in. ## Dream11 (India) URL: https://startup.talk/india/dream11/ Founders: Harsh Jain, Bhavit Sheth Founded: 2008; Based in: Mumbai, Maharashtra; Sector: Fantasy sports / sports technology / online gaming Harsh Jain says he heard "no" close to 150 times from investors before one said yes. He told an industry summit in Delhi in November 2019 that he kept going until Kalaari Capital agreed to invest in 2014. It had begun years earlier as a friendship. Jain and Bhavit Sheth were school friends who bonded over sports and FIFA. Jain went to high school in the UK in 1998, where he discovered fantasy sports and got Sheth and others hooked. He briefly worked at Microsoft, then joined the family business; Sheth had short stints at Godrej and a financial research firm. Both were unhappy and wanted to work in sports, gaming and tech. After finishing engineering in 2007, Jain pitched a startup just as the IPL was launching. His friends nodded, but only Sheth joined him. They started with a free, ad-supported site with fantasy sports at its core. Everyone around them, Jain has said, thought it a bad idea, and they were too stubborn to listen. The ad model failed because Indian ad rates were far lower than in the West. For about a year and a half they burned through money from friends and family, and Jain has admitted they were too stubborn to tell their families they had been wrong. Survival came sideways. They started a social media agency, won Mumbai Indians as a client, and used its cash to keep Dream11 alive until it found its footing. Banks refused to open accounts, Jain recalled, because they thought gaming meant gambling. During his MBA at Columbia Business School, American VCs told him to pitch in India instead. Sheth later said the biggest lesson was persistence: they got better with every rejection. What changed was a conversation. In 2014 Sheth met Vani Kola of Kalaari Capital, who backed them despite not being a sports fan, because she saw a capable team and a large market. Sheth said the money mattered less than the signal that they were on the right track. Kola also taught Jain to stay calm in a funding round. A 2017 court ruling that the game was skill-based eased what Jain called one of the most stressful stretches, when people questioned whether the business was legal at all. Users grew from 1 million in 2014 to 45 million by 2018. The same legal ground shook again. Karnataka suspended operations in October 2021, a suspension reversed in February 2022. On 21 August 2025 Parliament outlawed real-money games, and paid contests, over 90% of revenue, were paused. At its peak the company was valued at $8 billion. For a first-time founder, the lesson sits in the side business: the agency nobody planned bought the time the main idea needed. ## Freshworks (India) URL: https://startup.talk/india/freshworks/ Founders: Girish Mathrubootham, Shan Krishnasamy Founded: 2010; Based in: San Mateo, California, USA (founded in Chennai, Tamil Nadu); Sector: SaaS: customer support, CRM, IT service management On 8 October 2010, Girish Mathrubootham quit his job at Zoho and told his wife the same day, with no warning. He had two young sons and a recently booked apartment, and for weeks he had kept the idea even from her. He was born in Trichy in 1974. His father was a retired bank officer, and when Girish missed a merit seat in engineering college, his father borrowed money from relatives to pay for it. After HCL in the US he returned to India and took a ₹5,000-a-month job at AdventNet, later Zoho, over a ₹12,000 offer elsewhere, because it was in software. He spent nine years at ManageEngine, Zoho's division, rising to VP of Product Management. The trigger was small. In May 2010 he was reading Hacker News when Zendesk raised prices and a commenter said someone could build a better product. "That comment was like a slap on my face," he later wrote. Then came the weeks of stress. "I couldn't sleep because of the excitement and the fear," he said, naming school fees and the home loan. Shan Krishnasamy, a friend and gifted engineer, agreed immediately when Girish brought him in. Nights and weekends proved unsustainable with young families, so both resigned that October. The numbers were tiny: six people, a $5,000 monthly burn, salaries at 30-50% of market, rent of about $160, and no pay at all for Girish. A logo and website cost around $450. The turn came from a broken TV. Years earlier a shipped 40-inch set had arrived damaged and the insurer left his claim unpaid for five months, until he posted about it on an online forum and the insurer's president apologised within a day. "I was no longer the weak customer," he said. He understood the power of speaking in public. When Hacker News carried his post about leaving his job, top venture firms called. Accel offered a million-dollar term sheet. "Somebody external trusted us with a million dollars," he said. "We have to be onto something." Not everything was clean. In 2012 Zendesk's CEO publicly called Freshdesk a rip-off, which Girish answered calmly and later said brought more visibility than advertising could buy. Zoho, his old employer, sued in 2020 over alleged IP theft; it settled in 2021. Freshworks listed on Nasdaq in September 2021, raising about $1.03 billion, and in 2025 posted $838.8 million in revenue and its first GAAP-profitable year. Girish handed the CEO role to Dennis Woodside in 2024. The lesson is plain. The idea arrived as a customer's anger, but what turned it into a company was a friend who said yes at once and two men willing to work for months on a fraction of their pay. ## Zoho (India) URL: https://startup.talk/india/zoho/ Founders: Sridhar Vembu, Tony Thomas, Sreenivas Kanumuru, Kumar Vembu, Sekar Vembu, Shailesh Kumar Davey Founded: 1996 (as AdventNet; renamed Zoho 2009); Based in: Chennai, Tamil Nadu (also Austin, Texas for US operations); Sector: Enterprise and SMB software / SaaS Sridhar Vembu has written that Zoho's networking-software customers once shrank from over 150 to three. It was the dotcom bust, and some co-founders were leaving to pursue their own projects. He grew up in a village in Thanjavur district, Tamil Nadu, born in 1968. He studied electrical engineering at IIT Madras and then went to Princeton for his PhD, finishing in 1994. In the early 1990s he was at Qualcomm in California, staying with his brother Kumar. In his own telling, shopping trips bothered them: shelves full of Taiwanese goods and nothing from India. In 1996 they started AdventNet. In an interview with Sridhar Kumudam, he recalled that their parents were worried that two of their sons had come home without a job. The bust was the near-death. Secondary accounts report that he turned down an offer of about ₹150 crore to sell, and went without his own salary for four years, paying people from the company's cash reserves. What he draws from it, in his own words, is "we cannot rely on one market." That led to ManageEngine in 2003 and Zoho.com in 2005. The refusal of venture money came from a different moment. A VC once offered investment with a clause requiring an exit or liquidity guarantee within a set period. When he questioned it, he was told it was a standard clause. He could not promise an exit, so Zoho stayed bootstrapped. He now calls that a "fortunate moment" and says money always comes with strings attached. Then he did the unfashionable thing. In 2019 he left the Bay Area for rural Tenkasi, where Zoho opened offices in villages like Mathalamparai. In January 2025 he stepped aside as CEO for Shailesh Kumar Davey and became Chief Scientist. FY2025 revenue was ₹12,313 crore, net income ₹3,191 crore, with over 120 million users. Trouble hasn't left him. The 2020 IP lawsuit against Freshworks, founded by former Zoho staff, settled in 2021. In 2026 Zoho reportedly froze salary hikes to avoid layoffs, and a California court ordered him to post a $1.7 billion bond in his divorce case. He also has a son with autism. For founders, the striking thing is how ordinary the decisive acts were. Not a clever pivot or a famous raise, but a refusal to sign terms he could not honour, and a willingness to live on the company's own cash until customers came back. ## Groww (India) URL: https://startup.talk/india/groww/ Founders: Lalit Keshre, Harsh Jain, Ishan Bansal, Neeraj Singh Founded: 2016; Based in: Bengaluru, Karnataka; Sector: Fintech: stockbroking, mutual funds, investing platform "You can't do anything outside while you are still there," Lalit Keshre has said of Flipkart, where he was a Group Product Manager. So in 2016 he quit. Before that, there was a village. Reports place his birth in Lepa, a small village in Madhya Pradesh, in a farming family. The village had no English-medium school, so he was sent to Khargone to live with his grandparents. After Class 12 he learned about the Joint Entrance Exam and won a seat at IIT Bombay, where he finished a dual degree in electrical engineering in 2004. In college he began building a medical device, and later made video tutorials to help students prepare for IIT-JEE. At Flipkart he led Flipkart Quick. The idea for Groww grew from his own frustrations. He had invested in stocks since college and found Indian financial products "commission-centric." In 2016-17, he said, only about 20 million of 300 million middle-class Indians invested in equities and mutual funds. He and Harsh Jain, Ishan Bansal and Neeraj Singh, all from Flipkart, set out to fix it. Reports say the founders' own difficulty buying financial products, which they found opaque and complex, inspired the company. They started as a direct mutual-fund platform with three funds, and about six months in dropped extra features to keep it simple. When they took the idea to people early on, Keshre recalled in 2022, "they used to laugh at us." He says it made them more motivated. The product launched in May 2017. In November 2025, Groww went public at ₹100 a share with a ₹6,632 crore issue subscribed 17.60 times, listing at ₹112 on NSE. Reported FY25 profit was about ₹1,824 crore, against a loss near ₹805 crore the year before. Promoters held 27.98% before the issue. What they do show is a pattern: a boy sent to another town for a school, a product manager who knew he could not build a company on the side, and a team that took laughter as fuel. For founders, the take-away is small and concrete. Leaving the day job was not the heroic bit. It was accepting, in his own words about the old job, that nothing else could be done on the side. ## Physics Wallah (India) URL: https://startup.talk/india/physics-wallah/ Founders: Alakh Pandey, Prateek Maheshwari Founded: 2016 as YouTube channel; incorporated as a company 2020; Based in: Noida, Uttar Pradesh; Sector: Edtech (test prep, online and offline) Over tea at a roadside stall in Kanpur, Alakh Pandey's sister asked him a question meant to sting. If you quit engineering, she said, will you end up selling tea? He was in his third year at Harcourt Butler Technical University, studying on a loan, and she had travelled to talk him out of walking away. His answer, as it has been told, was that he would become a PhysicsWallah instead. The name came out of an argument, but the stubbornness behind it was built much earlier. Pandey grew up in Allahabad, now Prayagraj, in a family of four that lost its house when bad times came and ended up in a slum. His mother had severe asthma. He remembers waiting with his sister outside government hospitals, sleeping on pavements, because those were the only places the family could afford. His father failed at most of the businesses he tried. Relatives pooled money so the children could stay in school, and by Class VIII he, his sister and their mother were tutoring younger students to keep the household going. He wanted to crack the IIT entrance, but coaching fees were out of reach. That gap is the whole company in miniature. When he left college after about two years, convinced the classroom added little, his mother supported him. His sister did not. His father thought the YouTube plan was risky. Pandey has said his father and a girlfriend both left him around then, and that physics is what kept him going. So he taught the way he knew how: a marker, a whiteboard, a phone. The channel reached 10,000 subscribers within a year. Then 2020 arrived, with students stuck at home, and PW launched an app. On the first live class more than 40,000 students logged in at once, and the servers crashed. Pandey has said they never expected a response that big. What nobody expected was how quickly the teacher would become a corporate. Prateek Maheshwari, the co-founder who thinks in return on investment and spends on marketing only when it brings enrolments, helped turn a teacher's channel into a company. By June 2022 PW was a unicorn. Offline centres and acquisitions followed. The cost showed up in public. In 2023 former teachers alleged that Kota students were not getting an education worth their fees, and PW dismissed the claims as a distraction. That November it laid off around 150 people. In November 2025 PW listed on the stock exchanges with a ₹3,480 crore issue, and the founders each hold just over 40 percent. For FY26 it reported revenue of ₹3,899.54 crore and cut its net loss to ₹24.17 crore. The lesson for a first-time founder is hidden in that tea stall. Pandey did not search for a market; he built the thing he himself could not afford, and he said it out loud to the one person most likely to doubt him. ## Zepto (India) URL: https://startup.talk/india/zepto/ Founders: Aadit Palicha, Kaivalya Vohra Founded: 2020 (KiranaKart) / July 2021 (rebranded Zepto); Based in: Bengaluru, Karnataka (moved from Mumbai in 2024); Sector: Quick commerce / grocery delivery In the middle of the 2020 lockdown, two teenagers sat behind a WhatsApp group and tried to deliver groceries to their neighbours. Aadit Palicha has said they were 17. Neither had finished a university term. Both had been admitted to Stanford. Palicha grew up in Mumbai, fascinated by technology, and at 17 he had already built GoPool, a student carpool app in Dubai that never scaled. Kaivalya Vohra spent most of his childhood in Dubai, where he was deputy head boy at Dubai College, and was admitted to Stanford for computer science. Both attended orientation online, then came home within months. Vohra's parents have been described as among the earliest to believe in the idea. The first company, KiranaKart, was built around neighbourhood kirana stores. It sounded smart and it failed. Delivery times were inconsistent, quality was unpredictable, and the founders could not fix what they did not own. After about ten months they shut it down. What they did next was less glamorous and much harder. They took over the warehouses and the delivery themselves. The investor Contrary, which backed them in early 2021 with $730,000, had reportedly been willing to fund them only if they dropped out of Stanford. Y Combinator followed. In July 2021 they relaunched as Zepto, and by October it had a $60 million Series A. The valuation went from $2.5 million to $7 billion in under five years. Speed had a price. Anand Mahindra called quick-commerce delivery inhuman for the people doing it in 2022. Funding froze in 2023 when Silicon Valley Bank collapsed and trapped operating cash. Aadit has acknowledged early hiring mistakes. From 2024, regulators and users questioned dark patterns and drip pricing, and in April and May 2026 both founders were summoned by the Enforcement Directorate over foreign exchange matters tied to their foreign investors, a fact disclosed in the IPO filing. The company moved its headquarters from Mumbai to Bengaluru in 2024. It reported revenue of ₹22,624 crore for FY26, nearly double the year before, but losses of ₹5,905 crore. The IPO planned for July 2026 has been held off while it seeks about ₹1,000 crore in a pre-IPO round at a lower $4.5 billion valuation. The founders hold roughly 19.6 percent between them. The founders learned the hard way that the first idea is rarely the one that survives. They discarded a model that used other people's shops and took on the weight of running everything themselves. Growth that fast can look like proof. Zepto's story is still a bet on whether it is. ## boAt (India) URL: https://startup.talk/india/boat/ Founders: Aman Gupta, Sameer Mehta Founded: 2013 (Imagine Marketing); boAt brand launched 2014 / 2016; Based in: New Delhi / Gurugram; Sector: Consumer electronics / audio and wearables (D2C) An investor once told Aman Gupta and Sameer Mehta that he would put money in only if the two co-founders lived in the same city. They did not. They ran their company from different cities and, as they later told Entrepreneur India, held video calls even before the pandemic made that normal. They declined to move. Aman is a Delhi-born chartered accountant. He worked at Citibank and KPMG, then at Harman International, the audio group behind JBL. Sameer came from a trading and supply-chain background that taught him procurement, factories and logistics. In 2013 they started Imagine Marketing, first distributing Apple accessories in India. Distribution showed them something. Indian buyers had two choices: cheap, flimsy products or expensive global brands, and nothing sensible in between. The company launched boAt, starting with a charging cable. Early on, investors mostly passed, so the founders bootstrapped. Aman has described a business that forced them to be deliberate with spending. The numbers made the case. Revenue was about ₹226 crore in FY19, ₹609 crore in FY20 and ₹1,314 crore in FY21. Fireside Ventures, Qualcomm Ventures and an affiliate of Warburg Pincus came in. The founders still hold more than half the company, which they say gives them the upper hand in decisions, and they brought in Vivek Gambhir, who had led Godrej, as CEO and reported to him. Then Aman became a Shark on television and also an investor himself, with the pair backing about 50 startups. The lockdown, Aman said, gave the team the break it needed to think about what next. Sameer, for his part, flagged rising component prices and the push to depend less on China. The price was paid later. Revenue slipped for two years running, to ₹3,073 crore in FY25, and the wearables line shrank 40 percent to ₹330.4 crore. The company still turned a net profit of ₹60 crore that year. The IPO was cut from ₹2,000 crore to ₹1,500 crore, and in March 2026 Aman stepped back from daily operations to start OffBeat Studios. There is a founder lesson in that early investor condition. It sounded like a rule, but the founders say it was no longer necessary. Two people who split the work, one on the brand and one on the supply chain, built a company across two cities and still kept more than half of it. ## Mamaearth (India) URL: https://startup.talk/india/mamaearth/ Founders: Varun Alagh, Ghazal Alagh Founded: 2016; Based in: Gurugram, Haryana; Sector: Beauty and personal care (D2C, multi-brand) Ghazal Alagh has said the first product the couple wanted for their baby was easy to name and surprisingly hard to find. They wanted chemical-free care, and, in her words, they could not find options that were easily available in India. So they imported what they could, or asked friends travelling abroad to carry it back. It was slow. It was expensive. It nagged at them. Ghazal had come to this moment by an odd route. She studied for a BCA in information technology, then worked as a corporate trainer at NIIT. She took a course in figurative art at the New York Academy of Art and was recognised as an emerging woman artist of India. Before Mamaearth she ran dietexpert.in, a health startup that built customised diet plans.In 2016 they launched Mamaearth from Gurugram with baby products. Trust was the first problem. A new brand claiming to be toxin-free asks a mother to believe a stranger. So the Alaghs sought certification from a non-profit, MadeSafe, rather than asking people to take their word. Money was tight, which meant little advertising and a lot of word of mouth. The turn came from their own customers. Millennials, Ghazal has said, wanted the same safe products for themselves, so the company diversified into adult personal care. A baby brand became a house of brands. Fireside Ventures and Stellaris backed them early, followed by Sequoia, Sofina and Evolvence, and the company was declared a unicorn. Growth was not smooth. Honasa, the listed parent, was still posting losses a year earlier, including ₹18.56 crore in the September 2024 quarter. Then, in the quarter to September 2025, it reported a profit of ₹39 crore on like-for-like revenue of ₹566 crore, up 22.5 percent. Varun's headline line was that Mamaearth was back in the green. The Derma Co. now runs above ₹750 crore a year, and a prestige skincare brand called Luminéve launched exclusively on Nykaa. What stays useful for a first-time founder is how unglamorous the origin was. There was no market-sizing slide. There was a baby, an import bill and a friend with spare luggage space. The business started as a household problem that two people were willing to solve for themselves, and it only became a company when strangers turned out to have the same one. ## Delhivery (India) URL: https://startup.talk/india/delhivery/ Founders: Sahil Barua, Mohit Tandon, Bhavesh Manglani, Suraj Saharan, Kapil Bharati Founded: 2011 (May 2011, as SSN Logistics); Based in: Gurugram, Haryana; Sector: Logistics / e-commerce supply chain Sahil Barua called Bhavesh Manglani, a man he had met in Mumbai in 2009 through a mutual friend, and asked him to come along. Manglani, who had been introduced as a tech expert, said yes. That phone call is how a logistics company got its engineer before it had a single big customer. Barua was 29 when Delhivery began. He had graduated from IIM Bangalore and worked as a management consultant at Bain & Company, and Suraj Saharan and Mohit Tandon came from Bain too. Manglani and Kapil Bharati came from other organisations. Their reason for leaving consulting was simple and a little reckless. Barua later said it was the most interesting problem they could think of in the internet space. It started small. The company began in Gurugram in 2011 as a same-day delivery service for offline stores, and its first e-commerce client signed up in June. For the first six months, the founders have said, they bootstrapped with their own money. They won Chennai as their first city outside Delhi. By 2014 they were handling around 600 orders a day. Nothing about this was easy to sell. Investors needed convincing that a third-party logistics company could thrive in India. A rival, Chhotu.in, shut down after failing to raise funds, and an investor said of the sector that it was a difficult business. Delhivery was also dearer than the small operators and cheaper than Blue Dart and FedEx, caught between both. The founders were unhappy being only a delivery company, so they added fulfilment centres. Times Internet put in ₹6.6 crore in 2012, and in 2013 Nexus and Times added ₹35 crore. In 2019 SoftBank invested $413 million. By May 2022 the company listed, raising ₹5,235 crore at a valuation of about ₹35,283 crore. The founding team did not stay together. Reports differ on timing, but Manglani reportedly left in December 2020, and others followed by March 2021. The company went on to buy Spoton and then Ecom Express, and the integration cost it: Delhivery reported a loss of ₹50.37 crore in the September 2025 quarter, though revenue rose 17 percent to ₹2,559 crore. For FY25 it posted revenue of ₹9,372 crore and net income of ₹162 crore. The lesson: a start-up built on other people's customers has to win them one parcel at a time. Delhivery lasted because its founders kept refusing to be only a delivery company, and because the first hire was a call to someone they already trusted. ## Urban Company (India) URL: https://startup.talk/india/urban-company/ Founders: Abhiraj Singh Bhal, Varun Khaitan, Raghav Chandra Founded: 2014; Based in: Gurugram, Haryana; Sector: Home services marketplace / consumer internet The meeting happened at Hotel Grand in Vasant Kunj, Delhi, a day after three founders pushed their website live. Across the table sat investors from what was then SAIF Partners, now Elevation Capital. They had already sat through 55 to 60 local-services pitches without feeling convinced. This time, two interns had to step out of the room while the grown-ups talked money. The founders did not look like a typical Delhi startup. Abhiraj Singh Bhal and Varun Khaitan came out of IIT Kanpur and had worked in consulting at BCG; Raghav Chandra, with a computer science degree from UC Berkeley, was the one who could build the thing. Their idea came from an annoyance every Indian city-dweller knows. Finding a plumber, an electrician or a beautician you could trust meant calling a cousin or a neighbour, or gambling on an unorganised directory. Abhiraj and Varun had already tried one venture together. It folded in about six months. So UrbanClap began small. Their original thought, as the investors remember it, was that if Amazon could build online commerce for products, someone would build it for services. In the early days the team found service providers by hand and matched them with customers by hand. The office was 110 square feet in Bhikaji Cama Place. The company was doing around 40 orders a day. The hardest decision came about three years in. Roughly half of UrbanClap's revenue came from selling leads, a clean business that did not require owning quality. The founders concluded it would never produce the experience customers wanted, because the company could not control what happened after the lead was sold. They shut it down and moved to a full-stack model. Their investors recall the board presentation: no spin-off, no hedging. They simply stopped taking the money. That choice shaped everything that followed. Renamed Urban Company in 2020, it grew to the UAE, Australia and Singapore, became a unicorn in 2021 and listed on the NSE and BSE on 17 September 2025. Shares were issued at 103 rupees and opened at 162.25. The story since has been less tidy. Professionals have protested in Bengaluru over ID blocks, training fees and rating-based penalties, and the company's own full-stack promise puts it at the centre of every complaint a gig worker can make. In FY26 revenue rose about 36 percent to roughly ₹1,555 crore, but the company swung to a loss of about ₹235 crore as it spent on new ventures such as 15-minute home help. For a first-time founder, the lesson is uncomfortable and practical. The easy money in a business is sometimes the money that blocks the product you actually want to build. Urban Company's founders walked away from half their revenue while they still had a 110-square-foot office. ## FirstCry (India) URL: https://startup.talk/india/firstcry/ Founders: Supam Maheshwari, Amitava Saha Founded: 2010 (FirstCry.com launched Nov 2010); Based in: Pune, Maharashtra; Sector: Baby, maternity and kids products; omnichannel retail / e-commerce In 2010, Supam Maheshwari had a new baby and a problem. The diapers, clothes and toys he wanted for his child were not easy to find in India. So whenever he travelled abroad on business, he packed suitcases with them and brought them home. The idea for FirstCry started in those suitcases. Maheshwari had come a long way from a school called Apeejay and an engineering degree at Delhi College of Engineering, followed by a management diploma at IIM Ahmedabad. He was not new to building companies. In 2000 he and his friend Amitava Saha, who had studied at IIT Varanasi and IIM Lucknow, set up Brainvisa Technologies, an e-learning business. They sold it to Indecomm Global in 2007. The reported price varies by source, between about $16 million and $25 million. That money did not make FirstCry easy. In 2010, about 90 percent of baby products were sold in physical shops. Few Indians used credit cards online, and fewer trusted a website with something as tender as a baby's skin. The stock was awkward too. Maheshwari has said that selling anything from a 5-gram diaper pin to a 25-kg car seat takes real expertise in logistics. The company launched from Pune in November 2010. SAIF Partners backed it with $4 million in April 2011. What worked was admitting the internet alone would not be enough. In 2011 FirstCry began franchise stores in smaller cities. In 2013 it launched its own clothing brand, BabyHug. In 2016 it bought BabyOye from the Mahindra Group in a stock swap valued at ₹362 crore. Investors followed: Vertex, NEA, SoftBank with about $400 million, and later Azim Premji's investment arm. The price of that scale is visible in the accounts. FirstCry listed on the NSE and BSE on 13 August 2024, with revenue of ₹6,481 crore in FY24 and a net loss of ₹321 crore. In June 2025 former stakeholders filed an insolvency plea against its subsidiary GlobalBees, a reminder that acquisitions come with their own tails. In the September 2025 quarter, FirstCry reported revenue of about ₹2,099 crore and trimmed its net loss to roughly ₹50 crore. The lesson for founders is almost embarrassingly simple. FirstCry was not built on a market report. It began with one parent feeling a gap in his own home, and a co-founder who had already survived one company with him. The idea came first from need; the discipline of logistics and stores came later. ## MakeMyTrip (India) URL: https://startup.talk/india/makemytrip/ Founders: Deep Kalra, Rajesh Magow, Keyur Joshi, Sachin Bhatia Founded: 2000; Based in: Gurugram, Haryana; Sector: Online travel (OTA) On a Monday morning in the early 2000s, Deep Kalra learned how many of his people were willing to take less. He had asked his team of about 42 to consider pay cuts for senior staff, with stock in return. By the end of that weekend the company had 24 people. Kalra himself took no salary. He had not come to this from poverty. Born in Hyderabad and raised in Delhi and Ahmedabad, he studied economics at St. Stephen's College and then management at IIM Ahmedabad. He worked at ABN AMRO, helped set up bowling alleys for AMF in India, and joined GE Capital in 1999. That same year, a trip to Phuket with his wife made him see how the internet could change travel. The company he started in 2000 targeted Indians in the US who wanted to travel home. The money ran out quickly. eVentures had put in $2 million for roughly two-thirds of the company and then wanted out. Kalra bought the stake back for about ₹40 lakh, which he later described as the biggest negotiating lesson of his life. After the buyback there was money for about two months. One of the people he had to let go was his CFO, Rajesh Magow, who was his family's only earner. Kalra promised to bring him back once the company could pay him. Magow later put his own money in and became a co-founder. The next three years were not much easier. The dot-com bust and 9/11 meant cash rarely lasted beyond a couple of months. The team slept in the office and set daily ticket targets to hit monthly numbers. Keyur Joshi and Sachin Bhatia, who had joined as employees, took big pay cuts for stock and became co-founders as well. What held the company together was a narrow focus on NRI customers, who were already searching online even if most Indians were not yet booking. By 2010, MakeMyTrip was listed on the Nasdaq, raising about $80 million. A merger with Ibibo in 2016 brought Goibibo and redBus under one roof. It has not been calm since. In 2016 an indirect-tax case led to the arrest of the finance VP, which Kalra has called one of the lowest points of his life. In March 2026 a short-seller alleged the company still influenced hotel prices despite a 2022 competition ruling. In the year to March 2026, MakeMyTrip's revenue reached about $1.04 billion, while net profit fell to $51.8 million. The founder lesson is in that Monday. Kalra said he had to learn what the other side of a deal actually wants. People will take less for a company they own a piece of, but only if you take less first. ## InMobi (India) URL: https://startup.talk/india/inmobi/ Founders: Naveen Tewari, Mohit Saxena, Amit Gupta, Abhay Singhal Founded: 2007 (as mKhoj; rebranded InMobi 2008); Based in: Bengaluru, Karnataka; Sector: Mobile advertising / adtech Naveen Tewari has said he thought of quitting every day. He didn't. That tells you something about the first two years after he came back to India: he was in debt, early ventures had failed, co-founders had left, and his friends were buying cars and planning holidays. He came from a house with no business in it. His grandmother was a professor and his father, uncle and aunt worked in research and learning. Startups were never discussed. By his own account he privately asked himself whether life was just a repeat of earlier generations. He went to IIT Kanpur, then to McKinsey in the mid-1990s, as India was opening up, where he first saw what operating at scale looked like. The turn came at Harvard Business School. He met consultants, corporate managers and a small group of founders, and what he noticed was their quiet self-assurance. He has told an Entrepreneur India interviewer that this was the first time he thought there was a possibility of creation. He concluded entrepreneurs were not smarter than he was. They acted despite their fears. In 2007 he started mKhoj, an SMS-based search service, from an apartment in Bangalore. A year later it became InMobi, a name easier to say in English, and moved into mobile advertising. Kleiner Perkins invested in 2008. In 2011 SoftBank put in $200 million and InMobi became India's first unicorn. Tewari took it global early, when that was unfashionable. The years after were not smooth. Reported losses of about $45.5 million in FY15 and $40.9 million in FY16 led to layoffs of about 10 percent in 2016, which the company disputed. In 2018 it bought AerServ for $90 million. By FY23, InMobi reported revenue of $281 million and a net profit of $41 million. Then came the move few Indian founders have made: in 2025 Tewari and his co-founders raised about $350 million of debt to buy out SoftBank, lifting their combined stake above 60 percent. In 2026 the company was preparing an Indian listing, with reports of four banks appointed and a target valuation of $4 to 5 billion. What a first-time founder can take from Tewari is unglamorous. The pull to quit did not disappear when the company got big. He has said he did not stay because he was certain. He stayed because he decided persistence was part of the job, and then did it again the next day. ## Glance (India) URL: https://startup.talk/india/glance/ Founders: Naveen Tewari, Piyush Shah, Abhay Singhal, Mohit Saxena Founded: 2019; Based in: Bengaluru, Karnataka; Sector: Lock-screen content, consumer internet / AI commerce Naveen Tewari did not start Glance because InMobi was in trouble. He has said so himself. He was in the middle of rebuilding InMobi, the company that made him India's first unicorn founder, and he wanted something bigger: a consumer product, a 'complete internet platform', as he put it to Entrepreneur India. That ambition has an odd source. Tewari grew up in a family of researchers. His grandmother was a professor, and his father, uncle and aunt all worked in research and learning. Nobody discussed startups. He went to IIT Kanpur and then Harvard Business School, where, he says, he first saw entrepreneurs as ordinary people who acted despite fear. He thought of quitting InMobi every day. He didn't. Glance was the next bet. InMobi had spent more than a decade selling ads through other people's apps. Glance would put content directly where a phone owner looks first, the lock screen, and own the relationship with the consumer. The company launched in 2019 with $45 million from Mithril Capital. That same year it bought Roposo, a short-video app, to give the lock screen something to show. The next step was Google. In November 2020 Google and Mithril put in $145 million, valuing Glance at over $1.2 billion and making it the second unicorn in the InMobi group. In 2021 it bought Shop101, a social e-commerce company, and in 2022 it bought Gambit, an Indian gaming company. It also held Glance LIVE Fest, a three-day digital event. Reports in 2026 discuss an IPO for Glance's parent, InMobi. So what should a founder take from this? Glance is a case of a company started from a position of strength rather than need. Tewari had a unicorn, a team and a network of investors, and he used them to attempt something outside his core business. That is worth studying. It is also a reminder that without profit figures, even a billion-dollar valuation tells only half the story. ## ShareChat (India) URL: https://startup.talk/india/sharechat/ Founders: Ankush Sachdeva, Bhanu Pratap Singh, Farid Ahsan Founded: 2015 (Mohalla Tech incorporated 8 January 2015); Based in: Bengaluru, Karnataka; Sector: Social media / regional-language short video Thirty hours. That was how long Ankush Sachdeva said it took to code Moj, the short-video app ShareChat put out in 2020. He tweeted the number himself, and it landed like a dare to every rival who had spent months on the same job. Sachdeva and his co-founders, Bhanu Pratap Singh and Farid Ahsan, came out of IIT Kanpur, the kind of campus that sends its graduates to Bangalore and the Valley. They did something less obvious. Mohalla Tech was incorporated in January 2015, and the product they built around it, ShareChat, assumed that the next wave of Indian internet users would not be typing in English. They would want Hindi, Tamil, Telugu, Marathi, and a dozen more, and they would want to talk to each other, not to a brand. In 2016 ShareChat opened up to user-generated content, and the app began to look less like a messaging experiment and more like a town square. Money followed slowly, then fast. In September 2020 a group that included Twitter, Lightspeed and SAIF Partners put in $40 million. By April 2021 the company had raised about $500 million at a valuation above $2 billion. The fights came too. In 2018 ShareChat publicly accused ByteDance of copying its design for the Helo app. A Delhi High Court direction led ByteDance to change Helo's look. A small Indian company had stared down one of the largest internet firms in the world, at least on that point. The hard part was staying big. In May 2022 Google, the Times Group and Temasek invested $300 million at a $5 billion valuation, and ShareChat bought MX TakaTak from Times Internet for $700 million. Then the market turned. There were layoffs of 101 people in 2020, about 400 in January 2023, and about 200 in December 2023. In January 2023 Singh and Ahsan stepped down from their COO and CTO roles, leaving Sachdeva to carry the company he had started with two college friends. The company he leads now is smaller in ambition and sharper in arithmetic. In FY25 operating revenue was about ₹723 crore, nearly flat, while the adjusted EBITDA loss fell 72 percent to about ₹219 crore. Management traded growth for unit economics and pointed to microdramas, through its QuickTV platform launched in 2025, as the next bet. Reports in 2025-26 say it is preparing for a public listing. The lesson is not that speed wins. A 30-hour build gets you a headline. What kept ShareChat alive was the unglamorous decision, years earlier, to serve users that bigger platforms treated as an afterthought, and then the willingness, when the money stopped being free, to shrink the burn instead of the belief. ## Unacademy (India) URL: https://startup.talk/india/unacademy/ Founders: Gaurav Munjal, Roman Saini, Hemesh Singh Founded: 2015 (YouTube channel earlier; app launched December 2015); Based in: Bengaluru, Karnataka; Sector: Edtech (test prep, online learning) The video had 20,000 views, which for a student in Mumbai in 2010 felt like a stadium. Gaurav Munjal had shot it on a camera of about 0.02 megapixels, a Java tutorial for classmates who were struggling. He has often described the thrill of that number. Munjal was born in Rajasthan in 1990 and went to St. Xavier's in Jaipur. He then studied computer science at NMIMS in Mumbai, not at an IIT, and he has said that stung. School friends had made it into the IITs. He told interviewers he felt insecure, and that non-IITians do not get the safety net of a guaranteed job. When service companies came to his campus with packages of about ₹3.2 lakh a year, he had already promised himself in his second year that he would not take one. So he built things instead. With Hemesh Singh he made Flatchat, an app for finding flatmates, which was sold to CommonFloor. He worked at Directi. In 2015 he and Singh teamed up with Roman Saini, a man who had cracked the medical entrance at 16 and the civil services exam at 22. Saini served two years as an assistant collector in Jabalpur, then resigned to make UPSC preparation free online. He had once borrowed the coaching fee from a friend, and he said Unacademy was meant to bridge that gap. The app launched in December 2015. Sequoia India and SAIF Partners put in $11.5 million in 2017. In 2019 the company introduced a paid subscription, Unacademy Plus, and raised roughly $87 million, including a direct investment from Facebook. By 2021 it was valued at $3.44 billion, and it was buying other education companies one after another. Then came the bill. A data breach in 2020 exposed millions of accounts. Layoffs in 2022 and again in 2024 hollowed out the teams, and educators were put on notice as costs were cut. In 2023 a teacher was fired after a video urging students to vote for educated leaders went viral, pulling the company into a political storm. In 2024 Munjal announced zero appraisals in a video, wearing a $400 Burberry T-shirt, and the internet did not forgive him quickly. By FY25 revenue was around ₹820 crore against a loss of about ₹436 crore. In March 2026 upGrad signed a term sheet to absorb Unacademy in an all-stock deal reported to value it at about ₹2,055 crore, a fraction of its 2021 peak, with Munjal expected to stay on as chief executive. What a founder can take from this is uncomfortable. The same hunger that turns a 20,000-view video into a company can make a leader mistake growth for health. Munjal built a brand on proving the IIT crowd wrong; the harder lesson was learning when the proof was no longer the point. ## upGrad (India) URL: https://startup.talk/india/upgrad/ Founders: Ronnie Screwvala, Mayank Kumar, Phalgun Kompalli, Ravijot Chugh Founded: 2015; Based in: Mumbai, Maharashtra; Sector: Edtech (higher education, upskilling, study abroad) Before he sold a film studio to Disney, Ronnie Screwvala was a boy acting in professional Bombay plays after school. Othello. Death of a Salesman. He did it as a hobby, born into a Parsi family where his father worked as an executive at a British firm. He went to Cathedral and John Connon School and then Sydenham College in Mumbai. Then he did something stranger than theatre: he started a toothbrush manufacturing business. It was small and unglamorous and it taught him what it means to carry stock and chase customers. In 1981 he moved into cable television, working in an India where Doordarshan was the only terrestrial channel on air. That cable instinct became UTV, which grew into a studio, a games company and a content house. It listed in 2005. In 2012 Disney bought it entirely in a deal reported at an enterprise value of about $1.4 billion. Most people at that point would have stopped. Screwvala started again. In 2015 he joined Mayank Kumar, Phalgun Kompalli and Ravijot Chugh to build upGrad, an online education company aimed at working adults who wanted degrees, data skills and management training rather than the usual teenage test prep. It was a quiet category in a loud sector. It worked, though not without cost. In 2021 upGrad became a unicorn with a $185 million raise. In June 2022 it took in another $225 million at a valuation of $2.25 billion, and Screwvala himself put in $36 million. The company bought businesses and pushed into study abroad, universities and international markets. Chugh left in 2020, and Kumar exited in 2024. Losses ran large, and a mid-2020s workforce reduction followed. The numbers now look more orderly. In FY26 gross revenue crossed ₹2,070 crore and the net loss shrank by 52 percent to ₹130 crore. A confidential valuation report in May 2026 put the company at about $1.7 billion. In March 2026 it signed a term sheet to take over Unacademy, a rival worth far less than at its peak, and executives have said they are eyeing a public listing in 2027. The lesson is in the sequence. Screwvala did not trust a single big bang; he stacked small businesses, a toothbrush company, a cable network, a studio, until each taught him something the last could not. His fourth act is the one that shows what he learned: grow, then make the books honest. ## Udaan (India) URL: https://startup.talk/india/udaan/ Founders: Sujeet Kumar, Amod Malviya, Vaibhav Gupta Founded: 2016; Based in: Bengaluru, Karnataka; Sector: B2B e-commerce marketplace On 30 June 2026 a payment fell due that the company could not make. Udaan's offshore parent, Trustroot Internet, had issued $170 million of convertible notes, and it defaulted. Within days three creditors began insolvency proceedings in the Singapore High Court. Amod Malviya, Sujeet Kumar and Vaibhav Gupta had walked out of senior roles at Flipkart in 2016 to build a business-to-business marketplace for India's small traders. Their bet was plain. Kirana stores, chemists and general merchants bought stock from local wholesalers in a patchy, cash-heavy, slow system, and an app could sit between them and manufacturers. It grew like a bonfire. Udaan raised money from some of the biggest investors in the world, a total that has now passed $2 billion across 17 rounds. In FY22 its revenue neared ₹10,000 crore, and its loss was ₹3,076 crore. The company had been building warehouses, credit lines and delivery fleets, and burning cash to do it. Then the table was turned. When money tightened, revenue fell about 40 percent in FY23, then about 20 percent in FY25. The board and management cut costs and shrank the set of categories they chased. In June 2025 the company closed a $114 million Series G at a valuation 59 percent lower than before. Vaibhav Gupta, who leads the business, has been the face of the cost-cutting push. The repair has been partial. In FY26 operating revenue rose 12.6 percent to ₹5,134 crore and the loss narrowed to ₹852 crore. Management says it grew at roughly 25 percent a year for ten quarters, cut EBITDA burn by about 70 percent and improved margins by nearly 500 basis points. It is leaning on private labels, which make up 15 to 25 percent of staples sales, and on horeca360, its supply arm for hotels and caterers. Swiggy agreed to sell its LYNK distribution business to it for ₹500 crore in shares. Yet the auditors still flagged material uncertainty about whether the group can continue as a going concern, with liabilities exceeding assets by ₹791 crore. Udaan has said the Singapore proceedings will not materially affect Indian operations. For a first-time founder the point is a hard one. Scale is not the same as strength. A marketplace can double each year and still have no answer for the day the cheque stops. Udaan's story is being written now, and it asks whether the discipline came early enough. ## Infra.Market (India) URL: https://startup.talk/india/infra-market/ Founders: Souvik Sengupta, Aaditya Sharda Founded: 2016; Based in: Mumbai, Maharashtra; Sector: Construction materials marketplace and manufacturing (B2B) In May 2026, as Infra.Market lined up a final private round ahead of its listing, two of the investors in it were not funds. They were its own founders. Aaditya Sharda and Souvik Sengupta were putting fresh money into the company they started in Mumbai in 2016. That is a small detail, and it says a lot. For a decade the company has been selling the least glamorous things in the economy: ready-mix concrete, tiles, paint, bricks. Contractors and builders across India buy from it through an app and from its own plants. Nothing about it is flashy. Concrete does not trend, and tiles do not go viral. The business lives or dies on trucks arriving on time, on credit extended to builders and on the quality of what is poured. It turned a fragmented, cash-based market into a platform with names such as RDC Concrete, Millenium Tiles and Shalimar Paints. The shape of what they built is unusual for an internet company. Most marketplaces take a commission. Infra.Market also manufactures, which is why its revenue is large, nearly ₹20,000 crore by the estimates for FY26, and why its margins are thin, around 9 percent at the EBITDA level. Investors liked the story. Tiger Global, Accel and Nexus Venture Partners backed it, and in September 2025 a Series G of ₹732 crore valued it at about ₹24,600 crore. Over $740 million has been raised in total. Early in 2026 it added ₹1,250 crore in debt from Ascertis Credit. It filed its draft IPO papers confidentially in 2025 and got the regulator's clearance in January 2026 for an issue of up to ₹5,000 crore. There were wobbles. In FY25 profit reportedly fell by about 42 percent to roughly ₹220 crore as costs rose, even though revenue grew 27 percent to about ₹18,472 crore. Reports through mid-2026 describe caution about the timing of the listing, with the company targeting a window of four to six months. If the FY26 estimates hold, profit will have climbed back to between ₹300 and ₹325 crore. Those numbers are early estimates, not audited statements. The quiet insight for a first-time founder is that founders writing their own cheques into the last private round before a listing say something a pitch deck cannot. The cofounders have not explained their reasons publicly, but they chose to add to their stake rather than trim it. ## OfBusiness (India) URL: https://startup.talk/india/ofbusiness/ Founders: Asish Mohapatra, Ruchi Kalra, Bhuvan Gupta, Vasant Sridhar, Nitin Jain Founded: 2015 (company launched 2016 in some accounts); Based in: Gurugram, Haryana; Sector: B2B commerce (raw materials for SME manufacturers), lending (Oxyzo), SaaS Seventy-three times, investors looked at Asish Mohapatra's plan and said no. They were not rejecting an app or a fad. They were doubting the thing at the heart of the business: selling steel, polymers and other raw materials to small manufacturers on credit, and being paid later. Most venture capitalists, as the account of his fundraising goes, simply did not believe that model could make money. Mohapatra came from Odisha, then still spelled Orissa on the map, and got into IIT Kharagpur without any coaching. He went on to McKinsey and then into venture capital, a path that should have made him unusually good at pitching. In August 2015 he walked away from both to start OfBusiness. He did not begin with a theory of the market. He began, by his own telling in profiles, by sitting with balance sheets every day and learning how small manufacturers really worked. What he found was uncomfortable. Around 95 per cent of the transactions on his platform ran on credit, and he did not own the credit. That was the flaw in the model, and it was why the sceptics were right about the version they had seen. The turn came in December 2016, when a round of about ₹70 crore led by Zodius Capital let him build lending inside the company instead of renting it from others. The rejections, in hindsight, had pointed at the exact piece he had to build himself. By 2018 revenue touched ₹100 crore, and the company added tools like BidAssist and ProcureAssist to keep small buyers inside its ecosystem. Then the numbers went vertical. In August 2021, with Tiger Global leading a round of about ₹1,300 crore, OfBusiness became a unicorn, serving some seven lakh small businesses. It has not been smooth since. The IPO that was pencilled in for 2025 has slipped, and co-founder Nitin Jain left in December 2025 to start something new. The company has converted itself into a public company and is preparing a listing of $750 million to $1 billion, but the timeline has been deferred. The business, meanwhile, keeps earning. In FY26 it reported revenue of ₹20,645 crore and a net profit of ₹724 crore, up from ₹597 crore a year earlier, with its lending arm Oxyzo earning ₹375 crore. The lesson is hiding in those 73 noes. A rejection is not always a verdict on your idea. Sometimes it is a precise list of the part you have not built yet. ## Zetwerk (India) URL: https://startup.talk/india/zetwerk/ Founders: Amrit Acharya, Srinath Ramakkrushnan, Rahul Sharma, Vishal Chaudhary, Ankit Fatehpuria Founded: 2018; Based in: Bengaluru, Karnataka; Sector: B2B manufacturing marketplace / contract manufacturing and supply chain (capital goods, energy, infrastructure) In his second year at IIT Madras, Amrit Acharya built a vehicle out of scrap and won. The contest was Bosch's Junkyard Wars. The prize was an internship, but the real reward was a taste of something his family had not planned for him. His parents, an insurance employee and an NGO worker, both Arts graduates, had wanted a doctor or an IAS officer. Engineering, he has said, was a late arrival in his life. His grandmother, who was very good at maths, had been his first maths tutor. He got into IIT Madras in 2006. After graduating he joined ITC, where his first assignment was to build a factory. He has described the odd range of the job: talking to the company's CEO in the morning and to the people on the factory floor in the afternoon. At ITC he also met the colleagues he would later start a company with, among them Srinath Ramakkrushnan, another IIT Madras alumnus, and Rahul Sharma. Zetwerk was born in 2018 around a simple frustration. Large companies that needed machinery, steel structures and other custom parts had to hunt across thousands of small, scattered workshops. Quality was uneven, deadlines slipped. Zetwerk put itself in the middle, matching orders to factories and then policing the work. The bet worked. In 2021 a $150 million round made it a unicorn, with money eventually crossing $900 million from names like Peak XV, Accel, Lightspeed, Greenoaks and D1. The price came later. Revenue grew, but losses grew with it. In FY26 the company reported operating revenue of ₹15,913 crore, up 40 per cent, alongside a net loss of ₹1,606 crore. That loss carried a ₹796 crore exceptional charge and a ₹453 crore impairment in a civil infrastructure business it walked away from. Founders Acharya and Ramakkrushnan put up ₹600 crore of their own, raised through personal debt, to raise their stake. A dispute with a former senior executive and his company, Ayr Energy, over confidential information was settled in September 2026. In October 2026 Zetwerk went public at ₹151 a share in a ₹5,012 crore issue, valuing it at roughly $2.7 billion, well below the near-$5 billion private talk of earlier years. It listed on 23 October at about ₹159, then slid to around ₹138 by the close. For a first-time founder, the quiet lesson is in that first job. Acharya started by learning to speak to both the boardroom and the shop floor. A marketplace like his lives or dies on that fluency. ## Pine Labs (India) URL: https://startup.talk/india/pine-labs/ Founders: Lokvir Kapoor, Rajul Garg, Tarun Upadhyay Founded: 1998; Based in: Noida, Uttar Pradesh (registered office in Gurugram, Haryana); Sector: Fintech: POS terminals, merchant payments, issuer/prepaid, BNPL Before it counted card swipes at shopping counters, Pine Labs counted litres at petrol pumps. The company was incorporated in New Delhi on 18 May 1998 as Pine Labs Private Limited, to automate petroleum retail with smart cards and loyalty programmes. Two people signed it into existence, Rajul Garg and Tarun Upadhyay, with 100 shares each at ₹10 apiece, ₹1,000 each. Lokvir Kapoor is not on that first page. An IIT Kanpur mechanical engineer of the 1987 batch, with an MBA from IIM Bangalore and years at Schlumberger behind him, he came in afterwards, and by 2007 he held shares through the company's employee option plan. The first outside money was small and close to home. In March 2005 a company called Induslogic, Inc took 920,000 shares at one rupee each, about ₹9.2 lakh, and bought more the next year. Induslogic was the name the software firm GlobalLogic carried until 2006, and Garg and Upadhyay were among the founders of GlobalLogic. Within a few years of starting Pine Labs, the two had moved on to it. Around 2004 the founding team thinned out and Kapoor took charge as CEO. The company had to decide what it wanted to be. In 2005 it launched Plutus, a way for shops to accept credit and debit cards, with monthly instalments and loyalty offers built in. The petrol-pump cards had taught the team how to put technology into a shop's daily routine, and that skill travelled. It chose merchants. Pine Labs moved from fuel stations to the counters of retailers, selling point-of-sale machines that took cards. Over the next decade Kapoor steered it from a niche hardware seller toward payments infrastructure that banks and shops both leaned on. That patient bet drew serious money. In 2018 Actis and Altimeter put in $82 million, then Temasek and PayPal followed with $125 million, which made it a unicorn. Mastercard, Lone Pine, Fidelity, BlackRock and Alpha Wave came later. At its peak private valuation, the company was marked at $5 billion. The price of scale was not small. A data breach in 2021 exposed information on up to 500,000 users. A planned US listing was shelved after a confidential filing in 2022. Ahead of its Indian IPO, the valuation expectation was reportedly cut by about 40 per cent to around $2.9 billion. It finally listed in India on 14 November 2025 at an issue price of ₹221. The shares opened at ₹242, touched ₹283 during the day and closed at ₹251.30, a market value of about ₹28,856 crore. Its latest annual numbers show revenue of ₹2,274 crore for FY25 and a net loss of ₹145 crore. The company has since widened into prepaid cards, gift vouchers through its Qwikcilver acquisition, and credit tools for lenders, and in February 2026 announced a tie-up with OpenAI for AI-driven shopping payments. There is a plain lesson for founders in that long road. Pine Labs took 27 years to reach the stock exchange. The business it started with, petrol-pump cards, is not the one it listed. What lasted was the relationship with the people at the counter. ## BharatPe (India) URL: https://startup.talk/india/bharatpe/ Founders: Ashneer Grover, Shashvat Nakrani, Bhavik Koladiya Founded: 2018 (incorporated March 2018); Based in: New Delhi (Delhi NCR); Sector: Fintech: merchant QR/UPI payments, POS, merchant lending Ashneer Grover wanted Salman Khan. The company he worked for had about ₹100 crore in the bank, and the actor's fee was around ₹7.5 crore. He has told the story on stage himself, at a Lovely Professional University talk in April 2022: the haggling, the manager who needed convincing, the number sliding to ₹4.5 crore. In 2019, a small payments company called BharatPe got its brand ambassador. It was classic Grover, loud, confident and willing to spend before anyone could tell him not to. He was born in Delhi in 1982, the son of a chartered accountant and a teacher. He went to IIT Delhi, spent time in Lyon on a French Embassy scholarship, and finished an MBA at IIM Ahmedabad in 2006. Then came Kotak, American Express, a stint as CFO at Grofers, which he left in 2017 saying promised ESOPs never arrived, and a job at PC Jeweller. That last one, he says, planted the idea for BharatPe. The company's origin is less simple than his legend. Business Today, drawing on regulatory data, reports it was incorporated in March 2018 by Bhavik Koladiya and Shashvat Nakrani with a capital of ₹1 lakh. Nakrani, then a young IIT Delhi dropout who had earlier built a haircut-booking site, had called BharatPe his brainchild. Grover joined by taking shares transferred from them and held nearly 32 per cent by July 2018. What worked was an interoperable QR code that charged merchants nothing. Valuation went from ₹1.9 crore at the seed round to roughly ₹20,000 crore in August 2021, with Tiger Global leading a round of about $370 million. Then it came apart. In early 2022 the board began a governance review into alleged financial irregularities, and Grover resigned on 28 February. He said he and his family had been vilified. Years of lawsuits followed, with accusations, an arbitration plea and a 2023 petition against him. On 30 September 2024, both sides settled, the cases were dropped and his shares moved, in part, to a trust. BharatPe carries on under CEO Nalin Negi, with FY24 revenue of ₹1,700 crore and a profit of ₹66 crore. Grover moved on to television and a book called Doglapan. Founders should notice the order of events. The brand ambassador, the confidence and the valuation came first. The governance and the relationships had to catch up afterwards, and did not. ## Postman (India) URL: https://startup.talk/india/postman/ Founders: Abhinav Asthana, Ankit Sobti, Abhijit Kane Founded: 2012 (side project); company formally founded 1 Oct 2014; Based in: San Francisco, California (moved from Bengaluru in 2017); major office in Bengaluru; Sector: Developer tools / API platform (SaaS) In 1997, in the small town of Basti in eastern Uttar Pradesh, a boy in Class 5 sat down at a computer for the first time. Abhinav Asthana has recalled that the machine came into the house because his father, who worked as an engineer, bought one and taught himself to program. The family moved from Basti to Lakhimpur near the Nepal border, following the postings that came with a government career. The computer stayed constant. As a teenager he designed websites for pay, and his first earnings arrived in dollars. He once tried to build a Yahoo clone. It never worked. He went on to BITS Pilani's Goa campus, ran a site called BITS 360 for entrance aspirants, and interned at Yahoo in Bengaluru around 2009. That internship is where the story begins. Working with Yahoo's internal APIs was painful, and he has said the frustration led to Postman. In 2012 he built it as a side project, a free Chrome extension that let developers test APIs. He did no marketing. By the time Postman became a company, about half a million people were using it. Nobody asked him to build it, and nobody paid him for it. The decision to commit was not instant. Asthana left his CTO job at a startup called TeliportMe in 2013, and rather than ask his father for money, he lived on consulting projects. In 2014 he formally founded Postman with Abhijit Kane, whom he knew from TeliportMe, and Ankit Sobti, whom he had met at Yahoo. A $1 million seed from Nexus Venture Partners followed in May 2015. In 2017 the company moved its headquarters to San Francisco. The curve that came after was steep. A 2020 round put Postman past $2 billion. In August 2021 a $225 million Series D valued it at $5.6 billion. By April 2022 it reported around 20 million users, and later versions added AI features, including a rebuilt platform in March 2026 and the acquisition of liblab in November 2025. Postman has not reported a newer valuation since 2021, and revenue figures remain unconfirmed. The company's customers now include names like Meta, PayPal, Microsoft and Salesforce, and Asthana lives in San Francisco, a long way from the Basti house with the first computer. The pattern is worth keeping. Asthana did not start with a company. He started with a tool for his own irritation, gave it away, and let the users prove demand before he asked anyone for money. ## BrowserStack (India) URL: https://startup.talk/india/browserstack/ Founders: Ritesh Arora, Nakul Aggarwal Founded: 2011; Based in: Mumbai, Maharashtra (offices in San Francisco, New York, Dublin); Sector: Enterprise SaaS / software testing (cloud browser and device testing) The pitch to hire people was awkward. You would be joining a company that, in its early years, worked out of a Starbucks. That, by one account of BrowserStack's early days, was the sort of thing that made good candidates pause. Ritesh Arora and Nakul Aggarwal had met at IIT Bombay, studying Computer Science, and the two were by many accounts roommates. Before BrowserStack they had already tried other ventures, and those failures made them wary of outside money. They ran a small consulting business and were fans of Ruby on Rails, writing a popular blog and contributing to open source. The real itch came from something dull. They were building a website for the consulting business. Nakul built it quickly. Testing it was the nightmare: old versions of Internet Explorer, slow connections, device after device. Setting up a lab to check every browser would have taken days. So the two sat in a Mumbai coffee shop and asked if anyone else suffered. They found over a hundred developers complaining about exactly this on Twitter, Stack Overflow and Reddit. They built a tool, opened a free beta in 2011, and within a month about 10,000 people had used it. Developers talked about it publicly, jQuery's founder among them. Nobody was spending on marketing; the company reportedly spent no marketing money in its first eight years. Then came the quieter decision. Rather than hunt for investors, they charged $19 a month on a pay-as-you-go plan. It meant they were profitable almost from the start, and revenue reportedly reached about $20,000 a month within six months. Pricing got messy later, with 15 plans, so they cut it to three. It was not smooth. The browser world moved from Internet Explorer to Chrome. Mobile app testing took close to three years to build, against a few months for the web version. In 2024 a rival, Deque Systems, sued over its accessibility products; a US federal court dismissed the complaint in January 2025. Only in 2018 did they take outside money, a $50M Series A from Accel, followed in 2021 by a $200M round at a $4 billion valuation. By 2025 the company reported about $225 million in revenue. In April 2025 the founders pledged Rs 100 crore to IIT Bombay. For a first-time founder the lesson is unglamorous. They did not invent a grand category. They fixed an irritation they personally felt, checked that strangers felt it too, and charged for it on day one. ## Chargebee (India) URL: https://startup.talk/india/chargebee/ Founders: Krish Subramanian, Rajaraman Santhanam, Saravanan KP, Thiyagarajan T Founded: 2011; Based in: San Francisco, California; major base in Chennai, Tamil Nadu; Sector: SaaS / subscription billing and revenue management The office had a curtain in it. On one side was a living room; on the other, four men trying to build software that would bill businesses every month. That is how Chargebee started in 2011, in a co-founder's drawing room in Chennai, with late-night research and long discussions. The founders came from two worlds. Krish Subramanian and Rajaraman Santhanam had studied together and had both long wanted to run their own company. One account says Subramanian was working at Cognizant; another says he spent years at Zoho. Santhanam, Saravanan KP and Thiyagarajan T had been building SaaS and cloud products at Zoho. Subramanian and Santhanam quit their jobs and brought the other two in. The timing was awkward. Subscription and SaaS models were still a novelty, and subscription services were not popular in India. Selling a billing tool for businesses that barely existed locally meant looking outward, to customers abroad, from a Chennai living room. Money came slowly. An angel round in 2012. Then in 2014 Accel made its first big investment, a modest sum by later standards. In 2015 Tiger Global led a $5 million round. Nothing explosive, just steady compounding. Then the market caught up. In April 2021, a $125 million round led by Sapphire Ventures valued Chargebee at $1.4 billion, a valuation that reportedly tripled in six months. The company said revenue was doubling year on year. Less than a year later, in February 2022, Tiger Global and Sequoia co-led a $250 million round at $3.5 billion. It was a company that stayed small and patient for about a decade before the money arrived. By 2024 one third-party estimate placed annual recurring revenue at about $202.6 million. Santhanam has said the company plans to invest R&D in the Indian market, even though it sells mainly to the world. The lesson for a first-time founder: you can build a global product from a rented living room, if you pick a boring problem, a category that is about to grow, and the patience to outlast the years when nobody has heard of it. ## Innovaccer (India) URL: https://startup.talk/india/innovaccer/ Founders: Abhinav Shashank, Kanav Hasija, Sandeep Gupta Founded: 2014; Based in: San Francisco, California (large operations in Noida, Uttar Pradesh); Sector: Healthcare IT / data and AI platform Long before healthcare, there was a student club idea at IIT Kharagpur. Abhinav Shashank and Kanav Hasija, who reportedly met there as undergraduates, started something called the Kharagpur Consulting Group, an attempt to link the campus with industry. Shashank is described as having grown up in Lucknow, and studied mechanical engineering. Hasija went to the US to study patent law. Shashank worked in automation. The itch to build something together stayed with them. The company's origin is murkier than its press releases.What is better documented is a decision. In early 2016, the team chose to focus on healthcare. That was the turn. American healthcare was drowning in data nobody could use. Patient records sat in electronic health records, insurance claims, pharmacies, labs and remote monitoring devices, each speaking a different language. Innovaccer's pitch was to pull them together. But stitching data together was not enough. The company has described how unified data alone did not change outcomes, so it had to build applications on top, and later AI agents, as hospitals struggled with staff shortages. The money followed. A $105 million round in February 2021 made it a unicorn. Mubadala Capital led a $150 million round in December 2021 at $3.2 billion. In January 2025, a $275 million Series F took total funding to about $675 million. The company says it has assembled records for more than 80 million patients and has been buying its way into adjacent ground: Cured, Pharmacy Quality Solutions, Humbi AI and Story Health. The lesson that survives is about focus. A pivot is easier to survive when you have friends who trust each other since college. Innovaccer found its market not by inventing something new, but by deciding to stop doing everything and pick one painful industry. ## Icertis (India) URL: https://startup.talk/india/icertis/ Founders: Samir Bodas, Monish Darda Founded: 2009; Based in: Bellevue, Washington, USA (major engineering base in Pune, Maharashtra); Sector: Enterprise SaaS / contract lifecycle management (AI) In January 2026, the company announced that its co-founder and CEO, Samir Bodas, had died after a battle with cancer. He was 61. Bodas had arrived in the United States in 1982 to study at the University of Texas at Austin, and later earned an MBA at Wharton. In between were several ventures: iMandi, Jamcracker, Disha Technology, and Aztecsoft, where he was CEO. These were the names on a long entrepreneurial CV before Icertis. His partner, Monish Darda, came from BladeLogic. In 2009 they founded Icertis, with its engineering base in Pune and headquarters in Bellevue, Washington. The product was contract management software, not an obviously exciting category. They built it on Microsoft Azure and sold it as a service to large enterprises, with AI to flag problem clauses and compliance risks. The growth was patient. Early money was modest: $6 million from Greycroft and Fidelity in 2015 and $15 million in 2016. In 2019 a $115 million round valued it around $1 billion. In 2021 an $80 million Series F put the valuation near $2.8 billion, and SoftBank was reported to be close to a stake at $5 billion. SAP took a minority stake in early 2022. By the end of 2021, the company had around 1,700 employees. The takeaway for a founder with a long career behind them: experience is not the same as retiring. Bodas had built and sold companies for twenty-five years, and the biggest one came at the point many people stop. It also shows that a boring category with big enterprises as customers can become a multi-billion-dollar company. ## Darwinbox (India) URL: https://startup.talk/india/darwinbox/ Founders: Jayant Paleti, Rohit Chennamaneni, Chaitanya Peddi, Vineet Singh Founded: 2015; Based in: Hyderabad, Telangana (group headquarters listed as Singapore); Sector: HR tech / cloud HCM SaaS What exists is a set of facts about three people who decided to build HR software together and kept running the company jointly. Jayant Paleti, Rohit Chennamaneni and Chaitanya Peddi launched Darwinbox in 2015. The Indian entity is registered in Hyderabad, with a Singapore holding company above it.The founders run the company collectively. In August 2024 their CTO Vineet Singh was elevated to co-founder. The product is a cloud suite for human resources, payroll and talent. It was competing against global suites such as Workday and UKG, plus Indian rivals like Zoho People and Keka. In 2021 it appeared in Gartner's Magic Quadrant for Cloud HCM Suites as the youngest Asian player. The money arrived in stages. About $15 million from Salesforce Ventures in 2021, then a $72 million Series D led by TCV in January 2022 that made it a unicorn. Microsoft invested in January 2023. In March 2025, KKR and Partners Group put in $140 million, mostly buying existing shares, at a reported valuation around $950 million. The revenue picture is clearer than the human one. Revenue rose to about Rs 534 crore in FY25. International markets made up 63% of new sales. The company is not yet profitable, with one database showing a net loss of Rs 276 crore, and it has talked about an IPO in two to four years. It has also been buying back employee stock, about Rs 86 crore in June 2025. What can a founder take from this? That a category crowded by global giants can still be entered from Hyderabad by selling in markets where the giants are expensive or slow, and that founders who share control for a decade have to trust each other. ## Druva (India) URL: https://startup.talk/india/druva/ Founders: Jaspreet Singh, Milind Borate, Ramani Kothandaraman Founded: 2008; Based in: Santa Clara, California, USA (founded in Pune, Maharashtra); Sector: Cloud data protection / SaaS backup Jaspreet Singh has told the story of one particular door that wouldn't open. Druva was a tiny team in Pune, and one of the customers it wanted most was NASA. At first, Singh said, he couldn't even get into NASA's offices. Then, one day, an email arrived. NASA had tried the product, liked it, and wanted to meet. That email was the payoff for a bruising first two years. Singh, Milind Borate and Ramani Kothandaraman had worked together at Veritas Software, the backup giant. In 2008 they left to start their own company in Pune and named it after the Sanskrit word for the Pole Star. Singh, who reportedly studied computer science at IIT Guwahati, has said little in public about his childhood, so the early picture is mostly about the work. The work did not go well at first. Their first product, inSync 1.0, was ambitious, and the market was not interested. No one, Singh later said, wanted to buy high-end software from an unfunded company in Pune. The team scaled back, took the product apart and rebuilt it for enterprise customers around the world, starting with the unglamorous problem of backing up laptops. Then came money and a bigger bet. Sequoia backed the company in its Series A in 2010. In 2012 Druva moved its headquarters to Silicon Valley, because Singh wanted to sit where the buyers were and watch the cloud take over. That meant another rebuild: a product that lived in the cloud rather than in a customer's server room. He later described Druva as a story of pivots, not one clean idea. The turn came slowly, then all at once. The cloud model fit what enterprises wanted, and the company added server backup, then more acquisitions to widen its reach. In 2019 it raised $130 million led by Viking Global at a valuation above a billion dollars. By then it had passed $100 million in annual recurring revenue. In April 2021 came another $147 million at about $2 billion. The price of that path was a long stretch of looking small. 'Everything looks like an overnight success in Silicon Valley,' Singh told Inc42, and then said that was not Druva's story. It took more than a decade from the Pune office to the unicorn headline. Druva remains private, and its recent financials are not public. For a first-time founder, the useful part is the order of operations. Singh did not start with a grand vision. He started by listening to why a customer would say no, and he was willing to rebuild the product, and even move countries, to hear a yes. ## Rapido (India) URL: https://startup.talk/india/rapido/ Founders: Aravind Sanka, Pavan Guntupalli, SR Rishikesh Founded: 2015 (originally theKarrier); Based in: Bengaluru, Karnataka; Sector: Mobility / bike taxi, auto, cab, logistics The first customers Rapido had to win over were not riders but worried mothers and wary city officials. Aravind Sanka has said the biggest early barrier was a simple belief: that getting on the back of a motorcycle was unsafe. So the team made safety part of the product. Insurance from day one. Speed tracking that deactivates a captain who goes over a set limit. That obsession with an objection came from a company that began somewhere else entirely. Sanka studied mechanical engineering at IIT Bhubaneswar, interned at Tata Motors, then worked as a finance business partner at Flipkart. With Pavan Guntupalli, who studied at IIT Kharagpur, and SR Rishikesh from PES University, he started theKarrier in 2015, a business that moved goods for clients.The logistics business hit a ceiling. Sanka has said the team saw more room to grow by serving consumers, not businesses. Trucks were wrong for people. Cars meant higher costs, more competition and traffic. Bikes were everywhere, so the founders asked owners to register as captains, rather than buying fleets the way Ola and Uber had. The company started with 15 to 20 people. Early backers included Hero MotoCorp's Pawan Munjal and former Google India head Rajan Anandan in 2016. Growth was fast and messy. By 2018 it counted around 15,000 registered riders and about 30,000 rides a day. By November 2019 it said it had 1 crore registered users. But much of the country's law had not caught up. Bikes were seized in Coimbatore in 2018. The Madras High Court banned operations in Tamil Nadu in July 2019, then lifted it a month later. Assam banned it in 2020. In Bengaluru, its home, more than 120 bikes were impounded in January 2022 alone. Pune followed. Investors kept betting anyway. Swiggy led a $180 million round in April 2022. WestBridge led $200 million in September 2024, making Rapido a unicorn. This May, Prosus led a $240 million round at a valuation of about $3 billion, nearly triple the 2024 figure. FY25 revenue reached ₹934 crore, up 44%, with a net loss of ₹258 crore. The company says it plans to begin its IPO process by the end of 2026. The lesson is in how Rapido found its idea. It did not start with a big vision for mobility. It started with a business that worked, noticed where it stopped growing, and let the customer's biggest fear, safety, decide what to build first. ## BlackBuck (India) URL: https://startup.talk/india/blackbuck/ Founders: Rajesh Kumar Naidu Yabaji, Chanakya Hridaya, Ramasubramaniam Balasubramaniam Founded: 2015; Based in: Bengaluru, Karnataka (Sept 2025 report that it plans to move out of Bengaluru); Sector: Trucking / logistics marketplace and fintech At 30, Rajesh Yabaji sat for an interview and used a word most founders avoid. His business, he said, was dhool-mitthi: dust and mud. Trucks, drivers, yards. Not the glossy end of technology. But Harvard Business School had just written a case study about it, and he called that 'extremely humbling'. He came to trucking sideways. Yabaji graduated from IIT Kharagpur in 2010 with a degree in metallurgical and materials engineering. During a final-year internship, he has said, he fell in love with supply chain management and saw how inefficient India's freight industry was. He joined ITC and ran supply chain for its leaf tobacco business, a job that taught him how goods move at scale and how much of that movement is waste. In April 2015 he resigned. He walked away from a secure career at a major company to start BlackBuck with Chanakya Hridaya and Ramasubramaniam Balasubramaniam. The market they chose was brutal. Truckers ran on phone calls and personal relationships. Only about 7 per cent of truck drivers then had smartphones.The early signs, at least as one profile tells it, were good. A $5 million seed round in April 2015. Flipkart in with a $25 million Series B the same year. By the company's account, it turned EBITDA-positive within about four months. Accel co-led an $80 million Series D in 2019, and a $67 million round in July 2021 made BlackBuck a unicorn. COVID hit the people who made the platform work. Many truckers went out of business and small shippers lost access to trucks. Yabaji shut down projects that weren't paying off and waived platform fees for small and medium shippers. Volumes later grew as those shippers came back. November 2024 brought the public market: a ₹1,114.72 crore IPO priced at ₹273, listing on 22 November at about ₹279. Accel held 17 per cent and the three founders each held roughly 10.7 per cent going into it. The first quarters after were uneven: a ₹32 crore profit in the June quarter of FY25, then a ₹308 crore loss in the next, tied to a one-time item. Excluding it, the company would have posted a profit of about ₹39 crore. For a founder, the useful detail is the word dhool-mitthi. Yabaji picked an unfashionable industry because he had seen its waste up close, and then spent a decade earning the trust of people who did not want an app. ## CarDekho (India) URL: https://startup.talk/india/cardekho/ Founders: Amit Jain, Anurag Jain, Umang Kumar Founded: 2008 (CarDekho.com launched March 2008; parent GirnarSoft set up earlier as an IT consultancy); Based in: Jaipur, Rajasthan (also Gurugram); Sector: Automotive marketplace / classifieds, auto fintech and insurance The idea started at a car show. In January 2008, Amit and Anurag Jain went to the Delhi Auto Expo and left with a problem: buying a car in India meant too much guesswork. The brothers wanted to make it easier. By March that year CarDekho.com was live, launched, as one profile tells it, from a garage. They were not outsiders to technology. Both studied at IIT Delhi, Amit in civil engineering and Anurag in mathematics and computing. They had gone to St. Xavier's School in Jaipur. Amit had worked at TCS and Trilogy as a product manager, and Anurag at i2 Technologies and Sabre. Together they ran GirnarSoft, an IT consultancy, which meant they had a way to pay the bills while an idea took shape. They did not quit it. This is the unglamorous part of the story. The consultancy kept running, giving steady income while CarDekho grew. The first push for clients was by phone and email, and it was slow. The first client came through, and brought referrals. A third co-founder, Umang Kumar of Gaadi.com, later joined when CarDekho bought his company in 2014. The site gained audience. By January 2012 it counted 2.5 crore visitors. It moved from listings into new cars, dealerships and eventually finance and insurance, and bought companies along the way: Carmudi Philippines in 2019, Revv in 2023. A $70 million round from Ping An's fund in December 2019 and a $200 million round led by LeapFrog in October 2021 funded the push. Then reality set in. In September 2021 it shut more than 20 used-car stores to move to home inspections. Losses ran for years, from ₹343 crore in FY21 to ₹640 crore in FY24 on the figures of one profile, even as revenue climbed. The picture has improved: FY25 consolidated revenue was ₹2,795 crore, up 24 per cent, with the group's loss narrowing to about ₹266 crore and the standalone entity reporting a profit. The IPO has been in the pipeline for some time, with reports of a ₹3,500 crore or larger offering, but a listing date has not been confirmed. The lesson here is quiet. The Jains did not bet everything on the new idea on day one. They kept the old business paying the bills until the new one could stand on its own, and in doing so bought themselves the years it took for CarDekho to become real. ## Cars24 (India) URL: https://startup.talk/india/cars24/ Founders: Vikram Chopra, Mehul Agrawal, Gajendra Jangid, Ruchit Agarwal Founded: 2015; Based in: Gurugram, Haryana; Sector: Used-car marketplace, auto lending In April 2026, two of the four people who started Cars24 left their operating jobs in the same week. Mehul Agrawal gave up the COO role and Gajendra Jangid left as chief marketing officer. The company was preparing for an IPO. It was an odd moment for a goodbye, and a fair place to start the story. The four founders came from different worlds. Vikram Chopra studied at IIT Bombay and then at the University of Pennsylvania, and worked at McKinsey and Sequoia. Agrawal had been at BCG and had built the online furniture company FabFurnish. Chopra was an early part of it. Jangid, another IIT Bombay graduate, came from Schlumberger. Ruchit Agarwal, a Cornell graduate, had been a vice president at Bank of America Merrill Lynch.In 2015 they set up Cars24. They looked at the used-car market and saw what they called 'utter chaos': no standard way to buy or sell and no technology for pricing. The answer was to inspect cars, quote a price and close the deal fast. By 2019 it had 202 branches in 73 cities and sold about 150,000 cars a year. A buyback guarantee added trust. Then the money arrived. DST put $200 million in November 2020 at a $1 billion valuation. Through 2021 came further rounds from SoftBank, Tencent and Alpha Wave, and a $400 million round that December at $3.3 billion. Cars24 went into the UAE, Thailand and Australia. The bill came due fast. In December 2021 it shut supply-focused businesses in about 82 Indian cities. In May 2022 it laid off about 600 people. Senior executives left. India revenue then fell for two straight years after peaking in FY24, to about ₹6,233 crore in FY25, with a net loss of ₹543 crore. The recovery story is told by the company. For FY26 it reports adjusted net revenue of ₹1,411 crore, a smaller loss of ₹363 crore, and ₹20 crore of adjusted EBITDA in the last quarter, which it calls its first profitable one. Those numbers are unaudited. Chopra said in January the company hoped to list within six to twelve months. The lesson is uncomfortable for anyone chasing growth. Easy capital made Cars24 big in months, and shrinking back to something that earns money took years and cost people their jobs. ## Spinny (India) URL: https://startup.talk/india/spinny/ Founders: Niraj Singh, Ramanshu Mahaur, Mohit Gupta, Ganesh Pawar Founded: 2015; Based in: Gurugram, Haryana; Sector: Used cars / auto commerce It began over coffee. Niraj Singh and Ramanshu Mahaur were talking about something that had nothing to do with startups: how many of their friends had bought a used car, and how miserable it had been. The stories were all the same. Friends trudged from dealer to dealer, haggled over prices that never felt reasonable, and drove home with a car they could not fully trust. They did not have a bad purchase, Niraj later wrote, so much as 'a sense of doubt about the quality of their purchase.' That doubt stuck with him. He and Ramanshu took the idea to Mohit Gupta and Ganesh Pawar, and all four, in his telling, became convinced it was their next venture. Reports describe Niraj as an IIT Delhi alumnus who put in about $500,000 of his own money to begin. The first version, in 2015, was an obvious one: a marketplace where owners could list cars for other individuals to buy. It did not work. Used cars are a business of mistrust, and a website that merely connects strangers does nothing about that. 'The presence of information asymmetry throughout the market quickly became apparent to us,' Niraj wrote. Buyers could not see what the seller knew. So in 2017 the founders did the harder thing. Instead of introducing buyers to sellers, Spinny would stand in the middle. It would buy cars, put each through a 200-point inspection, price it with its own engine, offer a money-back guarantee, and handle the loan, the title transfer and the delivery. Ten months into the first model they had listed more than 3,000 cars and inspected over 100 a day. The retail model was far heavier on capital, and it had to be: every car Spinny owned was cash sitting in a lot. That is where the price shows. In FY23 the company lost ₹820 Cr. Marketing spend that year was ₹424 Cr; by FY25 it had been cut to ₹125 Cr, and employee costs fell from ₹392 Cr to ₹338 Cr. Procurement of cars is still about 80% of everything it spends. Sourcing carries its own risks, from dodgy paperwork to hidden accident damage, and refurbishment and re-registration add delay and cost. The discipline has started to show. In FY25, Spinny's revenue rose 25% to ₹4,657 Cr and its loss narrowed 28% to ₹424 Cr. In June 2025 it raised $170 Mn led by Accel's growth fund at a valuation of roughly $1.7-1.8 Bn, and it is reported to have confidentially filed for an IPO. The lesson is in the sequence. The founders started with a problem a friend could describe in one sentence, tried the cheap solution, watched it fail for a specific reason, and then paid for the expensive one. Trust was the product all along, and it cost money to build. ## Licious (India) URL: https://startup.talk/india/licious/ Founders: Abhay Hanjura, Vivek Gupta Founded: 2015; Based in: Bengaluru, Karnataka; Sector: Direct-to-consumer meat and seafood / foodtech Vivek Gupta had said no. He was a venture investor at Helion Ventures, he had just put money into BigBasket, and the pitch from Abhay Hanjura, meat delivered fresh to your door, did not move him. So Abhay invited him to lunch. He ordered a chicken dish from his own fledgling company. It arrived in a soan papdi box, dripping and smelling bad, and the meat spilled onto Vivek's clothes. 'What the hell!' Vivek exclaimed. Abhay later recalled that moment as his proof. The problem of buying meat in India was not an opinion. It had just landed on a good shirt. They were an odd pair: one who understood food, one who understood money. Together they started Licious in 2015 in Bengaluru's Whitefield with a team of five, sourcing chicken from farms within about 200km and processing it themselves. The early months were thin. At one point the company had about 15 days of cash left, a key investor backed out, and the founders put in their own savings to pay salaries. Vivek, an investor turned operator, is said to have delivered orders himself at weekends. Suppliers did not take kindly to the founders' quality demands. The business ran on the one thing the lunch had proved: people hated what they were getting. What changed things was control. Licious did not just sell meat; it took responsibility for the entire path from farm to kitchen, including the processing and the cold chain that kept it fresh. By October 2015 it had reached about 1,300 orders a month, and by FY26 net revenue stood at ₹1,166 Cr, up 47%. The company also grew beyond its own app into offline stores, from ₹26 Cr in FY25 to ₹177 Cr in FY26. But the cost of control is cash. Licious still burns it: EBITDA losses were ₹187 Cr in FY26, and the founders have talked of being IPO-ready while pushing a listing, once pitched at a valuation above $2 Bn, to 2027-28. The founders' own origin story carries a quiet lesson. The investor's no was not a verdict on the idea; it was a gap in what he had seen. Abhay did not argue the market size. He let the product fail in front of the person he needed to convince, and let the failure explain the business. ## Country Delight (India) URL: https://startup.talk/india/country-delight/ Founders: Chakradhar Gade, Nitin Kaushal Founded: 2015; Based in: Gurugram, Haryana; Sector: Dairy and fresh food subscription delivery Two men who spent their days reading other people's balance sheets decided, in the end, to sell milk. Chakradhar Gade and Nitin Kaushal had been colleagues at IBF Financial in 2011. Their work was to study businesses and judge them, which meant they knew exactly what a good one looked like. They wanted something people needed every day, and they settled on the fresh, creamy milk they remembered from growing up. Their first idea about cows was wrong. They treated cattle as non-depreciating assets, a view the founders later described as naive. Dairy turned out to have plenty of ways to lose money. Venture capital did not help. Fresh food delivery, and milk in particular, was not a category investors wanted to touch, and the founders had never raised money before. So they used what they had: their own savings. For years, until about 2017, they ran the company on thin budgets with no frills. They rented two-wheelers, worked with a handful of delivery partners and personally handled procurement, route planning and customer calls. The company's real trick was proof. Early customers got third-party lab reports with their milk. The founders built their own cold chain and made each bottle traceable back to its farm. The goal in the first weeks was modest: get a family to try it once and let quality bring them back. The lab reports later grew into a home milk-testing kit developed with DRDO. Marketing came mostly by word of mouth. Only then did money arrive, and in steps: $1.5M from Orios in 2018, $10M in 2019, $25M from Elevation Capital in 2020, and $108M in 2021 led by Venturi Partners and Temasek. Revenue reached ₹1,380 Cr in FY24, up from ₹917 Cr the year before. In March 2025, a Temasek-led round valued it at about $812-820 Mn. Trust is harder to keep than to win. In April 2025 testing platform Trustified reported that the company's Desi Danedar Ghee contained Novaluron above the regulator's limit, a blow to a brand that had sold itself on purity. The founders' path suggests something a first-time founder can use. They chose a boring category on purpose, found that no investor wanted to back it, and took that as a reason to learn the business with their own money first. By the time investors came, the company had a product people paid for every morning. ## Cult.fit (India) URL: https://startup.talk/india/cult-fit/ Founders: Mukesh Bansal, Ankit Nagori, Rishabh Telang and Deepak Poduva Founded: 2016; Based in: Bengaluru, Karnataka; Sector: Fitness and wellness Mukesh Bansal stood alone in a mall in Pune on a weekday, looking at shelves of clothes and shoes. He had been coming to malls every Wednesday and Thursday, scouting for a way to sell personalised products offline. That day he saw something different. Most of the stores around him were selling clothes, footwear and accessories, and the stock sat there while the buyers were at home or at work. In a Business Today interview he described it as a 'Eureka' moment. These things should be online. That was the start of Myntra, which he co-founded in 2007 and which Flipkart bought in 2014 for roughly $330 million. He was already a veteran of failure. Born in Haridwar, a graduate of IIT Kanpur in computer science, he had worked at several early-stage startups between 1997 and 2007, and said he 'saw failure upfront multiple times.' The lesson he took was that a new venture would most likely not work, and that you should start knowing that. In 2016 he and Ankit Nagori, both then at Flipkart, left to start CureFit, a bet that Indians would pay for fitness, food and mental wellbeing in one place. It was a sprawling idea. Cult became the gym business, Eat.fit the food arm, Mind.fit the yoga and meditation product. In 2017 Hrithik Roshan signed on as ambassador. In 2018 the company bought Fitness First India. In 2020 Eat.fit was separated into Curefoods, and Nagori went with it. Then the pandemic closed the gyms. Workforce reductions followed, some centres shut, and the company pivoted to digital classes. An Entrepreneur India report said sales fell 67.4% in the year to March 2021. In November 2021, Zomato led a round with Tata Digital participating at about $1.5 Bn, and in 2022 Cult.fit bought the franchise behind Gold's Gym in India. By 2024 Bansal had stepped aside from the day-to-day, with Naresh Krishnaswamy as CEO and Bansal as chairman. The numbers show the company has narrowed its focus. In FY26 revenue was ₹1,720.6 Cr, up 41.5%, and net loss fell 47.6% to ₹251.8 Cr. It runs 708 centres in 77 cities, and in July 2026 it filed to list, with Bansal and Hrithik Roshan among those selling some shares. What stands out is how Bansal frames the odds. A man who built one of India's biggest e-commerce exits still says most startups don't work. For a founder, that is permission to try plus a reminder to keep the company small enough to survive the bad year. ## PharmEasy (India) URL: https://startup.talk/india/pharmeasy/ Founders: Dharmil Sheth, Dhaval Shah, Siddharth Shah, Hardik Dedhia, Harsh Parekh Founded: 2015; Based in: Mumbai, Maharashtra; Sector: Online pharmacy / healthtech For a few days, some of the PharmEasy founders went into hiding. A competitor had filed a false police complaint. Arrest warrants followed. Their parents received threats, and some of the company's stores were vandalised. It is a strange scene for a business now known as a big healthtech name, but it comes from the founders' own telling in an Entrepreneur India profile. The five of them, Dharmil Sheth, Dhaval Shah, Siddharth Shah, Hardik Dedhia and Harsh Parekh, grew up in Ghatkopar in Mumbai and have been friends since kindergarten. Dhaval trained as a doctor and then went to McKinsey. Siddharth developed the first online pharmacy idea during his MBA at IIM Ahmedabad and incorporated it as Dialhealth.com. Hardik left Carnegie Mellon to join him. Harsh quit Airtel. Dharmil was running another business, 91streets, when the group decided to pivot. It did not work at first. Dialhealth stalled. A retail pharmacy chain that followed also failed. What they got from the failures was a lesson in the supply chain, and a wholesale distribution business that taught them how medicines move. Money was the harder part. When no one outside would back them, their families stepped in, mortgaging homes and assets for about ₹21 Cr in total. Siddharth later said they 'didn't have the luxury of making mistakes.' A partner issued a term sheet at a ₹50 Cr valuation and withdrew after due diligence. Within six months a new investor came in at ₹150 Cr. Then the rise: the unicorn status in 2021, the Medlife merger, and the purchase of 66.1% of Thyrocare for ₹4,546 Cr. At the peak, the company was valued at about $5.6 Bn. An IPO for ₹6,250 Cr was filed in November 2021 and withdrawn in August 2022 after markets turned. Founders and employees bought ₹45 Cr of shares from investors at market price. The fall was steep. A $300M Goldman Sachs loan at 17-18% required fresh equity of ₹1,000 Cr by August 2023, which it missed. Ranjan Pai's family put in ₹1,300 Cr in a rights issue and ₹800 Cr in 2024. Valuation fell about 90% to roughly $456 Mn, and the founders stepped back, with Siddharth leaving the CEO's chair in August 2025. Today the company is still losing money, ₹1,517 Cr in FY25, though 40% less than the year before, on revenue near ₹5,900 Cr, mostly from selling medicines to retailers. Dharmil's reflection to Entrepreneur India is the one worth keeping: 'I would have wanted to fail faster.' The expensive thing in a startup is rarely the failure itself. It is the time spent before you admit it. ## Practo (India) URL: https://startup.talk/india/practo/ Founders: Shashank ND, Abhinav Lal Founded: 2008; Based in: Bengaluru, Karnataka; Sector: Healthtech (doctor discovery, appointments, clinic software) In 2008, Shashank ND was trying to do something that sounds simple. His father needed knee surgery, and the family wanted a second opinion from a physician in the United States. That meant sending over medical records. In India, those records lived on paper, in folders, in the hands of whichever hospital had last seen him. It was a small, maddening errand, and it stuck. Shashank was still at NIT Karnataka, Surathkal, studying information technology alongside Abhinav Lal, a batchmate who was the better coder of the two. The two of them looked at the problem and concluded that the whole system lacked a digital spine. They skipped the usual route of campus placements to build one, working at first from Shashank's father's modest three-bedroom house in Bangalore. The first idea, a way for patients to search for doctors, ran straight into a wall. Doctors had no online schedules, so there was nothing to search. In 2009 the founders pivoted to Practo Ray, software to run a clinic: appointments, billing, records. Then came the hard part. Doctors had low digital literacy, were wary of software that made their billing and taxes visible, and did not adopt anything on their own. The pair built a field sales force and sold door to door, a feet-on-street operation that was slow and costly. That grind turned out to be the moat. By the time Practo.com launched for patients in 2012, thousands of clinics were already on the system, and Sequoia's India arm backed the company around then. Tencent led later rounds of $90M in 2015 and $55M in 2017, and the company grew into a platform for booking, consultations, pharmacy and diagnostics. But growth got loose. Practo wandered into fitness subscriptions and hardware, burning capital before pulling back to its core. Pharmacy discount wars squeezed margins, and the company chose not to match them. COVID brought a surge in teleconsultation and a clearer legal footing with India's 2020 telemedicine guidelines, yet losses stayed heavy: an operating EBITDA loss of ₹162 Cr in FY22. The discipline came late but it came. By FY25 Practo reported an operating EBITDA profit of ₹15 Cr on ₹234 Cr revenue, with GMV near ₹3,500 Cr. In 2025 former COO Jagnoor Singh took over as CEO while Shashank became executive chairman, and Abhinav Lal moved to chief scientist, as the company prepared for an IPO that has been promised since 2023 but not yet dated. The lesson sits in the first year. Practo did not start with a grand marketplace. It started by fixing the unglamorous supply side that the marketplace depended on, and it paid for that with years of knocking on clinic doors. If the thing your idea needs does not exist yet, you may have to build it yourself first. ## Pristyn Care (India) URL: https://startup.talk/india/pristyn-care/ Founders: Harsimarbir Singh, Dr Garima Sawhney, Dr Vaibhav Kapoor Founded: 2018 (August); Based in: Gurugram, Haryana; Sector: Healthtech (elective surgery care delivery, clinics, hospital network) Harsimarbir Singh's mother needed surgery, and Dr Garima Sawhney has said that this brush with the system was part of what pushed the founders to act. The family found what millions of Indian families find: a fragmented maze of small, owner-run clinics, uncertain pricing and not much comfort. By then Singh was no stranger to failure. He had started Hyden Power in 2010, and it failed. A venture called US Brands failed too. Whoopie Technologies and DeliverOye followed. He was a serial founder with more lessons than wins, and his childhood friends, two doctors, kept telling him the same thing about healthcare: patients got a bad deal and doctors knew it. So in August 2018 in Gurugram, Singh, Sawhney and Dr Vaibhav Kapoor started Pristyn Care. The pitch was narrow. They would focus on elective surgery, planned procedures in areas such as proctology, gynaecology and urology. Pristyn would own the patient's whole path: its own clinics for the first consult, its own doctors for the operation, partner hospitals for the theatre, and coordinators for insurance and follow-up. Singh has said most Indians land at small private facilities with a frustrating experience, and the company was built, in his words, to fix that. Investors saw the logic quickly. Sequoia Capital India wrote an early cheque in 2019, and rounds followed from Hummingbird, Greenoaks and then Tiger Global. By December 2021 a $96M Series E made Pristyn a unicorn at $1.4B. By then it ran more than a hundred clinics and hundreds of partner hospitals. In 2022 it bought Lybrate, bringing about 150 people with it. The price of speed showed up in public. In September 2022 Singh posted his interview tactics on LinkedIn, including making candidates wait six to eight hours and calling them in on Sundays. He deleted the post, but the screenshots had already travelled. In 2024 the company laid off around 120 people in a second round. Earlier, Singh had told Business Standard in 2023 that revenue would reach ₹2,000 Cr the following year. The actual figure for FY24 was ₹600.5 Cr, with a net loss of about ₹381 Cr. Still, the business is real: over 700 partner hospitals and 40-plus cities, and FY24 revenue up 32.6% on the year before. The next target is profitability. For a first-time founder, the lesson is in Singh's own record. His earlier failures taught him where not to look. His best idea came from a problem he and his friends had lived inside, and a hard promise about growth cost him credibility when it did not arrive on schedule. ## Dailyhunt (India) URL: https://startup.talk/india/dailyhunt/ Founders: Virendra Gupta, Umang Bedi Founded: 2007 (Verse); Dailyhunt/Newshunt acquired 2011; Based in: Bengaluru, Karnataka; Sector: Local-language content, news aggregation and short video (Dailyhunt, Josh) In 2011, Virendra Gupta had a plain problem. People wanted to read the news in Hindi on their phones, and the phones could not draw Hindi letters. "Reading messages in Hindi was tough in 2011," he later told Entrepreneur India. Gupta's route to that moment was not glamorous. He had a degree in engineering from MBM Engineering College in Jodhpur. In 2007 he started Verse, a mobile value-added services business that pushed SMS alerts on jobs, matrimony, property and education to subscribers across the country. It was a business of tiny messages and tiny margins. The app he bought changed that. Newshunt began in 2009 as a Symbian app written by two former Nokia engineers, Umesh Kulkarni and Chandrashekhar Sohoni. At the end of 2011 Gupta acquired it. He did not just buy and wait. He built a rendering engine so regional-language text would actually display, and tuned the app to run on low-bandwidth 2G phones. Dailyhunt's first technology innovation, he said, was this rendering engine. It was boring plumbing that mattered to hundreds of millions of people who did not read English. The app was renamed Dailyhunt in 2015. Money followed: Sequoia India in 2014, then $40M from Falcon Capital in 2015 and $25M from ByteDance in 2016. At one point the app had around 17-18 million daily users, who spent about 25 minutes a day on it, and was growing 80-100% a year. In 2018 Umang Bedi joined as president and became co-founder in 2020. The company, now called VerSe Innovation, added Josh, a short-video app, and in April 2022 reported an ~$805M raise at a valuation near $5B. The next four years were harder. Competition from global platforms squeezed the short-video business. In 2024 one investor, 360 ONE, valued it around $2.9B, which VerSe disputed. Then, in FY26, a ban on real-money gaming hit advertising, livestream income fell and operating revenue dropped 40% to ₹1,162 Cr from ₹1,930 Cr the year before. The net loss narrowed to ₹699 Cr, but in October 2026 Baillie Gifford marked its holding to an implied valuation of about $576M, roughly 90% below the 2022 peak. There is a founder lesson in both halves of the story. Gupta won early by solving a very specific, unglamorous problem for people nobody else was serving. The later slump is a reminder that a business dependent on a few advertising categories can lose a big part of its revenue to a single regulation. ## Pocket FM (India) URL: https://startup.talk/india/pocket-fm/ Founders: Rohan Nayak, Prateek Dixit, Nishanth KS Founded: 2018; Based in: Bengaluru, Karnataka (founded in Gurugram); Sector: Audio and storytelling entertainment (audio series, novels) The idea for Pocket FM arrived on a crowded Gurugram train. Rohan Nayak's job meant three to four hours of daily travel between Gurugram, Delhi and Noida, and he looked around at people staring into their phones with nothing to listen to but music. There was Netflix for video and Saavn and Gaana for songs. For stories, nothing. Nayak had come to products through an unusual route: a couple of years in investment banking, then a role as product lead at a mobile tech start-up, Cube26. He is reported to be an IIT Kanpur graduate. In 2018 he met Nishanth KS and, with Prateek Dixit, started Pocket FM. A seed round of ₹4.61 Cr from Lightspeed and the Times Group followed that December. The early months were rough. The team first thought of a podcast platform, then realised podcasts lacked the daily pull, so it switched to episodic audio fiction, closer to a TV show than a talk. But nobody came. The app drew about 10,000 downloads in its first 18 months. "We had a near-death moment at Pocket FM even before properly starting up," Nayak has said, crediting his co-founders and what he called naive optimism for getting through. Nishanth has described the first year and a half as constant experiments and quick pivots. The turn came from outside. In the 2020 lockdown, people were stuck at home with time and phones. Downloads climbed to 1 crore in about three months and Pocket FM became India's top audio entertainment app. The team also dropped the subscription model, which Indian users resist, for a micro-payment one: the first episodes free, then a small charge for each next episode. Then came the quiet part of the story. In 2022 Pocket FM entered the US. By 2025 only around 15% of users were outside India but they produced about 70% of revenue. That is how a company with a ₹1,052 Cr FY24 became one that says it hit ₹1,768 Cr in FY25, a figure the company has not audited publicly. In March 2026 Nayak said annualised revenue passed $400M. The cost of the growth has been visible. FY24 still showed a net loss of ₹165 Cr, nearly nine of every ten rupees of revenue came from micro-payments and subscriptions, and Pocket TV, its video experiment, was shut even as micro-dramas boomed in India. Rivals like Kuku FM and Pratilipi keep the pressure on. The lesson for a founder is how little the first version mattered. Pocket FM needed two years, around ten pivots by one account, and a pandemic, but it survived because the team kept changing the product while the idea stayed the same. ## Ather Energy (India) URL: https://startup.talk/india/ather-energy/ Founders: Tarun Mehta, Swapnil Jain Founded: 2013; Based in: Bengaluru, Karnataka; Sector: Electric vehicles (smart electric scooters, charging network) Tarun Mehta had received a probation letter at work. Instead of waiting for the next one, he left the office and went to the IIT Madras campus, where his old hostel friend Swapnil Jain was waiting. That was the real start of Ather Energy. Mehta grew up in Ahmedabad in a Marwari family. His father ran a one-man computer business, assembling PCs and selling antivirus and tax software, and Tarun tested the machines by playing games on them: Age of Empires, SimCity, tycoon games. He refused tuitions until Class 10 and took exactly one JEE class, two hours every other day, mainly because the cycle ride was short. He still made IIT Madras, where he picked the new engineering design department. He was nervous about electronics and disliked coding, so he took the mechanical path. In the hostel, his room sat directly under Swapnil's. They spent about five years building things together. Swapnil had made a Stirling engine in his second year, and the two kept at it for roughly three years after college without reaching production. Mehta started looking for jobs, and Swapnil took one at General Motors. The pull of the old project did not fade. Over weekends he took trips to Bangalore to pitch Swapnil on coming back, and on Friday nights he slept in a department room at IIT. They quit and began on batteries and chargers, with a mock scooter. Customers, they discovered, wanted the scooter, not the battery. Mehta has since said the ₹5 lakh incubation grant was all the capital he thought they would need, and that he underestimated the costs. The first real cheque, ₹25 lakh, came from Dr V Srinivasan because Mehta was the only founder who turned up to meet him. Flipkart's Sachin and Binny Bansal put in $1M in December 2014 and Tiger Global followed with $12M. Hero MotoCorp became the anchor investor from 2016. The product took years. The S340 was unveiled in 2016, the 450 series came in 2020, and a plant in Hosur started operating in January 2021. In 2019 Mehta raised the price to about ₹1.65 lakh, which he calls his biggest risk. Sales rose afterward. In May 2025 Ather listed on the NSE and BSE, and Mehta has named the listing ceremony his proudest moment. In Q2 FY26 it beat Ola Electric in quarterly revenue, with about ₹899 Cr and a net loss narrowing to ₹154 Cr, though the stock fell back after lock-ins expired. The lesson is small and practical. Ather did not begin with a grand plan; it began with two friends who kept building, and with one founder who admitted that he tends to jump while his partner remembers everything. ## Rebel Foods (India) URL: https://startup.talk/india/rebel-foods/ Founders: Jaydeep Barman, Kallol Banerjee Founded: 2011; Based in: Mumbai, Maharashtra; Sector: Cloud kitchens / food-tech In July 2025, Jaydeep Barman handed the chief executive's job to a colleague and took a different title: chairman and group CEO. The man who had started Rebel Foods stepped back from running it day to day, and Ankush Grover, a co-founder who had run the India business, became global CEO. Observers read the move as the kind of governance tidying that comes before a stock-market listing. The company they were preparing to list began in 2011 as a single bet. Faasos was a delivery-only food brand, at a time when most Indian restaurants thought of delivery as an afterthought. Barman and Kallol Banerjee were its co-founders, and the idea was simple: if the customer never walks in, the restaurant does not need a dining room. That logic took four years to harden into a business. In 2015 the company opened its first cloud kitchen, a kitchen that only cooks for delivery. In 2016 it went cloud-kitchen-only. Then came the part that made it unusual. Instead of one brand per kitchen, Rebel began running many brands from the same kitchens: biryani, pizza, bowls, desserts. By 2025 the portfolio had grown past 45 brands, among them Behrouz Biryani, Oven Story Pizza and The Good Bowl. Partnerships followed. In December 2020 it signed with Wendy's to run 250 cloud kitchens in India, and in February 2022 it took the master franchise rights for Wendy's offline restaurants too. Investors noticed. A 2021 round valued the company at $1.4 billion, with Coatue, Sequoia, Lightbox and the Qatar Investment Authority among the backers. The price of that growth was paid in losses. Rebel lost INR 656 crore on revenue of INR 1,195 crore in FY23. Fast food delivery is a thin-margin business, and a rival executive told Inc42 in 2025 that a clear timeline to EBITDA breakeven had yet to emerge. Managing dozens of brands is its own burden, with capital and attention spread across all of them. Rivals such as Curefoods and Wow! Momo kept pressing, and the company's February 2025 push into 15-minute delivery, QuickiES, reportedly had not yet found much of a following. Still, the numbers have moved the right way. In FY25 operating revenue rose about 14% to INR 1,617 crore and the net loss narrowed to INR 337 crore, from INR 380 crore the year before. Reports for FY26 put the loss near INR 282 crore on revenue of about INR 1,952 crore, and the company is reportedly eyeing a public listing. The lesson is in the sequence. Rebel did not start as a platform. It started with one brand and one stubborn idea about delivery, and only after the idea proved itself did it turn into a system that other brands could ride on. ## Vedantu (India) URL: https://startup.talk/india/vedantu/ Founders: Vamsi Krishna, Pulkit Jain, Anand Prakash, Saurabh Saxena Founded: 2014; Based in: Bengaluru, Karnataka; Sector: Edtech - live online tutoring In July 2022, Vedantu let go of about 100 people. In December it cut 350 more. For a company that had celebrated a billion-dollar valuation less than a year earlier, it was a hard reversal, and the reason was plain: the money that had funded Indian edtech was drying up. The company's founders, Vamsi Krishna, Pulkit Jain, Anand Prakash and Saurabh Saxena, are all described as IIT alumni. Their idea, launched in 2014, was live online tutoring for students in classes 4 to 12, mostly CBSE and ICSE, with extra tracks for JEE foundation, NEET and NTSE. The name joins two Sanskrit roots: Veda, knowledge, and Tantu, network. The early funding came in steps. A $5 million Series A from Accel and Tiger Global, then $11 million led by Omidyar Network, then $42 million in 2019. The pandemic turned the model from an option into a necessity. In July 2020 Coatue led a $100 million round, and in September 2021 another $100 million, led by ABC World Asia, made Vedantu a unicorn. Then came the spending. The company bought Instasolv, a doubt-solving app, in February 2021 and took a majority stake in Pedagogy that July. It lost INR 696 crore in FY22. There had also been a data breach in 2019 that exposed about 687,000 users. The layoffs of 2022 were the bill. What followed was less dramatic and more instructive. Vedantu cut costs, and the losses fell: about INR 158 crore in FY24, with revenue near INR 185 crore. Krishna told the press in 2025 that the company had turned profitable in the fourth quarter of FY25, with free cash flow of INR 6 crore and collections of INR 284 crore for the year. These are company figures, and the full-year accounts had yet to be filed when reports appeared. A company that raised hundreds of millions and then shrank to find its footing is not the usual unicorn story, and it should not be dressed up as one. But it is a useful one. The founders' next phase was not about growth at any price. It was about how little a business could burn and still serve students. For a first-time founder the point is blunt: funding rounds can be reversed by the market, and cash flow cannot be argued with. ## Eruditus (India) URL: https://startup.talk/india/eruditus/ Founders: Ashwin Damera, Chaitanya Kalipatnapu Founded: 2010 (some directories say 2011); Based in: Mumbai, Maharashtra (global presence); Sector: Edtech - executive education In February 2025, Ashwin Damera told YourStory that Eruditus planned to pass $1 billion in annual revenue within five years. It was a bold number for a company that, only a few years earlier, had been losing money at a startling rate. Damera and Chaitanya Kalipatnapu started Eruditus in 2010 with a narrow focus: executive-level courses from top universities. The company works with those institutions to deliver their programmes, rather than building its own brand of degree. That position, between the elite university and the working professional, became the core of the business. Growth drew heavy capital. By 2025 Eruditus had raised more than $814 million in equity. Its Series F, in October 2024, was led by TPG's Rise Fund and valued the company at about $3.1 billion, slightly lower than some earlier figures. A further $150 million, again TPG-led, kept the valuation at $3.1 billion. The cost of that scale was visible. In FY22 net losses were around $349 million, up about 32% on the prior year. In FY23 revenue jumped 75%, but the company still lost INR 1,049 crore. Then the direction changed. In FY24 revenue grew 12.4% to INR 3,733 crore, and adjusted EBITDA loss narrowed to INR 69 crore from INR 417 crore. The company said it was now pushing for profitable growth and, in September 2025, locked in a refinancing to support that. The larger story here is a pivot from growth at any cost to growth that pays for itself. It is a pivot many Indian startups have had to make since 2022, usually with more pain and less warning. For founders, Eruditus offers a quieter lesson than the usual unicorn tale: choosing a partner-led model, one that rents the best credentials in the world instead of trying to build them, can speed up trust. But it also leaves you competing for the same students against rivals who can copy the same playbook, which is why the margin story matters as much as the revenue story. ## Acko (India) URL: https://startup.talk/india/acko/ Founders: Varun Dua Founded: 2016; Based in: Bengaluru, Karnataka; Sector: Insurtech - digital general insurance In October 2021, Acko raised $255 million in a Series D round and became India's 34th unicorn. The company had been licensed to sell insurance only four years earlier. Varun Dua founded Acko in November 2016 in Bengaluru. The insurer received its IRDAI licence in September 2017 and began operations in 2018. It was built as a direct-to-consumer business from the start, with no agent network, selling motor, health and travel cover through its own channels and through partners such as Amazon India and Ola. The early backing came from Amazon, Accel, SAIF Partners and Catamaran Ventures, roughly $42 million by 2018. The model bet that people who were already buying something else online would buy insurance at the same moment. The company also reached into cars beyond cover. Acko Drive launched in Bengaluru in 2019, an online car purchase portal followed in 2020, and in 2022 it added car financing and maintenance. By 2024 Acko Drive operated in 12 cities. Not everything went smoothly. IRDAI fined the company INR 1 crore for lapses in its outsourcing and commission practices. And the economics stayed hard: FY25 revenue rose 35% to INR 2,837 crore, but the net loss, though down 37%, was still INR 424 crore. A 2026 report said the insurance arm posted a profit of INR 43.6 crore in FY26. Now Acko is heading to the market. It has hired Kotak, ICICI Securities and Morgan Stanley for an IPO of up to $350 million, targeting a valuation of $2 to $2.5 billion, above the $1.4 billion it fetched in 2021. A confidential filing is planned for the second half of 2026 and a listing in the first half of 2027. Peer Digit's volatile trading since its 2024 listing is the benchmark the public market will use. For founders, the pattern is worth noticing. Acko's growth has leaned on distribution partners, which gave it scale quickly, but it also means a large part of its fate sits in other companies' hands. Borrowed customers are cheap to get and costly to keep. ## Go Digit (India) URL: https://startup.talk/india/go-digit/ Founders: Kamesh Goyal Founded: 2016; Based in: Pune, Maharashtra; Sector: General insurance (digital-first) On 23 May 2024, Go Digit's shares opened on the NSE at INR 286, 5.15% above the issue price of INR 272. The listing capped a three-day offering that raised about INR 2,615 crore, part fresh capital, part existing shareholders selling. Founder Kamesh Goyal built Digit as a digital-first general insurer. It is Fairfax-backed, and that backing carried it through to the public market. At the issue price the company was worth roughly INR 24,948 crore. What the debut did not guarantee was a smooth ride. The stock traded around INR 350 in early January 2026, giving a market value of about INR 32,336 crore, and around INR 311 in late May 2026, near INR 28,573 crore. Insurance stocks move on claims, pricing and regulation, and Digit's price has been volatile. The business underneath kept growing. In FY26 gross written premium rose 10.4% to INR 11,294 crore, with a return on equity of 17.7%. Reported profit for the year was about INR 544 crore. The combined ratio, a measure of whether underwriting itself makes money, improved to 105.7%. Anything over 100% means the insurer is paying out more in claims and costs than it takes in premium, and relying on investment income to make up the gap. That last number is the honest centre of the story. Digit has scale and profit, but it has not yet shown that its insurance alone is profitable at the underwriting line. Peer Acko is heading to its own listing and Digit's trading history will be the yardstick. For founders, an IPO is not the finish. It is the moment the business gets graded every day, by people who read the combined ratio. ## Zeta (India) URL: https://startup.talk/india/zeta/ Founders: Bhavin Turakhia, Ramki Gaddipati Founded: 2015 (April); Based in: San Francisco, California (offices in Mumbai, Bengaluru and Dubai); Sector: Fintech / banking technology (cloud-native core banking, card processing, employee benefits) At 18, Bhavin Turakhia started a company with Rs 25,000 and his brother Divyank. That was roughly $675. He was born in Mumbai in 1979 into a middle-class Jain family and went to school at Arya Vidya Mandir in Bandra. College did not go smoothly. He began a science course at D.G. Ruparel College and left it, then finished a commerce degree at Sydenham College. By then the company, Directi, was already running with his brother Divyank. A BBC video about him later carried the title 'I followed my father's advice and became a billionaire.' Directi built the plumbing of the early web: domain names, hosting, reseller tools. Brands like BigRock and Reseller Club grew out of it. In 2014 the brothers sold four of those businesses, BigRock, LogicBoxes, Reseller Club and Webhosting.info, to Endurance International Group for $160 million. Forbes India put Bhavin on its 2016 rich list at number 95, with $1.3 billion. Most people would have stopped there. He did something odder. In April 2015 he and engineer Ramki Gaddipati started Zeta, a company aimed at the dullest, most regulated corner of tech: how banks run cards and accounts. In 2016 Turakhia put about $19 million of his own money into it. Banks are not known for trusting young software firms with their core systems, and Zeta had to earn that trust one contract at a time. The turn came through food vouchers. Sodexo, the meal-benefits giant, invested in July 2019 at a $300 million valuation and folded its Indian benefits unit into Zeta. In 2020 Zeta took its platform to the Philippines and Vietnam, with Sodexo as the first client. Then in 2021 SoftBank Vision Fund 2 led a $250 million round at $1.45 billion, and Zeta became a unicorn. Mastercard added $30 million in March 2022. It did not stand still. In 2024 Zeta launched a credit product on NPCI's Credit Line on UPI and said it processes around 2 million transactions a day. In February 2025 a strategic investor, reported to be Optum, put in $50 million at a $2 billion valuation. For FY24 the company reported revenue of Rs 893 crore and a net profit of Rs 119 crore. Asked about a listing, Turakhia said in 2025 that aggressive growth, not an IPO, was the priority. What stands out is the order of things. Turakhia had money and a name before Zeta, and still took the slow road: years of building a product no customer had yet asked for, then one anchor client, then the funds. The lesson for a first-time founder is that the anchor customer matters more than the big round. Sodexo gave Zeta revenue, a reference and a valuation in a single deal, and the later money followed it. ## Perfios (India) URL: https://startup.talk/india/perfios/ Founders: V.R. Govindarajan, Debasish Chakraborty Founded: 2008; Based in: Bengaluru, Karnataka; Sector: Fintech / B2B SaaS for lenders (bank statement analysis, credit underwriting, fraud, insurance) Seventeen years ago, lending in India mostly meant a branch, a stack of paper and a loan officer's hunch. Digital lending was barely a phrase. That was the world V.R. Govindarajan and Debasish Chakraborty walked into in Bengaluru in 2008, when they started Perfios as a software company. The idea they held onto was plain. They believed that in the finance of the future the scarce thing would be good decisions, not money. Cash would find borrowers. Knowing which borrower to trust would be the hard part. So Perfios built tools that read a bank statement, work out income and spending, and help a lender decide. For years that was patient, unglamorous work. Banks and NBFCs adopted new systems slowly, and a software vendor sold one institution at a time. Perfios did not report a profit until FY23, and that first profit was just Rs 7.8 crore. It took close to fifteen years to turn a loss-making start into a business that paid for itself. The wave that eventually lifted it was India's shift to digital lending. UPI, Aadhaar and account aggregators made it possible to approve a loan in minutes, and every lender doing so needed a way to check the applicant. Perfios was already inside those systems. The company says it now powers nearly 90% of India's digital loans and serves more than 1,100 financial institutions in 20 countries. Growth came partly from buying. In 2022 it bought Karza Technologies and FintechLabs, and handed the chief executive role to Sabyasachi Goswami. In 2025 it added Clari5, a fraud detection platform, CreditNirvana for collections, and IHX, a health-insurance claims platform. In 2024 an $80 million investment from Teachers' Venture Growth made it a unicorn. The numbers now look solid. For FY25 consolidated net profit was Rs 104.3 crore, up 46% from Rs 71.7 crore, on operating revenue of Rs 669.5 crore. Goswami has framed the ambition as 'Make in India for the world.' An IPO has been talked about since 2024, but no date has been disclosed. The lesson is in the timeline. Govindarajan and Chakraborty picked a problem that did not yet have a market, built for it, and waited. Founders tend to ask whether an idea is early. The better question may be whether, when the market arrives, you are already the thing it needs. ## Uniphore (India) URL: https://startup.talk/india/uniphore/ Founders: Umesh Sachdev, Ravi Saraogi Founded: 2008; Based in: Palo Alto, California (founded in Chennai, Tamil Nadu); Sector: Enterprise conversational AI and automation (Business AI Cloud) Two engineering students from Delhi walked into IIT Madras with a plan to become entrepreneurs and a professor to convince. They were Umesh Sachdev and Ravi Saraogi, and the professor was Ashok Jhunjhunwala, whose incubator on the campus took them in. The year was 2008. The pair had met at engineering college in Noida, and had first built a mobile-theft security product. Their first customers in India, among them MTNL and BSNL, were a start. Business Standard later reported that even those wins made the founders realise it would take more than this for the company to scale. A National Research Development Corporation grant of $100,000 helped them get going. The product that stuck was speech. Uniphore built software that listens to a customer call, recognises the language and turns it into data. India is a hard test for that: many languages, many accents, noisy lines. In 2014 it launched its first flagship products, auMina, Akeira and amVoice. By 2016 it supported over 70 languages and had opened in Singapore. A pivot followed in 2015. Uniphore stopped charging licence fees and moved to a subscription model. The same year Kris Gopalakrishnan, co-founder of Infosys, led its Series A, and IDG Ventures followed. In 2017 John Chambers, the former Cisco chief, led the Series B. Sachdev has said that Chambers pushed him to think beyond valuation and focus on building a team. The next step was moving the centre of gravity to California. Palo Alto became the headquarters, with the customer base shifting to large global enterprises: the company counts JPMorgan Chase, DHL, UPS, Verizon and Vodafone among them. A $51 million Series C came in 2019, $140 million in 2021, and in February 2022 a $400 million Series E led by NEA at a $2.5 billion valuation. Growth became acquisition-led. Emotion Research Lab, Jacada, Red Box, Hexagone, then ActionIQ, Infoworks and Orby AI. In June 2025 it launched a Business AI Cloud, and in October 2025 raised $260 million more, with Nvidia, AMD, Snowflake and Databricks among the investors, at a reported $2.5 billion. The pattern is useful. Uniphore started with a narrow technical skill, speech recognition in Indian languages, and kept moving it toward where the money was. For a founder, the lesson is that the product that wins your first customers may not be the one that builds the company. Be ready to follow the market, even across an ocean. ## Fractal Analytics (India) URL: https://startup.talk/india/fractal-analytics/ Founders: Srikanth Velamakanni, Pranay Agrawal, Nirmal Palaparthi, Pradeep Suryanarayan, Ramakrishna Reddy Founded: 2000; Based in: Mumbai, Maharashtra (also New York); Sector: AI, analytics and decision science for enterprises In August 1999, in a McDonald's in Bandra, Srikanth Velamakanni told a friend named Nikhil that he would start a business by the end of the year. A few months later, he did. He had grown up in Duliajan, a small oil town in Assam, where his father worked in materials management for an oil company. Honesty was a hard trade there. At about seven or eight, Srikanth saw men threaten his father with a pistol. His father believed that 'an honest businessman is an oxymoron,' and for years Srikanth accepted this. He went to IIT Delhi for electrical engineering, drawn more by the exam and by mathematics, and then to IIM Ahmedabad for an MBA. His plan was to join a multinational. Two things changed his mind, he said on a Blume Ventures podcast. Infosys co-founder Narayana Murthy gave a guest lecture on building a company on values. And venture money was arriving in India around 1999-2000. Months later, with former classmates, he co-founded Fractal in Mumbai. Selling analytics in 2000 was hard. To win ICICI, the team wrote a free 32-page white paper on credit-scoring models and then built the models. HDFC Bank later became one of the largest clients. Early money came from Gulu and Sasha Mirchandani, who, he said, backed the team more than the idea. His advice from those days: 'You need only one person to trust you for fundraising.' Then came the wound. Fractal started with six co-founders, later five. In 2006, at an investor's urging, they picked a CEO at a meeting in London. Srikanth was elected after RK declined the job. About six months later the founders proposed going back to equal partnership or splitting the company. The investors refused to split it. He later admitted, 'my ego is part of the problem,' and apologised. The company kept growing. Pranay Agrawal became CEO in 2016. Apax put in $200 million in 2019 and TPG $360 million in 2022, making Fractal a unicorn. In February 2026 it listed on the NSE and BSE, raising about Rs 2,834 crore, after an IPO subscribed 2.66 times. For FY25 it reported revenue of about Rs 2,765 crore and net profit of about Rs 221 crore. The lesson is in how the split ended. The founders could have broken the company in 2007. Instead the investors held it together, and a founder admitted his part in the damage. Co-founder fights are common. Fractal's survival suggests that saying so out loud is the way through them. ## Mindtickle (India) URL: https://startup.talk/india/mindtickle/ Founders: Krishna Depura, Nishant Mungali, Mohit Garg, Deepak Diwakar Founded: 2011; Based in: San Francisco, California (major office in Pune, Maharashtra); Sector: Sales readiness / sales enablement SaaS The first Mindtickle product was a game. In 2011, Krishna Depura and his co-founders ran online treasure hunts that used competition to keep employees engaged. Nothing about it suggested a billion-dollar company. Depura had met his co-founders, Nishant Mungali, Mohit Garg and Deepak Diwakar, back in 2006. They had five years of friendship and shared ideas before they sold anything. The treasure hunts worked as engagement tools, but they were a thin business. The founders then saw that the same approach could help new hires settle in. That covered only the early stretch of an employee's life at a company. The next step, and the one that stuck, was sales readiness: training, coaching and productivity tools for sales reps, with a way to measure whether any of it worked. Mistakes came along the way. In a 2022 interview with Entrepreneur India, the founders admitted two. They built the product before they had defined the customer's problem, and they under-invested in marketing and education. A mentor also warned them that serving only technology companies would cap their growth. They went after pharma and insurance, and then Fortune 100 and 500 customers. The pandemic helped. Remote work and digital transformation pushed companies to train sales teams online. In August 2021 SoftBank led a $100 million Series E that valued Mindtickle at $1.2 billion. Depura told Entrepreneur India the unicorn label was 'a byproduct of everything.' G2 ranked it the top enterprise product in its category for two years running, and it was hiring across the US, Canada, the UK and India with plans for Europe, the Middle East and Australia. The business has offices in Pune and San Francisco and hopes to stretch readiness beyond sales to every customer-facing team. The useful thing about Mindtickle is how unglamorous the path was: a game, then onboarding, then sales readiness. Depura described the pivots as reflections of changing markets, technology and customer needs. A first-time founder should read that as permission. The first idea is often the doorway, and the real company is in the room next door. ## Moglix (India) URL: https://startup.talk/india/moglix/ Founders: Rahul Garg Founded: 2015; Based in: Noida, Uttar Pradesh; Sector: B2B industrial commerce / supply chain Rahul Garg walked out of Google in January 2015 with a secure job behind him and, by his own account, something to prove. Moglix was not his first try. It was his third. He had grown up in Faridabad, the first entrepreneur in his family, and studied electrical engineering at IIT Kanpur, graduating in 2001. After a stint as a product manager at a chip company, he did an MBA at the Indian School of Business and joined Google, where he ended up running its ad exchange across India, Southeast Asia and Korea. From the outside it looked like a straight, safe line. What almost nobody knew was how crooked the line had been underneath. Before Moglix he had started two companies that, as he has acknowledged in interviews, did not work out. Both had co-founders. Both ended without the outcome he wanted. So when he quit Google in his mid-30s, he was not a first-timer with beginner's luck. He was a man with a record of failing, going back in. The idea came from watching consumer e-commerce work. If people could buy phones online, he reasoned, factories could buy bolts, bearings and safety gear the same way. Indian industrial purchasing was fragmented and slow. He wanted to be the Alibaba for it. The money did not come easily. Between 2015 and 2019, investors were wary of B2B. Accel and Ratan Tata backed him in October 2015, and Tata's cheque, after they met in 2016, was what Garg later called a moment of victory. For a founder with two earlier misses, an endorsement like that carried weight no pitch deck could. The company grew into industrial supply, financing through Credlix, and acquisitions in Singapore and India. A Series E of $120 million came in 2021 and a reported valuation of $2.6 billion in January 2022. Then the market turned. Moglix laid off about 40 people in 2023, around 3% of staff, and its FY22 numbers showed $307 million in revenue with losses widening to $22 million. Today Moglix says it serves about 3,000 manufacturing plants across India, Singapore, the UK and the UAE (2023). It is still unlisted, and the profit question is still open. The thing worth taking from Garg is not the unicorn tag. It is that he went into the third attempt carrying the first two, with experienced leaders around him and a clear thesis, rather than pretending they never happened. Failure, for him, was an input to the next company, not a line to be deleted from the story. ## Shiprocket (India) URL: https://startup.talk/india/shiprocket/ Founders: Saahil Goel, Gautam Kapoor, Vishesh Khurana, Akshay Ghulati Founded: 2017 (incorporated; origins in KartRocket, 2012); Based in: New Delhi; Sector: At four in the morning in Pittsburgh, Saahil Goel was already at work, and not for his employer. The day job, as a full-time consultant at an IT services firm, started later. This was the other job, the one nobody was paying him for. He told Entrepreneur India that he and his friend Gautam Kapoor started KartRocket, a platform to help small businesses sell online, and that he worked on it before dawn until he could move to India to do it full time. Press profiles also say he had been building websites since seventh grade. It was a long itch, scratched in the small hours. The pitch was simple: an Indian Shopify. It began when a family friend wanted to launch a shoe brand online and had nowhere good to do it. Goel moved home, and the product met the problem every early company dreads. Product-market fit, he has said, was the early challenge. Sellers could build a store, but that did not make them successful. What almost nobody outside the company noticed was where the pain really was. Small merchants were not stuck at building a storefront. They were stuck at shipping. At low volumes they could not get good rates or reliable pickups. Goel decided to stop doing many things and do one. He has said in interviews that focusing on building one thing at a time was crucial. So in 2017 Shiprocket was incorporated with Kapoor and Vishesh Khurana, and Akshay Ghulati joined as a co-founder in 2020. The first year was bootstrapped. Then angels came, and later Zomato and Temasek among others, for about $310 million across nine rounds. The company turned net profitable in March 2019, and by 2022 it was processing more than seven million shipments a month. In August 2022 it crossed a $1 billion valuation. The price of that focus was giving up the thing he first set out to build. The platform he dreamed of in the dark mornings in Pittsburgh became the side project, and the humble logistics layer became the company. Goel says he wants to build a platform that outlives him. The lesson he offers by example is quieter: the company you start is rarely the one you end up with, and the early mornings are only worth it if you are willing to let the idea change. ## Xpressbees (India) URL: https://startup.talk/india/xpressbees/ Founders: Amitava Saha, Supam Maheshwari Founded: 2015; Based in: Pune, Maharashtra; Sector: Logistics / express delivery Long before anyone called Amitava Saha a logistics founder, his business card said something far less glamorous: senior officer, Tata Steel. That is the part of his career that rarely makes the headlines. Saha studied at IIT (BHU) Varanasi and then did a PGDM at IIM Lucknow. Profiles describe him as a first-generation entrepreneur from a humble background, though that could not be traced to an interview he gave himself. What the record does show is a plain, steady apprenticeship. After Tata Steel came business development roles at NIIT Technologies and Aricent. Nothing flashy. Sales, targets, travel. What almost nobody connects to his name is how long he spent in jobs where somebody else owned the outcome. By the time he partnered with Supam Maheshwari on Brainvisa Technologies, he had years of sales and operations behind him. At Brainvisa he rose from director to senior vice president. Then came FirstCry, the baby-products e-commerce company they built together, where he was co-founder and COO. It was at FirstCry that the problem found him. Orders kept going out, and delivery decided whether customers came back. In 2015 the logistics arm became Xpressbees, in Pune, with Saha as chief executive. He discussed that spin-off in a 2016 podcast. The advantage was volume. The company had shipments from day one, a rare gift in a business that needs density to make money. Alibaba put in about $35 million in 2017. In February 2022 Blackstone, TPG and ChrysCapital led a $300 million round that took the valuation above a billion dollars. The cost has been the usual one for Indian logistics: money goes out faster than it comes in. Xpressbees reported a net loss of Rs 200 crore on revenue of Rs 2,831 crore in FY24, though it also reported a small positive EBITDA. By March 2025 it ran about 4,500 service centres and over 28,000 delivery partners. Saha's path suggests something quiet about founders. The loud origin story, the garage and the dropout, is not the only one. Sometimes the preparation is years of unglamorous learning about how things actually get sold and moved, and the company is simply the first problem big enough to use all of it. ## ixigo (India) URL: https://startup.talk/india/ixigo/ Founders: Aloke Bajpai, Rajnish Kumar Founded: 2007; Based in: Gurugram, Haryana; Sector: Online travel / rail and bus ticketing In March 2020 Rajnish Kumar sat on his bed and his watch buzzed. Heart rate, 120. He was not exercising. Bookings had gone to zero. Revenue had gone to zero. ixigo, a travel company he and Aloke Bajpai had been building since 2007, had nothing to sell while the country's trains and planes stood still. Open magazine reported that around then a buyer made an offer to take the company, an offer the founders found humiliating after fourteen years of work, with a demand to decide quickly. They said no. They were IIT Kanpur alumni who had worked in Europe, and had left that path to start a flight metasearch in India. What almost nobody knew was how thin their personal finances had become. After years of living without conventional salaries, Open reports, banks did not treat their founder income as income, and one of them had to put down a deposit just to get an ordinary credit card. That detail explains the rest. When ixigo bought the train platform ConfirmTkt in 2021, they signed personal guarantees to secure venture debt, which the magazine described as debt that knew their home addresses. In Covid, senior leaders took pay cuts of roughly 70 to 80 per cent, later converted into stock options. Both founders had young children and warned each other that time with them does not come back. Bajpai later went on a ten-day retreat in Gaya because his sleep was shallow, and his wife told him to go again. The turn came from a boring place. Indians kept checking train status. ixigo had built a rail product, and ConfirmTkt and AbhiBus added trains and buses to it. By the time travel reopened, the company was ready. In June 2024 ixigo listed on the NSE and BSE, and its shares debuted about 48.5% above the issue price. For FY24 it reported revenue of Rs 655 crore and a net profit of Rs 73 crore. Open says an employee had once asked Bajpai when they would do an IPO, and an uncle who had doubted the company's finances admitted he had been wrong. The credit card is the moral, though neither founder frames it that way. For years they bet their own paychecks on the company. A founder's belief is cheap until the bank refuses to see it as income. ## Upstox (India) URL: https://startup.talk/india/upstox/ Founders: Ravi Kumar, Raghu Kumar, Shrini Viswanath, Kavitha Subramanian Founded: 2008 (RKSV; retail brokerage Upstox from 2012); Based in: Mumbai, Maharashtra; Sector: Fintech / discount stockbroking In October 2024 Ratan Tata sold part of a stake he had bought eight years earlier, and Upstox said it had returned 23,000% on his original money. The number was so large it sounded like a typo. It says something about how quietly Upstox was built. The company's parent, RKSV Securities, was founded around 2008 by Shrini Viswanath, Raghu Kumar and Ravi Kumar, and it did not start as a consumer brand at all. It was a proprietary trading firm, a business of people trading with their own systems. Retail customers came later, in 2012, when the team opened the same plumbing to ordinary investors. Founder interviews revealing childhood, family or early money trouble could not be found, so the honest part of this story is the business one. Kalaari Capital wrote an early $4 million cheque. Ratan Tata took a 1.33% stake in 2016. In 2019 Tiger Global bought around 31% of the company and later added more, putting in about $50 million across two rounds. Then came the competitor everyone in Indian broking knows. Zerodha had already shown that low fees could win millions of accounts. Upstox had to lean on what it already owned: trading technology built for itself, and a brand tied to Tata's name. That worked. Operating revenue crossed Rs 1,000 crore in FY23, up 44% from Rs 765.6 crore the year before, and the company reported its first profit, Rs 25 crore. Reported FY25 profit was around Rs 215 crore. It claims over 10 million customers and a last valuation of $3.5 billion. The founders hold a large share still, reportedly around 36%, with Tiger Global at more than 38%, which means the people who started on a prop desk are now minority owners of the company they built, a trade-off every venture-backed founder makes. What the record cannot give is the human part: the doubt, the late nights, the first customer. A founder reading this should notice that absence. Not every company leaves a trail of colourful anecdotes, and some are built by people who never told theirs. The Upstox lesson is structural: build for yourself first, in a hard business, and the customers can come later. ## Juspay (India) URL: https://startup.talk/india/juspay/ Founders: Vimal Kumar, Sheetal Lalwani, Ramanathan RV Founded: 2012; Based in: Bengaluru, Karnataka; Sector: Fintech / payments infrastructure In 2014, Sheetal Lalwani had a job at Bloomberg that most people would have kept. She left it to join a two-year-old payments startup in Bengaluru that few had heard of, run by a former Amazon engineer named Vimal Kumar. The company was Juspay. It had been started in 2012 by Kumar and Ramanathan RV, at a time when India's online payments were a mess. Cards failed, banks timed out, and every merchant had to fight the same fight alone. Juspay's pitch was unglamorous: build the layer in between, and make the failures disappear. The public record is thin on the personal side. We do not know where the founders grew up, what their parents said, or what their first office looked like. What the record does show is how slow the early years were. An early cheque of around Rs 40 crore came from the founders of Accel and BookMyShow, money meant for product, not marketing. For years Juspay sold to other companies, not to consumers, which meant no app downloads to boast about and no brand to speak of. The nobody-noticed part is that the business was growing underneath. Amazon, Flipkart, Google, Swiggy, IndiGo and HSBC became clients. When the government's UPI system took off, Juspay already had the stack to sit under it, and it now says it handles more than 300 million transactions a day. There was a price. In 2019 co-founder Ramanathan RV left. In FY24 the company lost Rs 97.5 crore. In February 2024 the RBI finally granted it a payment aggregator licence, letting it move from advising merchants to moving their money itself. Then the numbers turned. In FY25, operating revenue rose 61 percent to Rs 514 crore, and a Rs 97.5 crore loss became a Rs 62.3 crore profit. In January 2026 WestBridge Capital led a $50 million round at a $1.2 billion valuation, making Juspay the first unicorn of the year. Part of that money bought out early investors and employees who had waited more than a decade. Lalwani has said the aim is to build a global payments company from India with a billion dollars in revenue. It is a big sentence for a company most users have never heard of, even though they have probably paid through it. For a first-time founder the lesson is a quiet one. Some companies are built in the places nobody sees: between two other companies, behind an interface someone else owns. The reward is slow and the credit goes elsewhere, but twelve years later the plumbing is worth a billion dollars. ## Khatabook (India) URL: https://startup.talk/india/khatabook/ Founders: Ravish Naresh, Ashish Sonone, Dhanesh Kumar, Jaideep Poonia Founded: 2018; Based in: Bengaluru, Karnataka; Sector: Fintech / MSME digital ledger The story most people tell about Khatabook begins in 2018 with a founder looking at a shopkeeper's notebook. The record is slightly stranger. Khatabook, the app that now sits in the pockets of millions of Indian shopkeepers, was not built by the people who made it famous. Vaibhav Kalpe built the original product. In October 2018 a small company called Kyte Technologies acquired it. Kyte was run by Ravish Naresh and a group of IIT Bombay friends, among them Ashish Sonone, Dhanesh Kumar and Jaideep Poonia. They had been through their own version of failure. In 2016 Naresh and his team had started a digital spend manager called Kyte.ai. It was not the thing that mattered. Naresh had been through bigger rooms before. DNA India reports that he was earlier co-founder and chief operating officer of Housing.com, a startup that had grown fast and in public. Having seen one company at scale, he knew what the next one needed: a use case so boring that no one could argue with it. The observation was simple. Indians had learned to pay by phone, but a corner store owner who gave credit to neighbours still tracked it in a paper ledger. Digital payments had arrived, but bookkeeping had not. Kyte folded the acquired product into its own work and pushed hard. The company went through Y Combinator and Sequoia's Surge programme, and cricketer MS Dhoni put in an undisclosed sum. By May 2020 B Capital led a $60 million round that reportedly valued Khatabook between $275 million and $300 million. At that point it claimed over 8 million active businesses and more than $200 million a day in transactions, in 11 languages. The next step was larger. In August 2021 Tribe Capital and Moore Strategic Ventures led a $100 million Series C at a valuation of almost $600 million. What the record does show is a useful correction to startup mythology. The founders did not need to invent the idea. They needed to see that someone else's small product was the right shape, buy it, and then carry it to millions. Sometimes the first job is not to build the thing but to notice it. ## Polygon (India) URL: https://startup.talk/india/polygon/ Founders: Sandeep Nailwal, Jaynti Kanani, Anurag Arjun, Mihailo Bjelic Founded: 2017 (Matic Network launched early 2018; renamed Polygon Feb 2021); Based in: Distributed team; founded in India (Mumbai); Sector: Web3 / Ethereum scaling (Layer 2) The money was meant for a wedding. Sandeep Nailwal had borrowed about $15,000 for it, and instead he put it into Bitcoin at roughly $800 a coin. He later called the move very risky. That sentence, from a Cointelegraph Magazine profile, says more about him than any valuation. He was born in 1987 in Ramnagar, a Himalayan foothill village with no electricity, and at about four his family moved to Delhi. They landed in the poor neighbourhoods on the east bank of the Yamuna, which he has described as the Bronx of New York. His father struggled with alcohol and gambling and earned perhaps $80 to $90 a month, much of it lost. School fees went unpaid. Nailwal recalls being made to stand outside the school, and calls it traumatic. At 10 he began looking after his baby brother after an accident, which he says ended his childhood. As a teenager he sold pens from a friend's shop at a markup and tutored other students. He could not afford the coaching that IIT entrance demanded, so he took a loan and studied computer engineering at MAIT in Delhi. An MBA, a job at Deloitte and another at Welspun Textiles followed. Then came the failed things. A GPS-based cargo optimiser. A B2B project platform. A B2B services marketplace he ran full time for about a year before deciding it would never grow the way he wanted. What almost nobody remembers is that he had heard of Bitcoin in 2013 and dismissed it. Only after it endured did he change his mind, and that wedding money went in. In 2016 he started Scope Weaver, a blockchain services firm, and became a regular voice on local Ethereum forums. There he met a programmer, Jaynti Kanani, who pushed him to fund his own ideas. Ethereum clogged in 2017. They and Anurag Arjun chose to build a scaling layer, raised $30,000 in seed funding and launched Matic Network in early 2018. The first two years ran on about $165,000. They delayed the token sale until they had a product, a decision Nailwal later regretted. A $500,000 Chinese fund commitment collapsed when Bitcoin fell, and the treasury, held in Bitcoin, lost half its value. The rescue was a $5.6 million Binance Launchpad sale after eight months of due diligence. In February 2022, renamed Polygon, it raised $450 million led by Sequoia Capital India. JPMorgan ran a live trade on it in November 2022. Then the hard years: a stolen 801,601 MATIC in 2021 and 60 layoffs in February 2024. The boy made to stand outside a school gate for unpaid fees now runs one of India's most visible global crypto projects. What he took from the gate, he has suggested, was a clear idea of the man he did not want to be. ## CoinDCX (India) URL: https://startup.talk/india/coindcx/ Founders: Sumit Gupta, Neeraj Khandelwal Founded: 2018; Based in: Mumbai, Maharashtra; Sector: Crypto exchange / Web3 On the last day of July 2025, Sumit Gupta had a problem that no founder wants. His exchange had just lost about $44 million in a hack. Newspapers were reporting that Coinbase was in talks to buy it, at under a billion dollars. Gupta had to tell the world, publicly, that this was not true. The personal record behind this man is thin. We know that he and Neeraj Khandelwal both studied at IIT Bombay, and that they started CoinDCX in 2018, when owning Bitcoin in India felt like an act of faith. We do not know where either grew up, what their families made of crypto, or how the first months felt. That is the honest limit of what we could verify. What the record does show is a pattern of surviving moments when a company could have ended. India's central bank had once banned banks from serving crypto firms. The country later imposed a 30 percent tax and a 1 percent deduction at source on trades, which cut trading volumes sharply. CoinDCX kept going. In August 2021 it raised $90 million at a $1.1 billion valuation, becoming India's first crypto unicorn. Soon after, another round of $135 million reportedly doubled its value. Coinbase, the American exchange, was already an investor, having first put money in during 2020. In July 2024 CoinDCX bought BitOasis in Dubai. Then the hack. About $44 million, or Rs 378 crore, left the exchange. The company said customer funds were untouched. In the following days Khandelwal told users that traffic had pushed its servers and asked them to ignore rumours. A buyer did not appear. Instead, in October 2025, Coinbase invested again, at a post-money valuation of $2.45 billion. The numbers show a company that is large, not rich: FY25 operating revenue of Rs 559.6 crore, up 43 percent, and a profit of Rs 1.7 crore. It claims over 2 crore users and trading volumes above Rs 2.4 lakh crore a quarter. The lesson is almost administrative. In a regulated-adjacent business, a founder's job is often to hold the line in public while the story is still unclear: deny the false thing, admit the true thing, and keep customers. Investors, it seems, noticed. ## Krutrim (India) URL: https://startup.talk/india/krutrim/ Founders: Bhavish Aggarwal Founded: 2023; Based in: Bengaluru, Karnataka; Sector: Generative AI / AI cloud and chips In January 2024 Bhavish Aggarwal made India's first AI unicorn. Eighteen months later, he was cutting staff. He was born in Ludhiana in 1985 and studied computer engineering at IIT Bombay, graduating in 2008. He then worked at Microsoft Research India for about two years, filing two patents and publishing three papers. A bad taxi ride is the story told about why he left to start Ola. That is how a researcher became a mobility founder. TechCrunch reported that Aggarwal put in $230 million of his own. A $50 million round led by Matrix Partners India in January 2024 valued Krutrim at $1 billion. The idea was to build India's own large language model, trained for Indian languages, with Indian chips and cloud behind it. For a moment it looked like the next Ola: loud, fast, backed by its founder's conviction. The story after that is about the gap between a promise and a product. In 2025, after the Kruti assistant launched, about 100 employees were let go. Reports say headcount fell from more than 550 in August 2025 to roughly 150 or 160 by March 2026, and that Kruti was pulled from app stores in April. A later round in 2026 reportedly cut about half the remaining staff. Fundraising targets fell, one report says from $500 million to $300 million, because investors showed little interest. The company's answer was to change the shape of the business. LLM and chip work paused, and AI cloud services came to the front. In May, Krutrim said revenue for fiscal 2026 had crossed Rs 300 crore and that it was profitable. We could not find independent confirmation. The lesson for first-time founders is hard and specific. A valuation is a promise, not a market. Aggarwal's backing bought time to find out which parts of the AI stack Krutrim could actually sell. The founder who can afford to be wrong for longer is not safe from being wrong. He is just wrong at a larger scale. ## Sarvam AI (India) URL: https://startup.talk/india/sarvam-ai/ Founders: Vivek Raghavan, Pratyush Kumar Founded: 2023; Based in: Bengaluru, Karnataka; Sector: Generative AI / Indian-language foundation models In 2010, a software engineer started work on the largest identity system anyone had tried to build, and by one profile he did it for free. Vivek Raghavan joined the Aadhaar programme as its chief biometric architect in a voluntary capacity. No salary was the point. The problem was. Most of what the public knows about him begins there. What the record does show is a long, unglamorous apprenticeship in building things for a billion people. After Aadhaar he worked in product at UIDAI, helped bring AI into identity checks and fraud detection, then moved to the EkStep Foundation as its chief AI evangelist. At AI4Bharat, the research lab at IIT Madras, he mentored alongside Pratyush Kumar. The two were working on open-source AI for Indian languages while the world was still arguing about whether chatbots could be trusted. What almost nobody noticed was the pattern. Each job had been public-interest work, often done without the usual startup rewards. In August 2023 the two men turned that habit into a company. Sarvam AI raised roughly $41 million that December from Lightspeed, with Peak XV and Khosla Ventures joining. That was a large cheque for an Indian AI seed, and a small one against the labs in San Francisco. The bet was narrow on purpose. Sarvam would build for Indian languages and for voice, not clone an English chatbot. It released Sarvam-1 in October 2024. In April 2025 the IT ministry picked it to build an indigenous foundation model under the IndiaAI Mission, with government-backed GPU compute, and UIDAI announced work on voice features for Aadhaar services. Then came the heavier models, 30B and 105B, shown at the India AI Impact Summit in February 2026, and in March the weights went out openly under an Apache 2.0 licence. There is a cost to that kind of ambition, and the sources only hint at it: public money brings public scrutiny, and the company has not disclosed revenue. In June 2026 it raised $234 million at a $1.5 billion valuation, with HCLTech putting in $150 million, which made it a unicorn. For a first-time founder the lesson sits in the early years rather than the valuation. Raghavan spent over a decade on problems that paid in credibility, not cash, and when he finally asked investors for money, he already knew the problem better than anyone in the room. The unpaid years, it turns out, were the research. ## MobiKwik (India) URL: https://startup.talk/india/mobikwik/ Founders: Bipin Preet Singh, Upasana Taku Founded: 2009; Based in: Gurugram, Haryana; Sector: Fintech / digital payments Bipin Preet Singh has said he left South Delhi for Dwarka because South Delhi cost too much. That was 2009, and the new flat was not just a home. It was the first office of MobiKwik. He was an IIT Delhi graduate, class of 2002, who had worked at chip companies including Intel, Nvidia and Freescale. His wife, Upasana Taku, was in her own corporate job. Neither was happy. A profile reports that Singh put in life savings of about ₹8 lakh, though that comes from a promotional source and should be held loosely. The couple, by their own telling, wanted out of the corporate track. Singh was the technical one. He built the site and the payment options himself, and in a Quartz interview he recalled that the early system needed rebooting so often that he carried his laptop everywhere. A payments company whose founder cannot leave his laptop is a fragile thing. The product was a mobile wallet at a time when few Indians trusted putting money into one. What almost nobody remembers is how long it stayed small. The RBI authorised the wallet in 2013. Funding came in 2015 with a $25 million round led by Tree Line Asia, and another $50 million in 2016. Then November 2016 happened. When the government pulled high-value notes, MobiKwik transactions rose 400%, and a company that had been explaining itself for seven years suddenly had users arriving unprompted. The trouble followed the growth. In May 2020 the app disappeared briefly from Google Play over an ads policy violation. In early 2021 a researcher claimed hackers had taken KYC data on nearly 100 million users; the company first denied it and later commissioned an outside forensic audit. In 2022 it suspended wallet transfers to major crypto exchanges. In September 2024 it froze withdrawals on its Xtra peer-to-peer lending product, and customers were angry. The IPO it filed in 2021 for ₹1,900 crore was shelved. In December 2024 it tried again, raised ₹572 crore, and listed on the NSE and BSE. The last full-year numbers in hand, for FY24, show revenue of ₹875 crore and a net profit of ₹14 crore: a company that took fifteen years to turn the corner, and only just. The lesson for a first-time founder is smaller than the headline. Singh did not win by being first or loudest. He stayed alive long enough, in a rented flat that doubled as an office, for the country to catch up to the thing he had already built. ## BlueStone (India) URL: https://startup.talk/india/bluestone/ Founders: Gaurav Singh Kushwaha, Vidya Nataraj Founded: 2011; Based in: Bengaluru, Karnataka; Sector: Omnichannel jewellery retail (D2C) Gaurav Singh Kushwaha is a jeweller who never trained as one. Before BlueStone he was an engineer, an IIT Delhi graduate who had worked at Amazon. He did not come from the jewellery trade, he came from the internet. In an industry run by family firms with generations of ledgers, that was the first strike against him. In 2011 he and co-founder Vidya Nataraj started BlueStone in Bengaluru as an online jeweller, at a time when Indians bought gold in person, from a man they knew, often for a wedding. That bet looked wrong for years. Jewellery is a trust purchase. You want to hold it, weigh it, argue about it. A website could show a ring, but it could not give a customer that comfort. What almost nobody expected was how BlueStone would answer: not by abandoning the internet, but by walking into the real world on its own terms. It built stores, many of them, and by FY25 about 93% of revenue came through the offline channel. The price of that walk was steep. In FY25 the company grew revenue about 40% to roughly ₹1,770 crore, but its net loss climbed to about ₹221.8 crore from ₹142.2 crore a year earlier, a 56% rise, which the business press tied directly to the store push. Stores cost money long before they earn it. Then came the market test. BlueStone opened its IPO on 11 August 2025 with the fresh issue trimmed 18% to ₹820 crore, a signal that investors wanted a lower price for the risk. The issue was priced at ₹517 a share. Kushwaha stayed the largest individual shareholder, at about 17.7%. Reports since suggest the bet may be paying off: FY26 operating revenue of about ₹2,441 crore and what one outlet called a first full year of profit, though accounts differ on the size of it. For a founder reading this, the useful point is how Kushwaha handled being an outsider. He did not pretend to know the trade. He used the internet to learn what customers wanted, and then went where they were. Being the outsider cost him a decade of losses. It also meant he was never defending an old way of doing things. ## Purplle (India) URL: https://startup.talk/india/purplle/ Founders: Manish Taneja, Rahul Dash, Suyash Katyayani Founded: 2011 / 2012 (others); Based in: Mumbai, Maharashtra; Sector: Beauty e-commerce and own brands The company now valued at over a billion dollars for selling lipstick was, at first, an idea about furniture. Three friends, Manish Taneja, Rahul Dash and Suyash Katyayani, set out more than a decade ago to build a furniture brand. They ended up in beauty. Taneja never planned to be an entrepreneur. As a schoolboy at Modern Vidya Niketan in Faridabad he was curious how things worked, according to a Financial Express profile. He went to IIT Delhi, and Purplle was reportedly his second startup. Dash studied at IIT Kharagpur and IIM Ahmedabad, Katyayani at IIT Kharagpur, and he became the company's CTO. What the record does show is the pivot. The idea moved from furniture to beauty, an aisle where Indian shoppers wanted choice and cared about authenticity. Purplle began, as many e-commerce firms did, as a marketplace. The turn came when it started holding its own inventory and building its own brands. What almost nobody outside the industry sees is how expensive that decision is. In FY25, procurement costs rose 5.6 times to ₹671 crore because Purplle was now buying stock rather than just listing it. Losses were the cost of the switch. In FY24 the company spent ₹1.25 for every rupee of operating revenue and lost ₹124.1 crore. Then the numbers turned. In FY25, operating revenue rose about 101% to roughly ₹1,367 crore, and the net loss fell 44% to ₹69.4 crore, with the cost per rupee earned dropping to ₹1.08. One outlet notes a one-time ₹20 crore non-compete fee sat inside that year. The company closed a ₹1,500 crore Series F round and became a unicorn, and reports point to an IPO filing around late 2026 with a valuation target near $1.25 billion. Nothing is filed yet. The founders held about 15.6% as of September 2024, which is the quiet price of a decade of fundraising: three people who built the company own less than a sixth of it. For someone starting up, the furniture detour is the point. Purplle's founders did not need the first idea to be right. They needed to be three people who could stay together while the idea changed shape under them, and who were willing to carry the cost of inventory when the easy marketplace model stopped working. ## Sugar Cosmetics (India) URL: https://startup.talk/india/sugar-cosmetics/ Founders: Vineeta Singh, Kaushik Mukherjee Founded: 2015; Based in: Mumbai, Maharashtra; Sector: Colour cosmetics / D2C beauty In 2006 Vineeta Singh was an intern at Deutsche Bank, and the bank offered her a job paying ₹1 crore a year. She said no. She was twenty-three. She had been born in Anand, Gujarat, in 1983, gone to Delhi Public School in R.K. Puram, then to IIT Madras for electrical engineering and IIM Ahmedabad for an MBA. The track was obvious. A crore a year at that age was the top of it. She stepped off. What almost nobody saw was what came next. Her first company, Quetzal, launched in 2007 to run background checks for recruiters, and it did not work. In 2012 she tried again with Fab-Bag, a subscription box of beauty products. Business Today has described three failed ventures in all before SUGAR. What is plain is the arithmetic: one refused crore, then years of businesses that did not hold. SUGAR came in 2015, with her husband, Kaushik Mukherjee, whom she married in 2011. Raising money was its own test. She has said one investor would back the company only if her husband joined full-time, a condition she described as gender discrimination. Later came a $50 million deal with L Catterton, and in 2022 actor Ranveer Singh invested. The numbers show the fourth try working. Revenue rose from ₹127.8 crore in FY21 to ₹420 crore in FY23 and ₹505 crore in FY24. From 2021 she also became a familiar face as a judge on Shark Tank India, which gave a lipstick brand the reach no ad budget could. Then the correction. In FY25 revenue fell around 18 to 20 percent, to about ₹404 to 415 crore depending on the source, and the net loss almost doubled to roughly ₹135 crore. Reports said the valuation had dropped from around ₹3,000 crore to about ₹500 crore, and the IPO she had said was two to three years away in early 2024 was pushed back by two years. She is also a triathlete who finished the Ironman in Austria in 2017, which says something about her tolerance for long, slow pain. The takeaway for a first-time founder is uncomfortable. The ₹1 crore she turned down is a famous detail, but the unglamorous ones matter more: two companies that failed before this one worked, and a brand that stumbled after it did. Founders rarely get one clean win. She seems to keep going through the interval between them. ## Skyroot Aerospace (India) URL: https://startup.talk/india/skyroot-aerospace/ Founders: Pawan Kumar Chandana, Naga Bharath Daka Founded: 2018; Based in: Hyderabad, Telangana; Sector: Space tech / private launch vehicles Naga Bharath Daka had already walked away from rockets once. He was an avionics engineer at ISRO, the kind of job an IIT Madras electrical engineer in India is supposed to hold for life. Then he left for three years at Xilinx, a semiconductor company, where the work was chips and not launches. Most profiles skip that detour. It is the part that makes Skyroot make sense. His co-founder, Pawan Kumar Chandana, took the straighter road. Schooled in Andhra Pradesh, drawn to maths, physics and later rockets, he went to IIT Kharagpur for a B.Tech and M.Tech in mechanical engineering, then joined ISRO as a scientist in 2012. He spent nearly six years there on launch-vehicle propulsion. Daka, on the electronics side, had been a colleague at ISRO. So one man knew how to make fire go the right way. The other knew what chips and avionics could do cheaply. Neither had a problem with the engineering. The problem was India. What almost nobody outside the community saw was how little ground there was to stand on. Chandana has said that when they started in 2018 there was neither a private space-tech policy nor an environment to fund space ventures. A rocket company needs hundreds of crores before it earns one rupee, and the only customer, the government, had not yet agreed to let a private firm launch anything. They registered the company on 12 June 2018 anyway, in Hyderabad. Four and a half years later, in November 2022, Vikram-S went up about 89.5 km, the first privately built rocket ever launched from India. It was a suborbital hop, but it changed what investors believed was possible. In October 2023 Temasek led a $27.5 million round that took total funding to about $95 million. Then came the job that actually mattered. In July 2026 Vikram-1 reached orbit on Mission Aagaman, placing its payload at 450 km and making India the third country after the United States and China with a private orbital launch. By May 2026 a further round of about $60 million had pushed the valuation past $1.1 billion, India's first space-tech unicorn. Vikram-2 is planned for 2027, and a Reflect Orbital contract is already signed. None of this is smooth. Launch capacity worldwide is short, the vehicle has to keep flying, and one failure can cost years. Skyroot's founders know that better than anyone; they have spent their careers around rockets that did not make it. For a first-time founder, the quieter lesson sits in that Xilinx detour. Daka stepped out of the field he loved, learned a neighbouring craft, and came back with something his rocket-only colleagues did not have. The years that look like a pause on a CV are sometimes the ones that complete the team. ## Agnikul Cosmos (India) URL: https://startup.talk/india/agnikul-cosmos/ Founders: Srinath Ravichandran, Moin SPM, Satyanarayanan Chakravarthy Founded: 2017; Based in: Chennai, Tamil Nadu (at NCCRD, IIT Madras); Sector: Space tech / small-satellite launch vehicles Srinath Ravichandran emailed and cold-called about 70 to 80 professors. Most did not answer. One did. That is how the founder of Agnikul Cosmos remembers the search for someone who would take a rocket idea seriously. He had nothing to show them but conviction. He was a finance man in Los Angeles who had been following ISRO launches since childhood in Chennai, glued to the television whenever one went up. His degrees did not point at space. He studied electrical engineering at the College of Engineering, Guindy, took an office job in Bengaluru, then a master's in financial engineering at Columbia and a Wall Street job. By his own account the finance career did not hold his interest. What almost nobody knew was what he was doing in the evenings. While working in Los Angeles, he was completing an aerospace engineering master's online from the University of Illinois, teaching himself the subject he wanted to work in without leaving a paycheque. Los Angeles was also where he saw the problem. Satellites, finished and ready, were sitting in university labs because there was no affordable launch vehicle to carry them. Having worked around jet engines, he said, he had some ideas for rockets too, but nobody had tested them because rockets were usually so big. The reply that mattered came from Prof. Satyanarayanan Chakravarthy, head of the National Centre for Combustion Research and Development at IIT Madras. He backed the idea and gave the team access to the propulsion lab. Moin SPM, then 27 and looking to move from manufacturing into aerospace, joined as co-founder. They formally started on 1 December 2017. Early on, the two pooled their own savings. Investors were harder than professors. Srinath has said he was never very conversant and had to learn networking. The pandemic pushed back the launch plans. Targets slipped, including early 2025. A planned commercial launch deal with Alaska Aerospace fell through. Still, the company kept clearing hurdles. A framework agreement with the Department of Space came in September 2021, a private launch pad at Sriharikota in November 2022, and a $26.7 million Series B in October 2023, bringing total capital to about $40 million. On 30 May 2024 Agnibaan flew its suborbital test. By early 2026 it was firing three semi-cryogenic engines at once, and in September 2026 the Technology Development Board signed Rs 200 crore for a reusable vehicle. The takeaway is plain. Srinath did not have the right degree, the right job or the right city. He had patience for seventy-odd silences and one yes, and he used his evenings to become credible before he asked anyone for anything. ## Lendingkart (India) URL: https://startup.talk/india/lendingkart/ Founders: Harshvardhan Lunia, Mukul Sachan Founded: 2014; Based in: Ahmedabad, Gujarat (operations also Bengaluru); Sector: Fintech / MSME lending (NBFC) In November 2025, three men from one family put their names on a petition to India's company-law tribunal. Harshvardhan Lunia, his father Raichand and his elder brother Anand were asking the court to remove the management of a company Lunia had built from his Ahmedabad home town. Lendingkart was supposed to be the story of a self-made business family's son who did everything right. He was a chartered accountant and an ISB graduate. He had worked at HDFC Bank and Standard Chartered, on small-business lending, so he knew exactly how badly an Indian shopkeeper was served by the formal system. He then took a job in London and, as the YourStory profile frames it, walked away from it to go back to Ahmedabad. His family, one profile notes, was baffled. He founded Lendingkart in 2014 with Mukul Sachan, an IIT and IIM alumnus. The pitch was simple: use data, not collateral, to lend working capital to small firms in days, not weeks. By September 2023 the company had more than 110,000 active loan accounts and about Rs 6,000 crore under management. What almost nobody saw from outside was where control was drifting. Fullerton Financial Holdings, a Temasek unit and the lead investor, grew its presence. In 2022 the idea of buying a personal-loan fintech, Upwards, reportedly came from Fullerton, and Lunia, according to The Ken, became less involved in decisions as Fullerton's CEO and COO moved in. Then came the fall. Valuation reportedly sank from about $700 million in May 2024 to under $100 million by October, and the company ran short of cash. Lunia was removed from management and the board while Fullerton raised its stake from 38% to 77%. Lendingkart Finance slipped from a Rs 60 crore profit in FY24 to a Rs 288 crore loss in FY25, with defaults eating Rs 523 crore. In FY26 revenue fell 62% to Rs 327 crore and the loss widened to Rs 333 crore. His side of the argument is now before a tribunal, and his father and brother are in it with him. For a founder, the lesson is hard and old. Taking the cheque of a large investor also means signing a story where you may not be the last chapter. Lunia built a company in his home city for a decade. The question that his petition poses, still unanswered, is what he actually kept. ## Slice (India) URL: https://startup.talk/india/slice/ Founders: Rajan Bajaj Founded: 2016; Based in: Bengaluru, Karnataka; Sector: Fintech / small finance bank Rajan Bajaj was 22 when he put his name on a company. According to profile summaries, the previous line on his resume was an internship as a business analyst at Walmart. Before that, there was another venture, Mesh Internet. Slice was not his first start. It was the one that would not let go. He had trained as a civil engineer at IIT Kharagpur. Civil engineers build bridges. He built a payments app instead. A YourStory profile frames the turning point as an Airbnb moment: seeing a young company reshape something ordinary and deciding to try one himself. SlicePay launched in January 2016 as a buy-now-pay-later product for students and young professionals, people who had no credit history and so, to a bank, did not exist. Then came years of pivots. The company went through several versions before it landed on a physical card in 2019, the year it got an NBFC licence from the RBI. What almost nobody saw was how narrow that road was. Slice was built on RBI's tolerance for credit lines on prepaid instruments. It worked. In November 2021 the company raised about $220 million and became a unicorn. Seven months later, in June 2022, the RBI barred NBFCs from giving credit lines on prepaid instruments, and Slice had to shut its prepaid cards. The product that carried its name stopped. Bajaj did something unusual with the shock. Instead of pivoting again, he looked for a bank. Slice took a stake of about 5% in North East Small Finance Bank, announced a merger in October 2023, and completed it a year later on 27 October 2024, India's first fintech-bank merger of its kind. The bank was renamed Slice Small Finance Bank in May 2025, and the RBI cleared Bajaj as MD and CEO. The money kept coming, though at lower prices. By December 2024 Slice had raised $387 million in all, from investors including Tiger Global and Insight Partners; in September 2026 it raised another $100 million at a reported $450 million valuation. For FY26 the bank reports revenue of Rs 1,403 crore and a net profit of Rs 48.4 crore. The founder-level lesson is not that the bank pivot was clever. It is that Bajaj built the whole thing at an age when most people are on their first job, got hit by a regulator, and chose to become part of the regulated system rather than to argue with it. Sometimes the most credible answer to the rulebook is to start playing inside it. ## Navi (India) URL: https://startup.talk/india/navi/ Founders: Sachin Bansal Founded: 2018; Based in: Bengaluru, Karnataka; Sector: Fintech / lending, insurance, mutual funds, UPI Sachin Bansal had the rank to get into IIT. He took the exam again anyway. His first JEE attempt had put him at All India Rank 2470 — high enough for a seat in most engineering colleges in the country. He sat the paper a second time and came out at 49. Then he chose IIT Delhi and computer science. That is the line almost no profile of the Flipkart co-founder leads with, and it is the one that explains the rest. He grew up in Chandigarh, where his father ran a business and his mother stayed home, and attended St. Anne's Convent School. After IIT he worked at Techspan for a few months and then at Amazon Web Services, until 2007. That October he and Binny Bansal, no relation, started Flipkart as an online bookstore with about Rs 4,00,000. Flipkart became the biggest story in Indian e-commerce. In 2018, when Walmart bought control, Bansal sold his 5.5% stake, and was reported to be worth a little over a billion dollars. A non-compete barred him from competing businesses for 18 months and from competing investments for 36. What almost nobody noticed at the time was how fast he moved. He set up BAC Acquisitions that December. In 2019 he put $100 million into Ola. He bought Maven Hive, and then the COCO insurance brand in 2020, and Navi Technologies was announced in April 2020. A draft IPO for Rs 3,350 crore followed in March 2022, though the listing never happened. The price of speed came later. In October 2024 the RBI restricted lending at Navi Finserv. The ban was lifted, and Navi Finserv still posted revenue of Rs 2,271 crore in FY25, up 19%, but profit fell 67% to Rs 222 crore, and the group swung to a loss of about Rs 126 crore. Bansal stepped down as CEO and became executive chairman. In July 2026 the company was reportedly working toward a fresh IPO in 12 to 18 months. The thing worth carrying away is the second JEE paper. Bansal did not stop at a good result. He re-took the test for a better one. Years later, he left a company worth billions to start over. People who are comfortable with a second attempt tend to be good at the third. ## Info Edge (India) URL: https://startup.talk/india/info-edge/ Founders: Sanjeev Bikhchandani Founded: 1995; Based in: Noida, Uttar Pradesh; Sector: Internet classifieds and investments (jobs, real estate, matrimony, education) The first office of what became India's most valuable job site was, by several profiles, a servant's quarter behind his father's house, and the founder paid his own father Rs 800 a month to sit in it. Sanjeev Bikhchandani was born in Delhi in 1963 to a Sindhi family. By one profile his father was a government hospital doctor and his mother a homemaker, and the family hoped for a doctor or an engineer. He took a different road: St. Columba's, then economics at St. Stephen's, then the MBA at IIM Ahmedabad in 1989. It was a respectable path. The job after it, marketing Horlicks for Hindustan Milkfood, was respectable too. He left in 1990. A partnership he then tried ended in 1993. He started Info Edge in 1995, selling things like salary surveys, small cheques, and a little pride. What almost nobody sees in the glossy version is who paid the bills. Several profiles say his wife Surbhi, an IIM-A batchmate with her own corporate job, carried the household for the first years. The details vary by telling, and I have kept them loose, but the shape is consistent: the founder was the one not earning. The turn came in April 1997, when Naukri.com went live on an Indian server. In its first year it reportedly earned a little over Rs 2 lakh, not much by any standard. But employers were slowly learning that a website could reach more candidates than a Sunday classified. By 2005 it was reported as the country's largest job site. Jeevansathi had launched in 1998, 99acres in 2005, and the company listed in 2006. The second act was less obvious. Info Edge put about Rs 86 crore into FoodieBay, which became Zomato, between 2010 and 2013, and held a controlling stake until 2015. By July 2021 it held 15.23% of Zomato. It also backed Policybazaar before the site opened, by one account. The weakness is concentration: Naukri produced over 70% of revenue in 2018. Even so, FY25 revenue was Rs 2,654 crore with net income of Rs 977 crore, and Forbes put Bikhchandani and family at $4.25 billion in October 2024. For a first-time founder the quiet lesson is about who carries you. The Rs 800 rent and the spouse's salary are small numbers beside Rs 2,654 crore, yet they are what made the first years survivable. Companies of this size usually begin as someone else's patience. ## Wakefit (India) URL: https://startup.talk/india/wakefit/ Founders: Ankit Garg, Chaitanya Ramalingegowda Founded: 2016; Based in: Bengaluru, Karnataka; Sector: D2C mattresses and home furnishings The first hundred Wakefit mattresses were carried to customers by the two founders themselves. Ankit Garg is a chemical engineer from IIT Roorkee, by one profile, and it was a mattress, not a technology, that started it. He bought one in 2014 or 2015 (accounts differ) and was unhappy with both the price and how it suited his sleep. At the time he was working at Tapzo, where his colleague Chaitanya Ramalingegowda was also building. Over coffee, the two began talking about sleep as one of the three pillars of health, with food and exercise, and about how little attention Indians gave it. Garg did not jump. In 2015, before any company existed, he bought a few hundred mattresses and sold them on Amazon to see what the numbers looked like. What almost nobody remembers about the brand's origin is how small the test was. One report says it made a profit of Rs 60 lakh, though that figure comes from a promotional source. The starting capital is murky as well: Rs 13 lakh in one telling, Rs 18 lakh of savings in another. Wakefit was incorporated in March 2016. Garg was on his second startup and Ramalingegowda on his third, which may explain the caution. They sold memory-foam and orthopaedic mattresses online and, when the first orders came, delivered them personally to learn what was going wrong. Money followed in steps. A Series A of about Rs 65 crore from Sequoia in December 2018, Rs 185 crore in 2020, Rs 200 crore in November 2021. COVID pushed them into beds, sofas and wardrobes. In 2022 they opened their own stores and a large plant at Hosur. The cost showed up in the accounts. In FY23 revenue was Rs 813 crore but the net loss was Rs 146 crore. The turn arrived quietly: Rs 35 crore profit in FY25, then Rs 189 crore on Rs 1,489 crore revenue in FY26. The company listed on 15 December 2025 after a roughly Rs 1,289 crore issue, and the shares dropped 7% in May 2026 on margin worries from input inflation. For a founder, the useful detail is how cheaply they asked the market a question. A few hundred mattresses and a few weeks of deliveries answered what a survey might not have. The pitch was real long before the pitch deck was. ## Bira 91 (India) URL: https://startup.talk/india/bira-91/ Founders: Ankur Jain Founded: 2015 (beer imports from 2008); Based in: New Delhi; Sector: Craft beer / alcoholic beverages For five years, by one account, Ankur Jain's father did not speak to him. The silence began when the family learned that the son wanted to make beer. Jain was born in Delhi to an architect and an interior designer, and in 1998 he left for the US, where he took a computer engineering degree at the Illinois Institute of Technology in 2002. He worked at Motorola, then built a healthcare business of his own and exited after about four years. It looked like a safe, successful immigrant story. He came home to India in 2008 and started importing beer. That was the point where his father reportedly stopped talking. What almost nobody tells is how long it was before it worked. He brought in about 30 varieties not sold in India. Only in 2015 did Bira 91 launch, first brewed under contract in Belgium. He raised $1.5 million from friends, then $6 million in January 2016 from Sequoia, its first bet in alcohol. More followed: $50 million in 2018 and $30 million from Japan's Kirin in 2021. By FY23, revenue crossed $100 million. Then it broke. A shift from a private to a public company before an IPO required new excise approvals state by state, and production and sales stumbled. FY24 revenue was Rs 638 crore against a net loss of Rs 748 crore. Debt was reported to have ballooned toward Rs 1,000 crore. Production halted in July 2025, salaries slipped, and in October more than 250 employees wrote to the board asking for his removal, citing governance failures. On 22 July 2026 Jain stepped down and gave up his family's 17.8% stake in a settlement; the lenders withdrew their claims and his personal guarantees were released. He said the company needed fresh capital, a clear balance sheet and a management team. For a founder the lesson is unromantic: a brand built on one person's will can grow quickly, and former employees described just that, a business where decisions revolved around Jain. The same force that carries a dream past a father's objection can carry a balance sheet past what it can bear. ## Nazara Technologies (India) URL: https://startup.talk/india/nazara-technologies/ Founders: Nitish Mittersain Founded: 1999; Based in: Mumbai, Maharashtra; Sector: Gaming and sports media (mobile games, esports, Sportskeeda, PokerBaazi stake) Nitish Mittersain started a gaming company in Mumbai in 1999, a year before the dotcom bust, when he was reportedly still a student. He has said he loved games from a young age. Nazara began as an online gaming portal. In India in 1999, that meant an audience that barely existed, on connections that barely worked. Then the bubble burst. In 2002 the company moved into mobile value-added services, selling games and content to telecom operators. It was a retreat that turned into a business. What almost nobody saw coming was the long game. From 2015 Nazara licensed Chhota Bheem for mobile; in 2018 it took control of Nextwave and 55% of Nodwin Gaming; in 2019 it bought 67% of Sportskeeda for Rs 44 crore. SEBI cleared an IPO in 2018, but the company halted it. It listed on the NSE and BSE in March 2021. The acquisitions kept coming: OpenPlay for Rs 186 crore, WildWorks, Comic Con India, and in 2024 a 47.7% stake in PokerBaazi's parent for Rs 832 crore. FY24 revenue was about Rs 1,138 crore with net income of about Rs 89 crore. In August 2026 reports said Mittersain was stepping down as CEO, with Raymond Stauffer named successor. For a founder, Nazara is a lesson in what a pivot looks like when it is not a slogan. The first idea needed a world that had not yet arrived. Staying alive in a different business bought the years in which that world could show up. ## Mensa Brands (India) URL: https://startup.talk/india/mensa-brands/ Founders: Ananth Narayanan Founded: 2021; Based in: Bengaluru, Karnataka; Sector: A company that became a unicorn in six months spent the next four years learning what that speed had cost. Mensa Brands was founded in 2021 by Ananth Narayanan, who ran the aggregator model: buy small internet brands and grow them under one roof. Inc42 reports it crossed a $1 billion valuation within six months of incorporation, the fastest Indian startup to do so. The numbers that did turn up tell the other half. The FY23 loss more than doubled to Rs 227 crore. In May 2023 the company laid off fewer than 30 people at its media arm; a headline put the figure at 200, but the company's statement rejected that. In FY24 the loss narrowed to about Rs 156 crore. Then FY25. Revenue from operations fell 5.8% to $168 million, about Rs 1,394 crore, and the loss widened 39% to $60.4 million. During the year, the company sold India Lifestyle Network, which runs MensXP and iDiva, to RPSG Group for Rs 21.4 crore. One blog says Mensa had paid an estimated $60 million for it in December 2022, which would be an 85% drop, though that source is less reliable. What almost nobody noticed in the headlines was the narrowing. Mensa, now BRND.ME, is focused on four brands: Majestic Pure for essential oils, MyFitness for snacks, Botanic Hearth for hair care, and PartyPropz. Majestic Pure sells only overseas and brings in roughly Rs 400 crore, which Narayanan thinks could reach Rs 1,000 crore in three to four years. He said in February 2026 that the company is already cash-flow positive. The next step is an IPO after moving its domicile from Singapore back to India. For a founder, the useful point is that fast valuation is a loan, not a prize. The company's second life began only when it stopped collecting brands and started choosing them. ## Ninjacart (India) URL: https://startup.talk/india/ninjacart/ Founders: Thirukumaran Nagarajan, Kartheeswaran K K Founded: 2015; Based in: Bengaluru, Karnataka; Sector: Agritech / farm-to-retail fresh produce supply chain For a company that sells vegetables, Ninjacart's most revealing number is one it chose to shrink. Its operating revenue fell from about Rs 2,007 crore to Rs 1,634 crore in FY25, roughly 19 percent, and the company said so on purpose: it was walking away from low-margin, non-core business. That is a very particular kind of decision for a founder-led startup. Growth is the story investors want. Ninjacart told them the opposite, that the way to a profit was a smaller company. Ninjacart began in Bengaluru in 2015 as a farm-to-retail supply chain: buy produce from farmers, move it quickly, and sell it to the small shops and restaurants that cannot afford waste. Accel and Nandan Nilekani were among its early backers, and it later drew bigger money from names reported to include Flipkart and Walmart. The model is brutal. Fresh produce rots. Margins are thin. Every rupee saved on a middleman has to cover trucks, crates and spoiled tomatoes. Scale makes it better and scale also makes it worse. The reckoning came in FY25. The loss stayed near Rs 256 crore, almost flat against the year before, even as revenue dropped. Management says its core businesses are now profitable at the operating level and that the company as a whole should break even in 2026-27. In 2025 it raised a first slice of $6 million from Accel and Nilekani and said it was aiming for an IPO within two years. Not everyone is persuaded. Falling sales with a flat loss is not the picture investors usually want before a listing, and the claim of operating profit is the company's own. For a first-time founder, the lesson is in the retreat. A supply-chain business can become big by taking every order and still lose on each one. The harder, less glamorous skill is knowing which customers to stop serving, and saying it out loud while the world is watching the top line. ## DeHaat (India) URL: https://startup.talk/india/dehaat/ Founders: Shashank Kumar, Shyam Sundar Singh, Amrendra Singh, Adarsh Srivastava, Abhishek Dokania Founded: 2012; Based in: Patna, Bihar / Gurugram; Sector: Agritech / farm inputs and produce marketplace Shashank Kumar's family did not farm. That is the first thing worth knowing about the man who built one of India's largest agritech companies. He graduated from IIT Delhi in 2008, the kind of degree that points to a corner office, and spent about three years at a consulting firm. A Financial Express feature says his immediate family had no direct tie to farming, yet agriculture stayed with him as what the writer calls an abiding passion. What almost nobody notices is the gap between those two lives. Consultants advise. Farmers bear risk. Kumar chose to cross it. DeHaat began in 2012 with five founders: Kumar, Shyam Sundar Singh, Amrendra Singh, Adarsh Srivastava and Abhishek Dokania. The idea was to serve a smallholder farmer's whole season in one place, inputs, advice and a buyer for the harvest, through local village entrepreneurs. What the numbers show is the scale it grew to. Kumar says revenue reached about Rs 3,000 crore in FY25, up 11 percent from about Rs 2,700 crore the year before. The books are not simple. In FY24 DeHaat reported a net loss of Rs 1,133 crore, though Inc42 says most of it was a non-cash fair-value adjustment. The company says it made an EBITDA profit of Rs 5 to 10 crore in the first quarter of FY26 and is aiming for a profitable full year. The business is hard in its own way. Agriculture is sold to millions of people who buy small, trust slowly and can be wiped out by one bad monsoon. Thin margins and heavy working capital make profit a slow climb rather than a switch. The takeaway for a founder is how little the origin story has to match the market. Kumar was not a farmer's son. He was an outsider with a good degree and a nagging interest, and he worked at it for more than a decade. Nobody needs to be born into a problem to earn the right to solve it. They have to stay with it after the consulting job is gone. ## Snapdeal (India) URL: https://startup.talk/india/snapdeal/ Founders: Kunal Bahl, Rohit Kumar Bansal Founded: 2010; Based in: New Delhi, Delhi; Sector: E-commerce marketplace Kunal Bahl did not come home to build a startup. He came home because the United States said he could not stay. He had to leave the United States over a visa problem. He had studied at Wharton, after DPS R K Puram in Delhi, and had a job at Microsoft. A career abroad ended on someone else's rule. A man who was born in New Delhi found himself back in New Delhi with a plan to make and no company. He called Rohit Bansal, a school friend. That is the part the later billion-dollar story leaves out. Snapdeal's founding is told as a vision. It started more like a phone call. The pair launched Snapdeal on 4 February 2010 as a daily deals site, copying a format that was hot in the US. It did not last long. Within months Bahl and Bansal told their investors they would shut the deals business and become a marketplace. By October 2011 they had. The pivot worked. Nexus and Bessemer backed it, then eBay, then SoftBank, which put in $647 million in 2014 and became the biggest shareholder. A $500 million round followed in 2015. In 2016 the company was valued at $6.5 billion. Then the price arrived. SoftBank pushed for a merger with Flipkart, and talks collapsed in July 2017 over valuation and special payouts to early investors. Bahl chose to stay and shrink. Snapdeal sold Freecharge and Vulcan Express and focused on small-town buyers. Revenue grew 74 percent from 2017 to 2021 while losses dropped nearly 95 percent, but FY23 revenue was only Rs 388 crore, against an operating loss of Rs 282 crore, and a planned $152 million IPO was shelved in December 2022. Bahl now runs AceVector, the group that holds Snapdeal and other businesses, and invests through Titan Capital. The quiet lesson is in the order of events. The company that nearly reached $6.5 billion was born from a forced exit, a call to a school friend and a pivot made within months. A bad turn does not end the plan. It can be the reason the plan exists. ## BigBasket (India) URL: https://startup.talk/india/bigbasket/ Founders: Hari Menon, V S Sudhakar, Vipul Parekh, V S Ramesh, Abhinay Choudhari Founded: 2011; Based in: Bengaluru, Karnataka; Sector: Online grocery / quick commerce In its first year, Hari Menon's first online store did not make a single sale. He has said so himself. Fabmart started in 1999 as a general store on the internet, long before India had many people online. Menon told a podcast that it was simply too early. People were afraid to type their bank details into a website. Zero transactions, in the first year. The obvious move was to quit. The team, he said, decided to stay on and run it at a lower scale. Fabmart sold music first, then books, toys and jewellery, and early in 2000 started selling groceries in Mumbai, Pune and Hyderabad. None of that is in the usual BigBasket origin story, which jumps to 2011. But it is where the product came from. What almost nobody sees is how long the same small group of people stayed together. V S Sudhakar, Vipul Parekh, Hari Menon and V S Ramesh had worked together on the earlier ventures. A 2012 report notes that the team behind Fabmart reunited to launch an e-grocery business. They were joined by Abhinay Choudhari. The headline of a 2022 YourStory podcast says Menon started BigBasket at about 50, an age at which many founders are advised to stop. The company took shape in Bengaluru with an initial $10 million from Ascent Capital. By 2016 it was India's biggest e-grocer in eight cities. In 2019 it became a unicorn. In 2021 Tata Digital bought 64 percent at about a $1.85 billion valuation, and in January 2023 a $200 million round valued it at $3.2 billion. The price came later. In October 2020 a data breach reportedly exposed data of more than 20 million users. Quick-commerce rivals arrived with 10-minute promises, and BigBasket answered with its own pilot in 2025. In FY24 revenue was about Rs 10,100 crore with a net loss of Rs 1,415 crore. Menon has since stepped down as CEO, though he remains on the board with Parekh. A founder in a hurry should note how long the first idea took to find its market. The sale that never came in 1999 was not the end of the story. It was the first draft of a business that needed the world to catch up. ## Dunzo (India) URL: https://startup.talk/india/dunzo/ Founders: Kabeer Biswas, Ankur Agarwal, Dalvir Suri, Mukund Jha Founded: 2014; Based in: Bengaluru, Karnataka; Sector: Hyperlocal delivery / quick commerce Dunzo started as a WhatsApp group. That sentence appears in almost every account of Kabeer Biswas and his company. What appears less often is what he did before it. He is a computer science engineer who then took an MBA, and worked in sales at Airtel. He left, took a break, and started Hoppr with Mohammed Imthiaz, his former boss. Hike Messenger acquired Hoppr in 2014. So by the time he began Dunzo that same year, he had already built and sold one company. What can be said is that Dunzo grew from a very small place. Friends in Bengaluru would post a task in a group and someone would run it. Blume and others put $650,000 in during March 2016. In December 2017 Google invested $12 million, its first direct investment in an Indian startup. Then the money got bigger and the problems with it. A data breach in 2019 exposed about 3.5 million users. A Google Play rule on tobacco and liquor in 2021 forced a stripped-down version, Dunzo Mo. In January 2022 Reliance Retail led a $240 million round, putting in $200 million for 25.8 percent. It was the high point. FY23 revenue was Rs 227 crore against a net loss of Rs 1,802 crore, and bigger quick-commerce rivals were faster and richer. Layoffs followed in 2023 and 2024, then delayed salaries. Reliance reportedly wrote off its stake. In January 2025 the app and website went offline, and Biswas moved to Flipkart's quick-commerce unit. The cost is easy to see from outside: a company that began as friends running errands for each other ended owing months of pay to its employees. For a founder, the lesson is plain. Dunzo's first idea was good and cheap. Its collapse came after it became expensive. A business can prove demand and still lose to whoever pays the least for each delivery. ## Arm Holdings (United Kingdom) URL: https://startup.talk/uk/arm-holdings/ Founders: Robin Saxby, Hermann Hauser, Acorn Computers engineering team (incl. Steve Furber) Founded: 1990 (incorporated 12 Nov 1990 as Widelogic Ltd); Based in: Cambridge, England; Sector: Semiconductor IP / chip architecture licensing Before it was Arm, it was a name nobody would remember. On 12 November 1990 the company was registered as Widelogic Limited, a placeholder, and three weeks later it became Advanced RISC Machines. That quiet paperwork hid a small miracle: twelve engineers from Acorn Computers, a Cambridge company best known for school computers, were about to be handed to a stranger. The stranger was Robin Saxby, born in Chesterfield in 1947. As a teenager he ran a television repair business, and he later described himself as destined for the electronics industry. He studied electronics at Liverpool, graduating in 1968, then moved through Pye, Motorola and European Silicon Structures. He was the kind of engineer who understood both the soldering iron and the sales call. What the partners had was odd. Acorn supplied the people. VLSI Technology supplied the tools. Apple put in US$3 million because it wanted a low-power chip for what would become its Newton handheld. Larry Tesler, an Apple vice president, helped find a chief executive, and the search landed on Saxby, who joined in 1991. Hermann Hauser, the Vienna-born physicist who had co-founded Acorn and who arrived in England at 16 to learn English, had helped push the spin-out into existence. Then came the hard part: nobody was buying. Arm was tiny, and the idea that a British company with a dozen engineers could out-muscle chip giants was close to a joke. Saxby's answer, credited with the company's rise, was to stop competing on manufacturing. Arm would license the design to anyone, take a fee up front and a royalty on every chip. Others would bear the cost of factories. Steve Furber later recalled that even the name was a concession to Apple, which did not want a rival's brand in the title. It took until 1993 to make a profit. After that the model compounded. By the time Saxby's ten years as chief executive ended, market value had passed US$10 billion, and Arm had become the quiet standard in mobile phones. The price came later. SoftBank bought the company in 2016. A planned US$40 billion sale to Nvidia collapsed in February 2022 under regulatory opposition, and disputes over Arm China strained relations. Arm returned to the public markets in September 2023, raising US$4.87 billion at a US$54.5 billion valuation, with SoftBank keeping about 90 percent. For the year to March 2026 it reported revenue of US$4.92 billion. The story is a little unfair to the people. The architecture survives, the licence structure survives, even the Cambridge roots survive. But the first move was made by a former television-repair teenager who looked at a room of brilliant engineers and decided the product was not the chip but the permission to build it. For a first-time founder the lesson is awkward: sometimes the most valuable thing you can do is give your best work away, on terms that make you the foundation of everyone else's success. ## Revolut (United Kingdom) URL: https://startup.talk/uk/revolut/ Founders: Nik Storonsky, Vlad Yatsenko Founded: 2015 (Revolut Ltd incorporated Dec 2013; app launched July 2015); Based in: London, England; Sector: Fintech / digital banking The first cheque into Revolut came from the man running it. Nik Storonsky put in about £300,000 of his own savings, money he had earned as a trader, and he was the company's first investor. It was a bet with no safety net, placed by someone who had learned to take hits. He was born in Dolgoprudny, a science town outside Moscow, in 1984. His father was a physicist and engineer. Nik boxed and swam as a child and later became a state champion during university. He said that he started reading economics and business books at six. The hunger came early, and so did the habit of preparing for a fight. He studied physics at the Moscow Institute of Physics and Technology and economics at the New Economic School, both finishing in 2007. In 2006 he had already moved to England, to work as an equity derivatives trader at Lehman Brothers in London. Credit Suisse followed, and he stayed there until 2013. The spark was dull and practical. Travelling for work, he noticed how much banks and card companies took in fees and bad exchange rates. He wanted a card that charged less. In December 2013 he founded the company, found a software engineer named Vlad Yatsenko from Deutsche Bank to build it, and had a prototype in early 2015. The public launch came in July 2015. What almost nobody saw coming was how long the real fight would last. Revolut spread quickly, but regulators did not let it call itself a bank. It applied for a UK banking licence in 2021, got a restricted one only in July 2024, and a full one in March 2026. In between came a cyberattack in 2022 that exposed data of more than 50,000 customers, criticism over account freezes and fraud handling, and in 2026 European restrictions on some of its activity. The numbers kept growing regardless. In 2025 it reported revenue of £4.52 billion and net income of £1.3 billion. A secondary share sale in July 2026 valued it at US$115 billion, and by September it had over 80 million customers in 40 countries. Forbes put Storonsky's fortune at US$18.8 billion in 2026, and he renounced his Russian citizenship in 2022. The quiet part of the story is the savings. Not a grant, not a famous investor, but one trader's own money, enough to hire an engineer and find out if the idea held. For a first-time founder: the first investor who truly shows conviction is you, and the size of that cheque tells everyone else how to read your pitch. ## Fluidstack (United Kingdom) URL: https://startup.talk/uk/fluidstack/ Founders: Jamie Cox, Gary Wu, Cesar Maklary Founded: 2017; Based in: London, England; Sector: AI cloud / GPU data centre infrastructure The co-founder who built one of Europe's most valuable AI infrastructure companies was studying classics. Jamie Cox, then at Oxford, was reading the ancient world while his friend Gary Wu studied economics down the corridor. In 2017 they started a company with an idea that sounded almost too small: gamers' graphics cards sat idle, AI researchers could not find enough compute, so why not connect them? That was the early product, a marketplace that rented out unused GPUs. It was scrappy, and so were the numbers. Revenue in 2022 was US$1.8 million. Cox, according to a VnExpress profile, left Oxford the year after the company was founded to take up a Thiel Fellowship, the programme that pays young people to skip college. He was a dropout in a field that does not ordinarily lead to server racks. What almost nobody saw was the pivot. During the Covid-19 pandemic, the marketplace changed shape. Instead of stitching together hobbyist GPUs, Fluidstack started renting larger pools from companies and university labs. In 2020 Cesar Maklary joined, a man who had worked in aerodynamics for a Formula One team. His background was in engineering at speed, and the company needed that. The next four years were steady, not spectacular. Revenue reached US$30 million in 2023 and US$66.2 million in 2024. Then came the AI boom, and the customer list changed. Fluidstack worked with Meta, Mistral, Poolside and Black Forest Labs. In November 2025, Anthropic announced a US$50 billion US computing investment and chose Fluidstack to build custom data centres in New York and Texas. It also appeared to be the first publicly known operator of TPU data centres outside Google, as part of Anthropic's planned deployment of up to one million Google TPUs. Not everything held. The company withdrew from a planned €10 billion data centre project in France as it shifted its focus to the United States, and its headquarters now sits in New York. Valuation rose from US$7.5 billion earlier in 2026 to above US$18 billion after a US$1.5 billion round led by Jane Street, with projected 2026 revenue of US$660 million. The founders, aged around 29 to 31, are reported to have crossed into billionaire territory. But the quiet detail is the humblest one: a company that now builds data centres for frontier labs began as two students noticing a mismatch between a teenager's gaming PC and a researcher's waiting list. The lesson for first-time founders: a small, almost silly observation about idle resources can hold a very large market, if you are willing to change the product three times to find it. ## Nscale (United Kingdom) URL: https://startup.talk/uk/nscale/ Founders: Josh Payne, Nathan Townsend Founded: 2024; Based in: London, England; Sector: AI cloud / GPU data centres Josh Payne's career started underground. Before the data centres, the Microsoft contracts and the US$14.6 billion valuation, he left high school in an industrial part of Australia and went to work in coal mines. Profiles describe it as his early career, though they do not say for how long or where. It is a strange place to start for a man who now runs one of Britain's most valuable AI infrastructure companies. The skill he carried out of the mines was not code. It was a feel for power, land and the heavy things that make production possible. Years later he moved to London, though sources do not say when or why. His first company was Arkon Energy, which he founded in 2020 and ran as executive chairman until late 2024. It supplied infrastructure for renewable-powered bitcoin mining and, more usefully, secured power and sites. That was the hidden asset. When the AI boom turned energy into the scarcest input in computing, Arkon's contracts looked like a map of where the electricity was. The obvious pivot was not entirely comfortable. Mining and AI are different customers with different tolerances, and Arkon eventually wound down. Nscale was launched in 2024, with Nathan Townsend named by Computer Weekly as a co-founder, and took over a Norwegian site from Arkon. Then the money arrived at an unusual speed. A US$155 million Series A in December 2024. A US$1.1 billion Series B in September 2025 from Aker, NVIDIA, Nokia and Dell. A US$433 million SAFE. And a US$2 billion Series C in March 2026, bringing Sheryl Sandberg, Susan Decker and Nick Clegg onto the board. Microsoft became the anchor customer, with deals covering over 200,000 NVIDIA chips and sites in Norway, Portugal, the UK and the US. The price has been scrutiny. Some press questioned its credibility and how British it really is. Revenue is real but small against the promises: US$19.1 million in 2024, US$33.0 million in 2025 and US$140.6 million in the first half of 2026, set against what its filings describe as US$103.4 billion in contracted take-or-pay agreements. The memorable detail is one nobody planned. The company shares its name with a model-railway scale, a hobby of tiny tracks, while building campuses meant to reach 2 gigawatts. The lesson for a first-time founder is that your unglamorous first career may be your edge. Payne knew what physical infrastructure costs and where the power is, which is often exactly what the glamorous software crowd cannot learn quickly. ## Isomorphic Labs (United Kingdom) URL: https://startup.talk/uk/isomorphic-labs/ Founders: Demis Hassabis Founded: 2021 (incorporated Feb 2021; announced Nov 2021); Based in: London, England (also Lausanne, Cambridge MA); Sector: AI drug discovery In 1984, an eight-year-old boy walked into a shop with chess winnings and came out with a ZX Spectrum 48K. Demis Hassabis had bought his first computer with prize money, and he taught himself to program it. The house he came from was not rich. His father, Costas Hassapis, was a Greek Cypriot and an aspiring singer-songwriter who sold toys from a van. His mother, Chinese Singaporean, had been raised as an orphan, trained as a nurse and worked as a retail clerk and part-time cleaner. The family name was Hassapis, which Demis later changed by one letter. Chess was the escape hatch, and he was good: master standard at 13, with an Elo rating of 2300. At 17 he co-designed and lead-programmed Theme Park at Bullfrog Productions with Peter Molyneux. Success arrived early, and so did an offer. He turned down a seven-figure sum to stay in the games industry, and went to Queens' College, Cambridge, graduating in 1997 with a double first. What almost nobody remembers is what came next. In 1998 he left Lionhead to found Elixir Studios. It made Republic and Evil Genius, but in April 2005 the studio closed and its technology rights were sold. It was a public failure for someone who had been a prodigy. He went back to school, earned a PhD in cognitive neuroscience at UCL in 2009 under Eleanor Maguire, and began to build the bridge between brains and machines that would become DeepMind. The turn was AlphaFold. It showed that a machine could predict how proteins fold, a problem that had stalled biologists for decades. Isomorphic Labs was incorporated in February 2021 and announced in November 2021 as a London drug-discovery company spun out of DeepMind inside Alphabet. In January 2024 it signed deals with Novartis and Eli Lilly. In October 2024 Hassabis shared the Nobel Prize in Chemistry with John Jumper. The money followed in two steps: US$600 million in April 2025, its first external round led by Thrive Capital, and US$2.1 billion in May 2026 with Alphabet, GV and the UK Sovereign AI Fund joining. By February 2026 it said its own drug design engine had beaten AlphaFold 3 on a protein-ligand benchmark. The real test is still ahead: a drug that works in people. Software scales quickly, but medicine does not. For a first-time founder, the odd lesson is that the closed studio mattered. Elixir's failure sent Hassabis to neuroscience, and that detour became the thing that nobody else had. ## Checkout.com (United Kingdom) URL: https://startup.talk/uk/checkout-com/ Founders: Guillaume Pousaz Founded: 2012 (rebrand of Opus Payments, founded 2009 in Singapore); Based in: London, England; Sector: Fintech / payments (merchant acquiring) In 2005 Guillaume Pousaz walked out of his economics degree in Lausanne. His father had just been diagnosed with cancer, and the plan he had been following suddenly looked like somebody else's. That plan was neat. Born in 1981 on the edge of Geneva, he had studied mathematical engineering at EPFL, one of Europe's hardest technical schools, and then started an economics degree at HEC Lausanne. He wanted to be an investment banker, and an offer from Citibank in London was there for the taking. He did not take the straight line. After leaving, he moved to California. The reason, as reported, was surfing. The Times later labelled him a quiet surfer. Nobody who has watched his company since would call it idle, but the detour tells you something. A man who walks away from a bank offer because life has just shifted has already decided that security is not the main thing. What almost nobody remembers is how slowly it started. In 2006 he joined a payments firm, International Payments Consultants. He then co-founded NetMerchant, a service moving money from the US to Europe, which ran until 2009. That year he bought SMS Pay in Mauritius and set up Opus Payments in Singapore. It was small, and it was not glamorous. In 2012 it became Checkout.com. Then came the unfashionable part. Checkout joined Visa and Mastercard in 2013 and became American Express's global acquirer in 2015. In 2017 it picked up a UK e-money licence. For years there was no big funding announcement. The first large outside round, $230 million, arrived only in 2019, and by January 2022 a $1 billion raise valued it at $40 billion, making it Europe's most valuable fintech startup at that moment. The fall was just as public. By December 2022 its own internal valuation had been cut to $11 billion, and $9.35 billion the next year. Pousaz, now in London with a wife and three children, had to run a company whose headline value had been cut by roughly three quarters. He did it by getting profitable. In 2024 Checkout reported profit, with core revenue growing 45 percent, and a September 2025 buyback of employee shares put it at $12 billion. The quiet part of the story may be the useful one. The boy who left school in grief and went to the beach built a company that never needed the world to look at it. For a founder, the lesson is plain: a detour taken for a real reason is not lost time. Pousaz came back to payments with a long view, years of small jobs behind him, and no need to impress anyone in the meantime. ## ElevenLabs (United Kingdom) URL: https://startup.talk/uk/elevenlabs/ Founders: Mati Staniszewski, Piotr Dabkowski Founded: 2022; Based in: London, England (founders Polish); Sector: AI / voice synthesis and conversational audio At Imperial College London, a maths student started a conference for people who liked maths for no reason except that it was fun. It was called Mathscon. Mati Staniszewski organised it before he had a company, a salary, or any idea that a movie dub would one day change his life. He was born in 1995 in a town outside Warsaw, and moved to the capital for high school at Copernicus Bilingual. There he met Piotr Dabkowski. They were friends first, and stayed friends. That detail matters, because most start-up partnerships are made over a cap table. This one was made over lunch breaks. The two went separate ways. Staniszewski studied mathematics in London, then worked at Opera Software and at BlackRock, where he joined the Portfolio Analytics Group and helped launch the Aladdin Wealth platform. After that he spent time at Palantir as a deployment strategist, running big implementations for governments and companies. Dabkowski became a machine learning engineer at Google. What bound them back together was a small, almost silly annoyance. They had grown up with American films dubbed into Polish by a single flat voice reading every role. Watching poorly dubbed films, as the story is told, planted the idea that voice technology could carry emotion across languages. That was not a grand thesis. It was an irritation they could not leave alone. They named the company after 11 November, Poland's Independence Day, and added Labs to signal research. In January 2023 they raised a $2 million pre-seed round led by Credo Ventures. By June a $19 million Series A, co-led by Andreessen Horowitz, put the value near $100 million. A public beta had already crossed a million registered users. The growth came with a cost. In 2024 fake robocalls imitating Joe Biden were traced to tools built on ElevenLabs. Staniszewski said the company was dedicated to preventing misuse. The same models that comfort a person who lost their voice can fool a voter, and a founder has to own both. The numbers went up quickly: $1.1 billion in January 2024, $3.3 billion a year later, $11 billion by February 2026. In March 2026 the company pledged $1 billion towards voice restoration for a million people with permanent voice loss. The quiet insight is in the first scene. Mathscon was never a business, and nobody paid for it. But a founder who spends student years building something fun for strangers is practising a skill: gathering people around an idea before there is money in it. ## SumUp (United Kingdom) URL: https://startup.talk/uk/sumup/ Founders: Marc-Alexander Christ, Daniel Klein, Jan Deepen, Stefan Jeschonnek, Petter Made Founded: 2012 (conceived 2011); Based in: London, England; Sector: Fintech / card readers and small business payments For a few years, SumUp was in a race it could have lost. On the other side stood Payleven, a rival with the same idea, the same little card reader, and the money of Rocket Internet behind it. Marc-Alexander Christ, one of SumUp's co-founders, later put it plainly in a German interview: the merger with Payleven got the company out of its fight for survival. That is not the sentence founders usually say in public, and it is the best clue to the shape of the story. SumUp was conceived in 2011 and founded in 2012, with a first product in August. Five people stood behind it: Christ, Daniel Klein, Stefan Jeschonnek, Jan Deepen and Petter Made. Klein came with an earlier scar and an earlier win, having helped found Moneybookers, later renamed Skrill. Before any Series A, the founders funded the company themselves. The five did not stay five. Made left the operating business at the end of 2013, and Deepen and Jeschonnek followed in the summer of 2014. All remained shareholders. Founding partnerships tend to thin out before the real money arrives, and this one was no exception. The turn came in April 2016. SumUp and Payleven merged. It was called a merger, but the SumUp brand won, Klein became chief executive of the combined business, and all Payleven merchants were migrated onto SumUp's technology. Payleven's co-founder Konstantin Wolff joined as marketing chief. The race was over. After that, SumUp could grow without looking over its shoulder. It bought the e-commerce platform Shoplo in 2019. In 2020 and 2021 it bought Goodtill, Tiller and Paysolut, and paid $317 million for the loyalty startup Fivestars. Not everything worked. A Chile venture with Banco del Estado was taken over wholly by the bank by September 2020. By 2022 it had about 3,000 staff in 23 locations and more than 3.5 million merchants across over 30 markets. A June 2022 round of EUR590 million valued it at around EUR8 billion, and in May 2024 Goldman Sachs led a EUR1.5 billion raise. What can a first-time founder take from this? Sometimes the strongest move in a race is to stop racing the person who looks like you. SumUp's founders, who had started with their own money, were ready to share the stage with a better-funded rival and let the best product win. Few founders would admit, as Christ did, that the deal had been survival. ## Wise (United Kingdom) URL: https://startup.talk/uk/wise/ Founders: Kristo Kaarmann, Taavet Hinrikus Founded: 2011; Based in: London, England; Sector: Fintech / cross-border payments In June 2022 the Financial Conduct Authority's published list of penalised people included the chief executive of a company that had spent a decade telling customers to stop paying hidden fees. Kristo Kaarmann had reportedly failed to pay 720,000 pounds of tax for the 2017-18 year. That is not where the story starts, but it is the part few tell. Kaarmann had been a management consultant at Deloitte. Taavet Hinrikus had been among the first employees at Skype. Both were Estonians living in London, and both had the same irritation. Hinrikus was paid in euros by Skype in Estonia and needed pounds in London. Banks charged at least 5 percent to move money, and most of that was buried in the exchange rate. According to the account on Wise's own history, the banks tried hard to make that hard to understand. The two worked out a trick between themselves: one sent euros to the other's Estonian account, the other sent pounds into Hinrikus's British one, both at the real rate. No bank margin. No fee worth the name. Wise was founded in 2011. A 1.3 million dollar seed round was followed in 2013 by 6 million dollars led by Valar Ventures. A 280 million dollar round in 2017 pushed it toward a 3.5 billion dollar valuation by 2019 and 5 billion in 2020. The cost of the company's honesty brand came into view in 2021. In July Wise went public through a direct listing in London, valued at 11 billion dollars. The same year Hinrikus stepped down as chair, replaced by David Wells. And the tax default surfaced: the FCA's list ran for 12 months from September 2021, and the story became public in June 2022. A founder who built a company on transparency now had a very public flaw of his own. The business kept growing anyway. In the year to March 2026 Wise reported net revenue of 2.5 billion dollars, up 19 percent, and income before tax of about 660 million. Primary trading moved to Nasdaq in May 2026, with London kept as a secondary listing. Kaarmann held a reported 1.134 billion pounds on the 2023 Sunday Times Rich List, Hinrikus 861 million. What should a first-time founder take from this? The most durable companies are built on a grievance everyone shares. But a founder who sells trust gets judged by a stricter standard, and the personal record counts as part of the product. ## Octopus Energy (United Kingdom) URL: https://startup.talk/uk/octopus-energy/ Founders: Greg Jackson Founded: 2015 (incorporated 2014); Based in: London, England; Sector: Energy retail / climate tech Greg Jackson has told the BBC that he remembers the power being cut off at home when he was a child. He runs one of Britain's biggest household energy suppliers now. The fact sits oddly beside the scale, and it explains a great deal. He was born in September 1971, listed as Hanover in the infobox, and grew up in Halifax, West Yorkshire, then Saltburn-by-the-Sea. In the early 1980s he got into electronics and coding. At 16 he left school to program computer games. It was not a rebellion. It was work, and he liked it. Then he went back. He did sixth form college, studied economics, and won a place at Pembroke College, Cambridge. It is the kind of path that makes admissions officers tell nice stories, but Jackson's was messier than the usual: a school-leaver programmer who returned to the classroom on his own terms. By 30 he had owned or managed a number of small firms employing between four and 60 people. He founded e-commerce company C360 and telemedicine provider Consultant Connect, and sat on boards including peer-to-peer lender Zopa. Then he joined Octopus Group's idea of a new energy company. Octopus Energy was incorporated in October 2014 under the name Mercury Energy Supply, renamed in 2015, and began supplying customers that December. What almost nobody notices is how deliberately it was built as software. The customer platform became Kraken, later licensed to other suppliers. In 2019 a campaign called Portraits from the Precipice appeared on more than 5,000 sites. In 2020 Origin Energy paid A$507 million for 20 percent, making Octopus a unicorn; Generation Investment Management came in at a 4.6 billion dollar valuation in 2021. The real test arrived in 2022. When rival Bulb collapsed, Octopus took on 1.5 million of its customers. That was a rescue and a risk. By the end of 2024 it had overtaken British Gas as Britain's largest household supplier, though a net loss of 255 million pounds followed in 2025 on revenue of 13.7 billion. In December 2025 Octopus said it would spin off Kraken at an 8.65 billion dollar valuation. The quiet insight: a founder who has sat in a house where the lights went off treats the bill as a human problem, not a data one. Jackson built a company around the grudge, and then built the machine to serve it. ## Kraken (United Kingdom) URL: https://startup.talk/uk/kraken/ Founders: Greg Jackson, Amir Orad Founded: 2016 (Kraken platform built inside Octopus Energy); Based in: London, England; Sector: Energy software / utility customer-management platform Greg Jackson has told the BBC that he remembers the day the energy went off in his childhood home. Not a storm, not a fault. The supply was cut because it had not been paid for. He grew up in Halifax in West Yorkshire and then in Saltburn-by-the-Sea on the Yorkshire coast. He was drawn to electronics and coding in the early 1980s, and at 16 he left school to program computer games. That did not last. He went back to sixth-form college to study economics and ended up at Pembroke College, Cambridge. What almost nobody connects to the software company is that childhood memory. Years later Jackson founded an energy retailer, Octopus, and the thing he is known for wanting is a bill that does not frighten people. But a retailer cannot be kind at scale if its systems are clumsy. Before turning 30 he had owned or run several small businesses with anywhere from 4 to 60 employees, including an e-commerce firm called C360 and a telemedicine provider, Consultant Connect. He knew what it was to carry payroll on a thin margin. So Octopus built its own customer platform and called it Kraken. It ran on Amazon's cloud and held accounts, bills, meters and tariffs in one place. In an industry where the old suppliers ran on stitched-together legacy systems, this was unusual. And then came the turn nobody planned: other utilities noticed. Hanwha's Australian retail arm chose the platform in 2018. Good Energy licensed it in late 2019. In 2020 E.ON, npower and Origin Energy signed up. EDF agreed in 2021 to move about five million customers across. The cost of success was a different kind of company. A retailer and a software vendor want different things, and in August 2024 Amir Orad was appointed as Kraken's own chief executive. By January 2025 the platform was reported to serve 60 million customer accounts. Then came the separation. In December 2025 Octopus said it would spin Kraken off at a valuation of about US$8.65 billion, with Octopus keeping a 14 percent stake. Roughly a US$1 billion minority stake changed hands. It is worth sitting with the oddity of that. A billion-dollar software company grew out of a man's refusal to treat a bill as a mere transaction. Jackson's own early failures and small firms taught him how fragile a customer relationship is when money is tight. For a first-time founder the lesson is quiet. The product that wins is often the one built to solve the thing that hurt you personally, and you only discover it is a company when strangers ask to borrow it. ## Wayve (United Kingdom) URL: https://startup.talk/uk/wayve/ Founders: Alex Kendall, Amar Shah Founded: 2017 (21 Aug 2017); Based in: London, England (founded in Cambridge); Sector: Autonomous driving / embodied AI The first boardroom was a bedroom. Alex Kendall has said the larger bedroom of a rented house near Cambridge was where he and Amar Shah held meetings, and the small one held the servers. The car itself was being rebuilt in the garage. Kendall grew up on New Zealand's mountainous South Island. As a student he built one of the country's first homemade drones from parts he found at the local electronics shop. Nothing about that suggests a billion-dollar company. It suggests a boy who liked to make flying things from whatever was on the shelf. He went to Cambridge and finished a PhD in deep learning and computer vision in 2017. Shah, also at Cambridge, had a PhD in machine learning. In August 2017 the two founded Wayve with an idea that most of the self-driving industry thought was wrong. Others were building detailed 3D maps and thousands of rules. Wayve wanted a car to learn to drive the way a person does, from cameras and experience. What almost nobody saw was how lonely that bet was. For about five years the approach was met with scepticism and the company was largely overlooked, Kendall has said, though he now sees the challenge as a motivator. They had raised roughly US$3 million to start. They fitted an electric Jaguar I-PACE with the software in under six months. The first drive was around the block. The turn came in 2019. Wayve reported the first successful test of an AI car driving a route it had never seen, using end-to-end deep learning and a basic GPS map. Eclipse led a US$20 million round that year. Microsoft, Virgin and others followed in 2022 with more than US$200 million. The price was personal as well as financial. In 2020 Shah, the first CEO, left, and Kendall moved from chief technology officer to chief executive. The sources do not say why, and nothing here should be read into it. What is documented is that the founder who had been the quiet engineer became the person who had to sell the company to carmakers. It worked. In May 2024 SoftBank led a US$1.05 billion round with Nvidia and Microsoft. In February 2026 a round of about US$1.5 billion from Mercedes-Benz, Stellantis, Nissan and Uber valued the company at US$8.6 billion, and Uber robotaxis launched in London that September. For a founder, the picture is simple and slightly funny. The company everyone now wants started in a rented house with servers in a small bedroom. Being early and unfashionable is survivable if the idea is real and the room is cheap. ## Just Eat (United Kingdom) URL: https://startup.talk/uk/just-eat/ Founders: Jesper Buch, Klaus Nyengaard, Christian Schnack-Petersen, David Buttress Founded: 2001 (launched Aug 2001 in Kolding); Based in: London, England (founded in Kolding, Denmark); Sector: Online food delivery marketplace In 2005 a technology entrepreneur named Bo Bendtsen bought out almost everyone who had started Just Eat. Not quite everyone. One founder, Jesper Buch, stayed. That is the part of the story that rarely makes the headlines. Just Eat began in Kolding, a Danish town, launching in August 2001 with 15 employees. The concept had been developed in parallel by two groups, who then chose to merge and work together. The early years were small and quiet. Then Bendtsen bought the founders and early investors out and moved the business to the United Kingdom. Buch kept his place, and he later pulled Klaus Nyengaard to London. In 2006 the company hired a Welsh executive, David Buttress, as chief executive. Why Britain? The sources do not give the founders' reasoning. What is documented is the model: restaurants paid a joining fee of GBP699 and a 13 to 14 percent commission on each order, and more than 90 percent of revenue came from those commissions. It was a toll on a habit Britain already had. The money followed. In 2009 Index Ventures and Venrex invested GBP10.5 million. Greylock and Redpoint put in GBP30 million in 2011, and Vitruvian led GBP40 million in 2012. On 3 April 2014 the company listed in London. By 2018 it had 3,600 employees and 26.3 million customers, and a 2017 Guardian report put its value at about GBP5.5 billion. The cost showed in the scrutiny. In 2017 an audit found 35 unlicensed restaurants with no hygiene rating on the platform, and a 2018 BBC investigation found takeaways rated zero listed on the app. In 2023 it cut 1,700 UK delivery jobs and moved drivers to freelance status. The merger with Takeaway.com, announced in 2019 at about GBP9 billion, ended Just Eat's independence. Buttress left in 2017 to run the Dragons rugby club. Danish operations had stopped by March 2026. There is an odd symmetry. A company born in a small Danish town was rebuilt in London by a buyer who removed nearly all its founders, and the one who remained is the name people still link to it. For a founder, the lesson is about ownership: being the one who is still in the room when the plan changes may matter more than having been there first. ## Fractile (United Kingdom) URL: https://startup.talk/uk/fractile/ Founders: Walter Goodwin Founded: 2022; Based in: London, England; Sector: AI inference semiconductors The speed of it is the strange part. A company founded by a robotics PhD student in 2022 was valued at about US$1 billion after a US$220 million round, and then, within about three months, reported to be raising at US$6.5 billion. Walter Goodwin was the student. He worked at the University of Oxford's Robotics Institute when he started Fractile, a chip company with a deliberately narrow target. He was not trying to out-train the biggest models. His bet was about what happens afterwards, when people actually use them. What almost nobody outside the chip world noticed was a plain observation. Running a trained AI model, known as inference, is limited by how fast data moves between memory and compute. Fractile's design puts compute and memory on the same die using SRAM, rather than fetching data from separate memory. It is a bet that the bottleneck is not arithmetic but waiting. That is also a risky bet for a young founder. Chips take years and large sums to prove. Here is the money: a small early round of roughly US$15 million, then a US$220 million Series B from Accel, Factorial Funds and Founders Fund. The turn came in 2026 with a US$250 million chip deal to supply inference hardware to Anthropic. Reports in August said Fractile was in advanced talks to raise at US$6.5 billion. The price is the part founders should read carefully. The chips do not ship until 2027, and the contract is still in its initial phase. A valuation this high is a promise, not a result. Goodwin now has to make silicon that does what the slides said. The lesson is not that a student can become a billionaire-adjacent founder in a season. It is that a narrow, technical claim about where the real bottleneck sits can be worth more than a broad one, if one customer believes it enough to sign. After that, the hard part starts. ## Blockchain.com (United Kingdom) URL: https://startup.talk/uk/blockchain-com/ Founders: Ben Reeves, Peter Smith, Nicolas Cary Founded: 2011 (August, York); Based in: London, England (founded in York); Sector: Crypto wallet, exchange and brokerage In early 2012 two young men applied to Y Combinator together. One was Ben Reeves. The other was Brian Armstrong, who would go on to found Coinbase. They did not finish the application as partners. The disagreement was about a single design question: should the users control their own wallets, or the platform? Reeves held to the first view and carried on alone. That quarrel is the origin of one of the biggest rival pairings in crypto, and few people retell it. Reeves had started Blockchain.info in August 2011 in York, England, with a website that let anyone watch bitcoin transactions, a block explorer. The sources do not give his birthplace or university. What is documented is the setting. Reeves, Nicolas Cary and later Peter Smith worked from his flat in York, and Roger Ver gave the first funding. The early years were as fragile as the currency. In February 2014 Apple removed the app from its App Store, which caused backlash in the bitcoin community. It was reinstated in July 2014. By then Peter Smith had joined as chief executive, and in October 2014 the company raised US$30.5 million, then the largest funding round in digital currency. The company grew slowly through a long winter. A US$40 million round in June 2017 valued it at US$280 million. Then crypto surged. In March 2021 a US$300 million round, with US$100 million from Baillie Gifford, valued it at US$5.2 billion. The cost of that speed arrived in 2022. Three Arrows Capital's insolvency left about US$270 million in loans in default, and in July the company laid off roughly a quarter of its staff, about 150 people. In November 2023 a US$110 million Series E, led by Kingsway Capital, kept it going. Today about 60 percent of revenue comes from brokerage and market services and about 40 percent from retail users, and it was valued at US$7 billion in November 2026. The quiet lesson in the York flat and the failed application is that a principle can split a partnership before the company even exists. Reeves chose self-custody over a partner, and a decade later the company he kept building still carries that choice. ## Monzo (United Kingdom) URL: https://startup.talk/uk/monzo/ Founders: Tom Blomfield, Jonas Huckestein, Paul Rippon, Gary Dolman, Jason Bates Founded: 2015 (as Mondo; registered as Monzo Aug 2016); Based in: London, England; Sector: Digital bank / fintech Mondo is not a name you will find on a bank card today. A trademark challenge forced the young company to rename itself in public, through a naming contest, and by August 2016 it was registered as Monzo. It was a small embarrassment, and the start of a habit: Monzo did its thinking out loud. The five people behind it, Tom Blomfield, Jonas Huckestein, Paul Rippon, Gary Dolman and Jason Bates, had met while working at Starling Bank. Their itch was simple. Banks in Britain were slow, opaque and hard to leave. Blomfield became chief executive in 2015, and the team set out to build something that felt like an app first and a bank second. Money was the first test. In 2016 they went to the crowd. A £1 million round on Crowdcube sold out in 96 seconds, a record. Customers were not just users, they were shareholders. In 2017 came £71 million at around £280 million, and in 2019 a £113 million round valued it near £2 billion. What almost nobody remembers is how the name itself was settled. Monzo was picked after outsiders had a say, and that open instinct carried into the product: neon hot coral cards that began as a prototype choice and became deliberate branding, and instant notifications every time you spent. Then came the price. In 2019 Monzo disclosed that 480,000 customer PINs had been logged internally. The BBC's Watchdog looked into account freezes and closures. In 2020 a £60 million down round cut its valuation to about £1.24 billion. Blomfield moved from chief executive to president and left in January 2021. Profit finally arrived: £15.4 million for the year to March 2024. In 2025 the FCA fined it £21 million over weak financial crime controls, with regulators finding implausible addresses such as Buckingham Palace and 10 Downing Street. The company left the US market in March 2026 and Diana Layfield took over as chief executive that February. Its latest figures show revenue of £1.71 billion, over 15 million customers and deposits above £25 billion. For a first-time founder, the lesson sits in the order of events. Monzo grew faster than its controls, and the bill came with interest. Moving quickly won customers. Boring systems kept them. ## King.com (United Kingdom) URL: https://startup.talk/uk/king-com/ Founders: Riccardo Zacconi, Sebastian Knutsson, Thomas Hartwig, Melvyn Morris, Toby Rowland, Lars Markgren, Patrik Stymne Founded: 2003 (August, Stockholm, as Midasplayer.com); Based in: London, England (founded Stockholm); Sector: Mobile and casual games It was Christmas Eve 2003. King was four months old and almost out of money. Then Melvyn Morris, an angel investor who would later become chairman, put cash in. The company that would one day own Candy Crush lived on because of a single holiday cheque. The founders were not newcomers. Riccardo Zacconi and Toby Rowland had worked on the dating site uDate.com, and several others had been at Spray, a web portal that failed. Midasplayer.com, as it was first called, was built by people who had already seen one dot-com bet collapse and wanted to try again, in Stockholm, with Zacconi, Sebastian Knutsson, Thomas Hartwig, Lars Markgren and Patrik Stymne beside him. The company turned a profit by 2005, the year it renamed itself King.com and took $43 million from Apax Partners and Index Ventures. For years it was a quiet portal of small browser games. What almost nobody outside gaming noticed was how patient the company was. Then in April 2012 it launched Candy Crush Saga, taking the saga format from Bubble Witch Saga. Revenue jumped from $62 million in 2011 to $1.88 billion in 2013, and passed $2.6 billion in 2014. There was a price. When King listed on the New York Stock Exchange on 26 March 2014 at $22.50 a share, valuing it at about $7.08 billion, shares fell 15.6% on the first day. Candy Crush mascots went to the exchange to celebrate, anyway. The bigger problem was dependence: Candy Crush made up nearly half of revenue, trademark disputes over Candy and Saga had to be settled, and Facebook's games market was shifting under it. Activision Blizzard agreed to buy King in November 2015, closing the $5.9 billion deal in February 2016. Microsoft bought Activision in 2023, and in 2025 cut ten percent of King's workforce. The lesson for a first-time founder is how thin survival can look. A company later worth billions came within one Christmas Eve of not existing, and what kept it alive was a person who believed in it before the product did. ## Mimecast (United Kingdom) URL: https://startup.talk/uk/mimecast/ Founders: Peter Bauer, Neil Murray Founded: 2003; Based in: London, England (North America: Lexington, Massachusetts); Sector: Cloud email security and archiving / cybersecurity The ticker on the screen was MIME. Not by accident: in June 2010 Mimecast had hired Dr Nathaniel Borenstein, co-creator of the MIME email standard itself, as its chief scientist. A company that lived inside email had put the inventor of one of its foundations on staff, and five years later it listed on the Nasdaq under his creation's name. Peter Bauer and Neil Murray started Mimecast in 2003, Bauer as chief executive and Murray as chief technology officer. What mattered was the bet. Companies were drowning in email, running their own servers, losing mail and facing legal requests to produce it. Their answer was to move the work to the cloud: management, archiving and protection, sold as a service for Google Workspace, Microsoft Exchange and Office 365. London became the global base, with Lexington, Massachusetts for North America. What almost nobody saw coming from outside was how the email layer would turn into a security layer. As attacks moved through the inbox, a firm already sitting in the mail path was well placed to filter it. It bought Ataata and Solebit in 2018. Then the cost of that position showed. In January 2021 one of its security certificates was compromised, possibly letting attackers intercept communications with Microsoft email servers. Some investigators suspected the group behind the SolarWinds breach. In May 2022 Permira acquired the company in a deal reported at about $5.8 billion, ending its life as a public company. In January 2024 Marc van Zadelhoff replaced Bauer as chief executive, and Mimecast later bought Code42. For a founder, the quiet lesson is positioning. Mimecast did not invent email; it stood at the point where every message had to pass, and it let that spot grow in value. The same spot also made it a target. Being essential is both a business model and a risk. ## Darktrace (United Kingdom) URL: https://startup.talk/uk/darktrace/ Founders: Poppy Gustafsson, Nicole Eagan, Mike Lynch (Invoke Capital), Jack Stockdale, Stephen Huxter, Emily Orton Founded: 2013; Based in: Cambridge, England; Sector: AI cybersecurity The company that later promised to spot attackers hiding in quiet networks was itself born inside someone else's company. Darktrace was founded in 2013 at Invoke Capital, owned by Mike Lynch, and many of its early managers came straight from Autonomy, Lynch's software firm: Poppy Gustafsson, Jack Stockdale, Nicole Eagan among them. That is the hidden thread. It was not a garage story. It began with mathematicians and cyber-defence specialists in Cambridge, backed by a patron who could open doors, and with a team that already knew how to build and sell enterprise software. Their idea was unusual. Instead of keeping a list of known threats, Darktrace used unsupervised machine learning to learn the normal pattern of life of each network, device and user. Anything that behaved oddly stood out. Its Antigena product could respond on its own, and a visual tool showed activity as it happened. The company says it can catch zero-day attacks such as Log4j. Becoming known took visibility. It struck a partnership with McLaren Racing in 2020 and extended it in 2021. In April 2021 it listed on the London Stock Exchange at about £2.5 billion, and within months the value neared £7 billion. Then came the price. In January 2023 the New York hedge fund Quintessential Capital alleged accounting errors, especially in contracts from before the listing. Darktrace disputed it. By July an EY review found a small number of errors and inconsistencies, none material. The share price never regained its high. In April 2024 Thoma Bravo offered $5.3 billion. The deal completed on 1 October 2024, ending the listing. Gustafsson stepped down as chief executive in September 2024, and Jill Popelka's successor Ed Jennings took over in March 2026 after her 16 months. The last public year, 2023, showed revenue of $545.4 million and net income of $59.0 million. For a first-time founder, the lesson is about borrowed strength. Darktrace's start was helped by an experienced network, and its trouble came when public scrutiny hit a young story. Being well introduced opens the door; it does not answer the questions asked inside. ## Ineffable Intelligence (United Kingdom) URL: https://startup.talk/uk/ineffable-intelligence/ Founders: David Silver Founded: 2025 (founded Nov 2025); Based in: London, England; Sector: Artificial intelligence (reinforcement learning research) Before he asked investors for $1.1 billion, David Silver made a promise about what he would do with the money. He pledged to give away all of his equity proceeds from his new company, the largest commitment in the history of Founders Pledge. The founder of a start-up valued at $5.1 billion decided in advance not to keep it. He was born in 1976. He studied at Christ's College, Cambridge, graduating in 1997 with the Addison-Wesley award, and made a friend there named Demis Hassabis. Silver did not go straight into research. He co-founded Elixir Studios, a games company, where he was chief technology officer and lead programmer, and the studio won awards for technology and innovation. In 2004 he went back to academia, at the University of Alberta, for a PhD in reinforcement learning, finished in 2009. With Sylvain Gelly he built MoGo, among the strongest Go programs of its time. What almost nobody outside AI remembers is how slowly that path ran. He consulted for DeepMind from 2010 and joined full time in 2013, after a Royal Society fellowship and a lectureship at University College London. At DeepMind came AlphaGo, which received an honorary 9-dan professional certificate. He later won the 2019 ACM Prize in Computing and became a Fellow of the Royal Society in 2021. He founded Ineffable Intelligence in November 2025 and left Google DeepMind in January 2026. By February it was reportedly seeking a $1 billion seed at a $4 billion valuation. In April 2026 the round was announced at $1.1 billion (£880 million) and $5.1 billion. The price is the open question. The company is only months old, and its value rests on one person's record. For a first-time founder, the quieter point is the order of decisions. Silver settled what the money was for before he took it. Deciding your limits first can make everything after it easier to explain. ## Zepz (United Kingdom) URL: https://startup.talk/uk/zepz/ Founders: Ismail Ahmed, Catherine Wines, Richard Igoe Founded: 2010; Based in: London, England; Sector: Fintech, digital cross-border remittances Before he ran a company that moves money to more than 130 countries, Ismail Ahmed lost his job for telling the truth. He had been working at the United Nations Development Programme, helping run a money transfer project. Then he reported corruption inside the UNDP's Somalia programme. The reward was dismissal. Years later he was found to have been unfairly treated and was awarded GBP 200,000. Almost nobody who uses the app knows that. He was born in May 1960 in Hargeisa, then British Somaliland, and grew up there. A World Bank scholarship took him to the University of London to study economics, and he worked on a World Bank agricultural project in Hargeisa. Then the civil war came. His family helped smuggle him out, and he reached Britain as a refugee. The early years in exile were plain work. He took part-time jobs, including picking strawberries in Kent, and sent money home to family. He kept studying: a PhD in economics from the University of London and an MBA from London Business School. The man who would build a remittance company first lived the problem, as the person at the sending end paying the fee. That is the itch behind the company. When he later looked at the cost of sending money home, banks and traditional transfer firms charged heavily. Ahmed developed the idea of a mobile transfer service that would undercut them. WorldRemit launched in 2010, with Catherine Wines and Richard Igoe as co-founders. His UNDP years mattered here too, because he had seen how money moved, and how it was skimmed, in a place like Somalia. Growth came in steps. TCV led a $100M round in 2015, LeapFrog a $40M round in 2017, then $175M in 2019. In 2021 the company bought Sendwave and renamed the group Zepz. Along the way came the usual price of being a money company: a US regulator fined the operator of Sendwave $1.5M in 2023, and in 2025 Australia's AUSTRAC found weak monitoring of suspicious transactions at WorldRemit. Chief executives came and went, including Breon Corcoran in 2018 and Mark Lenhard in 2022. Today the network covers more than 5,000 corridors in over 70 currencies. In October 2020 Ahmed was ranked first on the Powerlist of the UK's most powerful people of African heritage. The lesson for a first founder is not subtle. Ahmed lost a salary for refusing to look away, then built a business whose whole pitch is that nobody should be skimmed. Your worst professional injury can become your product's promise, if you are honest about why it hurt. ## Fuse Energy (United Kingdom) URL: https://startup.talk/uk/fuse-energy/ Founders: Alan Chang, Charles Orr Founded: 2022; Based in: London, England; Sector: Energy supply, renewables generation In 2022, a man who had spent six years at the heart of one of Europe's fastest fintechs walked away to start an energy supplier. It sounded absurd. About 30 British suppliers had just gone bust. Alan Chang had been one of Revolut's early hires, working his way through operations to chief revenue officer, and reportedly seen as a close lieutenant of its founder. He left in September 2022. With him went Charles Orr, another early Revolut person who became Fuse's chief operating officer. Sources on Chang's birthplace and schooling are thin, and profiles even disagree on his degree, so the personal side of this story stays modest. What is documented is a reason: Chang has said he entered energy because of rising prices and climate change. The first move was a shortcut. In February 2022 the company acquired Paddington Power, a firm that already held an Ofgem electricity supply licence, later renamed Fuse Energy Supply. That avoided a long wait for permission. It also reflects an instinct from his Revolut years: own the licence, own the pace. Money came early. A $78M seed was led by Balderton and Lakestar. Then came assets rather than just software. A small generator, Balnamoon Renewables, was bought in July 2022. Public supply began in July 2023 after a beta phase. In November 2024 Ofgem granted a gas licence. Then the numbers began to move. Customers rose from about 50,000 homes in July 2025 to about 200,000 by December and past 300,000 by June 2026. In December 2025, $70M at a roughly $5bn valuation made it one of Britain's most richly valued private companies, up from about $1bn five months earlier. Its annual sales were GBP 129.7M, and the Sunday Times ranked its three-year sales growth at 484%. The risk is plain. Energy retail is a thin-margin business, and the last wave of fast-growing suppliers collapsed when wholesale prices turned. Fuse is building its own solar farms, including Netley Central in September 2025, partly to hold more of its own supply. For a first-time founder, the useful part is less about energy than about timing. Chang entered a market that had just been emptied of competitors and moved faster than anyone expected. Sometimes the best moment to start is straight after everyone else has been hurt. ## Lendable (United Kingdom) URL: https://startup.talk/uk/lendable/ Founders: Martin Kissinger, Victoria van Lennep, Paul Pamment, Jakob Schwarz Founded: 2014; Based in: London, England; Sector: Fintech, AI-driven consumer lending Before Lendable lent a penny, one of its founders was filming young people who were trying to fix the world. Victoria van Lennep co-founded GoYoung, an educational tool and documentary about young social entrepreneurs around the world. She also won a fellowship from the Fondation Wiener-Anspach for a master's in nature, society and environmental policy at Oxford in 2012-13. Then she took a job as thought leadership deputy editor at the Economist Intelligence Unit. Nothing in that path points to consumer credit. That is the part nobody expects. In 2014 she joined Martin Kissinger, with Paul Pamment and Jakob Schwarz, to start a lender in London. This is less a tale of one wound than of a team that decided a loan should be quicker, cheaper and less bureaucratic. A 2017 Independent profile described the aim as undercutting banks on speed, red tape and price. The first money was small. A GBP 2.5M seed from angels, including Passion Capital, paid for the early lending engine. It used UK open banking data to approve loans and pay out quickly. And here was the shift that made it a company rather than a clever app. Lendable did not hold the loans on its own balance sheet. It originated and serviced them for institutional investors, a two-sided marketplace. The cheques followed the proof. Waterfall Asset Management put in GBP 100M in March 2017, and the company is reported to have turned profitable that year. A year later Goldman Sachs committed GBP 200M. By early 2021 a secondary share sale made Lendable a unicorn. In March 2022 Ontario Teachers' led a GBP 210M round at around GBP 3.5bn. Expansion came with risk. It launched Autolend car finance and entered the United States in 2021, then Mexico in 2026, and in October 2025 launched a mobile product with Gigs. A lender depends on funding and bad-debt discipline. The figures are now large: GBP 446M revenue and GBP 114M operating income in 2025, and by Experian data the UK's biggest personal loan issuer by volume. For a founder, the quiet lesson is in that CV. The people who built a profitable lender came from journalism and social enterprise. Obsession with fixing a boring, unfair process can matter more than a finance pedigree. ## Rightmove (United Kingdom) URL: https://startup.talk/uk/rightmove/ Founders: Countrywide, Halifax, Royal & Sun Alliance and Connells (founding shareholders) Founded: 2000 (incorporated 16 May 2000; site launched Jul 2000); Based in: Milton Keynes, England; Sector: Online property portal In 2000 four companies that normally competed for the same house sellers agreed to put their listings in one place. That odd truce is how Britain's biggest property portal began. The founders were Countrywide and Connells, two big estate agency groups, plus Halifax and Royal & Sun Alliance, which owned agencies through their own financial businesses. Rightmove was incorporated on 16 May 2000, and the website launched that July. There is no single founder to follow. This is a story about an idea rather than a person. The idea cut against instinct. Agents guard their listings. But a shared site meant a buyer could search every agent at once, and an agent who stayed off it would be invisible. Rightmove charged estate agents for listings and extra advertising, and kept private sellers out. That choice protected its customers and, in time, made the portal hard to leave. For years it grew quietly in Milton Keynes. It floated on the London Stock Exchange on 15 March 2006, ending its life as a shared agent venture. It bought a majority of Holiday Lettings in 2007, The Outside View in 2016 and Van Mildert in 2019. People at the top changed. Peter Brooks-Johnson became chief executive in 2017 and resigned in May 2022. Johan Svanstrom took over in February 2023. The sharpest test was external. In September 2024 Australia's REA Group made an informal approach worth about GBP 5.6bn. Rightmove's board called it opportunistic and rejected it, and REA later walked away without a formal offer. The margin tells the rest: in 2025 revenue was GBP 425.1M and operating income GBP 287.9M, with net income of GBP 217.1M. Few companies keep two-thirds of sales as operating profit. The takeaway for a founder is dull in the best way. Rightmove did not win through a famous founder. It won by making every agent's listing more valuable inside one place than outside it. A network that your own customers cannot afford to skip is rarer than a clever product. ## Synthesia (United Kingdom) URL: https://startup.talk/uk/synthesia/ Founders: Victor Riparbelli, Steffen Tjerrild, Matthias Niessner, Lourdes Agapito Founded: 2017; Based in: London, England; Sector: Generative AI video The first seven months of Synthesia were reportedly paid for with Bitcoin profits. That is the line Victor Riparbelli has used, in interviews summarised online, to explain how a company with no revenue stayed alive. He grew up in Denmark, a boy drawn to science fiction. He studied computer science and business informatics at the IT University of Copenhagen, and a semester at Stanford made him want to start something. He says he was around startups for nearly a decade before Synthesia. Then he moved to London, attracted by its ecosystem of virtual reality companies. The idea arrived through a person. He met Matthias Niessner, a researcher in AI video, and the two founded the company in 2017 with Steffen Tjerrild and Lourdes Agapito. The first pitch was dubbing and translation, so that a film or advert could appear in any language with the actor's lips matching. A 2018 demo on the BBC's Click got attention. Customers were slower. For about two years the team was around ten people and struggled with sales. Riparbelli has called the early years 'kind of horrible'. Cash came in small pieces: $3.1M in 2019, $12.5M in 2021, $50M that December. The turn was a change of customer. Around 2021, enquiries from big companies pushed Synthesia away from entertainment and towards training videos. Instead of hiring cameras and actors, a firm could type a script. Riparbelli describes a rule of 'utility over novelty', dropping projects built just to impress. The price was trust. A 2023 Freedom House report said governments used the tools to make propaganda, and the actor Dan Dewhirst found his likeness in Venezuelan videos. Synthesia now says it requires consent and screens new users. In October 2025 it raised $200M at a $4bn valuation, with revenue past $100M, and said it served most of the Fortune 100. It reportedly turned down a $3B offer from Adobe in April 2025. For first-time founders: the product that paid for the early months was a side bet, but the one that made the company was a customer who asked. Listen to who keeps calling. ## Ocado (United Kingdom) URL: https://startup.talk/uk/ocado/ Founders: Tim Steiner, Jason Gissing, Jonathan Faiman Founded: 2000 (began as L. M. Solutions; trading with Waitrose from Jan 2002); Based in: Hatfield, England; Sector: Online grocery and warehouse automation technology For eight years Tim Steiner bought and sold bonds. He joined Goldman Sachs in 1992, and the job took him from London to New York to Hong Kong. Then in 2000 he walked out of it to start a company that would deliver milk and bread. He had grown up in North London and gone to Haberdashers' Boys' School in Elstree, then studied economics at the University of Manchester. Nothing in that path points to a warehouse floor. The idea belonged to three men, not one: Steiner, Jason Gissing and Jonathan Faiman, all Goldman bankers, who began as a small outfit called L. M. Solutions in the year the dot-com bubble burst. That timing is the part people forget. They were building an online grocer just as the internet money dried up. Their first real customer did not arrive until January 2002, when Waitrose agreed to let them deliver its food. John Lewis later held a stake in the company, and sold it in 2011. What almost nobody notices is how unglamorous the bet was. Online grocery loses money on every van trip unless the warehouse is clever. So Ocado kept building its own robots and software, and went public in London in July 2010. Years later that decision became the company. Casino signed in 2017. Sobeys and Kroger followed in 2018, Coles in 2020. The grocer had become a technology licensor. The price came in February 2019, when a fire destroyed much of its Andover warehouse. The rebuilt site did not reopen until August 2021. In the same period M&S agreed to pay GBP 750m for half of the retail business, a deal that has since turned into a payment dispute. By mid-2026 the shares were reported to be below the IPO price, and some shareholders wanted Steiner replaced. Steiner had also been paid GBP 58.7m for the year to December 2019, a number critics noticed. For FY2025 the group reported revenue of GBP 1,361.5m, an operating loss of GBP 241.4m and net income of GBP 395.2m. Steiner's story is a reminder that a trader's habit, taking a long position and waiting, can look like genius or stubbornness depending on the year. He was a man who stayed with one unfashionable bet for twenty-six years. ## Thought Machine (United Kingdom) URL: https://startup.talk/uk/thought-machine/ Founders: Paul Taylor Founded: 2014; Based in: London, England; Sector: Cloud-native core banking software In 2018 a British high-street bank did something large banks almost never do. It put money into a startup that wanted to replace the system at the heart of how a bank counts money. That startup was Thought Machine, founded in London in 2014 by Paul Taylor. This is a story about stakes rather than childhood. The stakes were these. A core banking system holds every account, balance and loan. Banks run decades-old versions because replacing them can break everything. Taylor's pitch was cloud-native software, Vault Core, which lets a bank define products with smart contracts. The 2018 Series A of $25M included Lloyds Banking Group, which also became a customer. What followed was a ladder of bigger cheques. A $125M Series B in 2020 was led by Molten Ventures, Eurazeo, the British Business Bank and SEB. In 2021 came $200M at a $1B valuation, and in 2022 a $160M Series D at $2.7B with Temasek, Intesa Sanpaolo and Morgan Stanley. By then JPMorgan, Standard Chartered and ING Poland were in the client list. Andy Maguire, a former HSBC group COO, became chair in 2020. The cost of selling to banks is time. Contracts are slow, and a failed migration is public. After the 2022 round the company did not publish a new valuation. Taylor has reportedly said he wants less emphasis on valuation and more on commercial results, and a London listing has been pushed to 2028 at the earliest. The company says it passed $100M in revenue in 2025, and one outlet reports it turned free cash flow positive in the second half of that year. The lesson for a founder is narrow but real. When your customers move slowly, one believing early customer can be worth more than a hundred polite meetings. Everything after Lloyds was easier to say yes to. ## Improbable (United Kingdom) URL: https://startup.talk/uk/improbable/ Founders: Herman Narula, Rob Whitehead, Peter Lipka Founded: 2012; Based in: London, England; Sector: Simulation software, metaverse infrastructure, games tools In the early days Improbable did not have an office. It had Herman Narula's parents' house in Hertfordshire. Narula was born in Delhi in 1988 and brought to Britain when he was three. His father, Harpinder Singh Narula, runs a family construction company, DSC, where both of Herman's older brothers work. His mother, Surina, is a philanthropist. Herman went to Haberdashers' Aske's in Elstree and then to Girton College, Cambridge, to study computer science. At Girton he met Rob Whitehead. They shared two things: a love of games and a reluctance to go into finance, the usual exit for Cambridge graduates. What almost nobody mentions is who paid for the start. The first money was about $1m in a loan from the Narula family, plus $1.2m from angel investors. For a family that built its wealth on construction, that was a bet on a son's game-engine idea. The records read here do not say how the loan was discussed at home, and it would be wrong to guess. The idea was SpatialOS, software for building simulations too big for one machine. In March 2015 Andreessen Horowitz led a $20m round. In May 2017 SoftBank led $502m, then the largest venture investment in a European tech company. NetEase added $100m in 2018 at a valuation above $2bn. Then it nearly broke. In 2019 Unity blocked several SpatialOS games, and Epic Games helped put together a $25m fund for affected developers. The dispute was later resolved, but the company's core product had been shaken. It moved toward defence, the MSquared metaverse venture and, in 2023, sold its defence division. The same year it reported its first profit, GBP 11m on GBP 66m revenue. A family loan carries a different weight than venture money. Investors can walk away from a bad bet. A father cannot, and a founder knows it. Narula has spent over a decade making sure the loan was never his family's mistake. ## OnlyFans (United Kingdom) URL: https://startup.talk/uk/onlyfans/ Founders: Tim Stokely, Guy Stokely, Leonid Radvinsky Founded: 2016; Based in: London, England; Sector: Creator subscription platform The way Tim Stokely's father reportedly put it was blunt: this is going to be the last one. The father was Guy, a retired investment banker from Barclays. The son, Tim, was born in Harlow, Essex, in 1983, the youngest of four children. Even at school he had a hustle. He took orders for a local fish and chip shop and added a markup for delivering them. Then came the string of attempts. He studied at Anglia Ruskin University and went on to launch several sites, among them GlamWorship and Customs4U, adult-performance sites, and a platform linking customers with tradespeople. None of them became OnlyFans. The sources do not describe how each one ended, only that there were several before the one that stuck. In 2016 Guy lent him GBP 10,000. In November that year OnlyFans launched as a place where performers could directly monetise their content. Tim's brother Thomas became chief operating officer and Guy took over finance. It was a family company before it was anything else. What almost nobody expected was the money. In 2019 Leonid Radvinsky, who owned the adult site MyFreeCams, bought 75% of the company behind it. In 2020 revenue was about $350m. In 2021 it reached roughly $900m to $932m. Then came the week that tested everything. In August 2021 OnlyFans said it would ban sexually explicit content from October, blaming pressure from payment processors. Stokely later told the Financial Times that banks including BNY Mellon and JPMorgan Chase had withdrawn support. After a backlash from the creators who earned its money, the ban was reversed six days later. The company has carried heavier problems since: child-safety investigations, impersonation lawsuits, and a block in China in December 2024. For the year to November 2025 it reported revenue of $1.6bn and pre-tax profit of $715m. A sale at $8bn was sought in early 2026 and found no buyer, and a minority stake later changed hands at $3.15bn. A founder who has failed several times is carrying something a first-timer does not: a father's quiet doubt. The cheque was GBP 10,000. The line that came with it may have cost more. ## XTX Markets (United Kingdom) URL: https://startup.talk/uk/xtx-markets/ Founders: Alexander Gerko Founded: 2015; Based in: London, England; Sector: Algorithmic trading and market making In July 2024 a man who had paid more tax than anyone in Britain lost a court case about tax. Alexander Gerko had been named the UK's largest taxpayer by the Sunday Times that January, with GBP 664.5m. The Court of Appeal case, against HM Revenue and Customs, concerned the treatment of a deferred payment plan from his years at the hedge fund GSA Capital. He lost. He was born in Russia in December 1979 to a Russian Jewish family. He studied mathematics at Moscow State University, earning bachelor's, master's and doctoral degrees, and added an economics master's at the New Economic School and another degree at the Independent University of Moscow. He wrote research papers, including one on testing market timing. In 2004 he started trading equities at Deutsche Bank and then moved to foreign exchange. In 2006 he moved to the UK. From 2009 he was at GSA Capital, the hedge fund where that deferred pay arrangement later led him to court. What almost nobody pictures behind the billionaire is a man whose route to London ran through blackboards and exam halls. In January 2015 he founded XTX Markets as a spin-off of GSA, and kept about 75% of it. The firm used machine-learning models to quote prices for currencies and, later, European equities. In 2016 it ranked ninth by FX volume, the first non-bank in the top ten. By 2019 it was the largest FX spot liquidity provider. In 2022 revenue rose 68% to GBP 2.5bn, with profit of GBP 1.1bn. Money has pushed him toward mathematics itself. XTX has committed over GBP 250m to charity since 2017, much of it for maths education, and launched a $10m AI prize for a model that can win gold at the International Mathematical Olympiad. In 2022 he gave up his Russian citizenship and became British. The cost of a mathematician's rise is that the private rules he once negotiated become public. For a founder, the quiet lesson is that early pay deals and tax structures outlive the job, and can come back as headlines. ## CMR Surgical (United Kingdom) URL: https://startup.talk/uk/cmr-surgical/ Founders: Mark Slack, Martin Frost, Luke Hares, Keith Marshall, Paul Roberts Founded: 2014 (incorporated January 2014 as Cambridge Medical Robotics; renamed CMR Surgical March 2018); Based in: Cambridge, England; Sector: Medtech, surgical robotics The most important introduction in British surgical robotics may have happened in a room full of expectant parents. CMR Surgical began with two men talking. Mark Slack was a surgeon fixated on a stubborn fact about keyhole surgery: far less of it was being done than could be done. Luke Hares was an engineer. Hares brought in people he had worked with before in Cambridge: Keith Marshall, Paul Roberts and Martin Frost. In January 2014 they incorporated Cambridge Medical Robotics. Frost has described the aim in simple terms: bring minimal-access surgery to people who do not have it, and in doing so change surgery. Keyhole operations mean smaller cuts and faster recovery, but they are hard on the surgeon, who stands for hours twisting long instruments through tiny holes. A robot could take that strain. The obvious route was to build another da Vinci. They did not. Their machine, Versius, uses small, separate, modular arms that can be wheeled into any theatre and set up quickly. That mattered for hospitals without a dedicated robotic suite. A surgeon in India told reporters it let him sit at an open console and do six procedures in one day. Nobody had expected a start-up from a business park outside Cambridge to attract serious money. It did. A $240M round in 2019 made it a unicorn, and in 2021 it raised $600M at a $3bn valuation, which the company called the largest private medtech financing ever. The name changed from Cambridge Medical Robotics to CMR Surgical in 2018, a sign it wanted to sound like a global player. Then came the price. In April 2023 about 350 staff were made redundant, as investors grew cautious and selling expensive hardware proved slower than the pitch deck. The company kept going with a $165M raise that September and a GBP 10m factory in Ely able to make up to 500 systems a year. The US was always the prize, and FDA clearance for an upgraded Versius finally came in October 2024. In 2025 it raised more than $200M, and the Financial Times reported it was exploring a sale at up to $4bn. No buyer has been confirmed in anything I could find. There are small, human moments in the numbers too. In 2024 a seven-year-old boy from Gosport became the first child in the UK to have Versius surgery, in an NHS trial. More than 20,000 cases had been completed by March 2024. The lesson for founders is almost embarrassingly ordinary. The hardest part of this company was not the robot. It was getting five people with the right skills into the same conversation, and then keeping them in it for a decade while the money, and the market, kept moving. ## Deliveroo (United Kingdom) URL: https://startup.talk/uk/deliveroo/ Founders: Will Shu, Greg Orlowski Founded: 2013; Based in: London, England; Sector: Online food delivery For the first eight months of Deliveroo's life, the chief executive was also the delivery boy. Will Shu rode the orders himself, every day. Not as a stunt. He wanted to know what the customer felt when a bag arrived late, and what the restaurant felt when a rider showed up and the kitchen was not ready. Almost nobody who later argued about gig-work knew that the man at the top had started the job on a bike. He was born in Connecticut in December 1979 to Taiwanese American parents, studied at Northwestern, and in 2001 landed in London as an analyst at Morgan Stanley. Banking meant long nights. The office was full of tired people and the delivery options were bleak. He later moved through hedge-fund jobs, then went back to the US in 2010 for an MBA at Wharton. The itch to build something stayed. He started Deliveroo in 2013 with Greg Orlowski, a software engineer and childhood friend. For two years it grew mainly by word of mouth in London before it opened in Paris, Berlin and Dublin in 2015. Money followed fast: GBP 2.75M in 2014 from Index and Hoxton, then Accel, then rounds so large they became routine, including $575M led by Amazon in 2019. The fall-out was loud. In 2016 striking London riders gave the company a nickname, "Slaveroo", that outlets later repeated. Years of court cases over pay, sick pay and self-employed status followed. A founder who had once ridden the same streets now had to answer for the people riding them. There were retreats too: Germany, Spain, the Netherlands, Hong Kong, Australia. The low point was public. Deliveroo listed in London on 31 March 2021 at 390p and closed the day at 284p, down 14%. It was widely called one of the worst debuts in the city's history. At the float the company gave GBP 16M to its most active riders, up to GBP 10,000 each, an acknowledgment, perhaps, of who had built the thing. Then, calm. By FY2024 revenue was GBP 2,071.9M with a net loss of just GBP 0.1M. DoorDash bid, the boards agreed on a GBP 2.9bn deal in May 2025, and it completed in October. What stays with me is the habit rather than the exit. Shu reportedly still takes the odd shift himself, every few weeks. Founders talk about staying close to the customer. He literally went back to the street. It is the cheapest form of market research there is, and the one most people stop doing the moment they can afford not to. ## Quantexa (United Kingdom) URL: https://startup.talk/uk/quantexa/ Founders: Vishal Marria, Imam Hoque Founded: 2016; Based in: London, England; Sector: Enterprise AI, decision intelligence, data analytics Vishal Marria was about six when he started working. After school in Clapham he would hurry to his parents' cash-and-carry in Brixton, along with his three siblings. Every day of the week. Nobody running a venture-backed analytics firm today would call that a training programme, but he later credited the counter with shaping his business instincts. One podcast listing even claims a frightening raid on the shop when he was nine. I could only find that once, so take it as unconfirmed. His parents had come from India in the 1960s. They started with deliveries, moved to off-licences, then to cash-and-carry. Money in that world is counted in coins and margins, and the lesson is simple: you see waste at once. He went on to study computer science at Royal Holloway, finishing with a first. Then came jobs at Detica, SAS and EY, where he worked in risk. Here the wound turned into an itch. Banks needed to know who they were dealing with, yet analysts spent days manually combining datasets that did not agree with each other. Marria saw a gap for software that could connect the dots on its own. What almost nobody saw coming was how big the pay-off would be for a company with no famous founder, no Silicon Valley address and a product that is hard to explain at dinner. He has said the decision to leave and start up felt like needing to "place a big brave hand on my heart". In 2016 he and Imam Hoque founded Quantexa in London. The cost of building for banks is patience. Regulated customers move slowly, and the product has to work before anyone believes it. Then in April 2023, despite UK venture funding falling sharply, Quantexa raised $129M at a $1.8bn valuation and became the only new British unicorn of the year at that point. Warburg Pincus and Singapore's GIC had led its recent rounds. The next step was bigger. In March 2025 it raised $175M, led by the venture arm of Ontario Teachers', at a $2.6bn valuation. Annual recurring revenue had passed $100M in October 2024, Zurich had signed on as the first insurer to use its fraud platform, and an IPO was reportedly being considered for as soon as 2026. The pattern is familiar if you grew up behind a counter. Marria turned a childhood spent watching what customers actually needed into a company that reads what customers actually did. Founders from ordinary trading families often carry the same quiet edge: they learned to look at data before anyone called it that. ## CuspAI (United Kingdom) URL: https://startup.talk/uk/cuspai/ Founders: Chad Edwards, Max Welling Founded: 2024; Based in: Cambridge, England; Sector: AI for materials discovery, climate and semiconductors Chad Edwards was a boy in rural Wales with an ADHD diagnosis, the sort of label that can leave a child behind at school. What pulled him back, he has said, were science teachers with the patience to keep him interested. He describes himself as a visual thinker who treats science as literal, not theoretical. Later, when he talks about chemistry, he says he always saw it as a form of computation. He was the first in his family to go to university. He chose chemistry. The route from there was not a straight line to Silicon Valley. One profile reports a PhD in computational quantum chemistry across Manchester and Karlsruhe. Then came jobs at Google, at the chemicals company BASF, and at quantum computing firms where he helped sell technology to its first big customers. He was seeing the same bottleneck again and again: inventing a new material takes years of trial and error, and the world needed new ones for batteries, carbon capture and chips. The turn came through a phone call. Max Welling, a Dutch physicist turned machine-learning pioneer, was leaving his vice-president role at Microsoft Research when Edwards reached out with his plans. Welling had a physics PhD under Gerard 't Hooft, had sold his first company, Scyfer, to Qualcomm in 2017, and had worked on the variational autoencoder with Durk Kingma. Edwards later found that they shared the same outlook, which he took as validation. In April 2024 they started CuspAI in Cambridge. Two months later, in June, they came out of stealth with $30M and the Nobel-winning AI researcher Geoffrey Hinton on the advisory side. The cost of being new in a hyped field is that everyone watches the claims. Materials discovery has seen glossy announcements before, and the proof is a real compound made in a real lab. The company has also had to compete for scarce researchers with bigger labs. It has not lacked backers. A $100M round followed in September 2025, and in July 2026 CuspAI announced $450M at a $2.6bn valuation, with Kleiner Perkins and NEA leading and Bezos Expeditions and the UK Sovereign AI Fund among investors. The valuation more than quadrupled in less than a year. It also launched what it calls an AI Materials Foundry. There is a quiet lesson in the founder, not the funding. The kid who might have been written off in a classroom ended up running a company whose job is to look at old problems with fresh eyes. Teachers who stay patient do not just save a grade. Sometimes they hand a future company its chief executive. ## OakNorth (United Kingdom) URL: https://startup.talk/uk/oaknorth/ Founders: Rishi Khosla, Joel Perlman Founded: 2015 (bank licensed March 2015; products launched September 2015); Based in: London, England; Sector: Fintech, SME lending bank, credit analytics software Rishi Khosla walked into University College London in 1992 at just 17. He was born in London in 1975 and went to secondary school in South Harrow. His father had an engineering degree and later sold construction equipment in the Middle East, which meant long absences and, for a stretch, a household that moved. The family spent time in Delhi, living with relatives. That detail is easy to skip, but it frames a man who would later build a bank for businesses that traditional lenders overlooked. After UCL came a degree in economics and a master's in accounting and finance at the London School of Economics. In 1995 he joined the Dutch bank ABN AMRO, then GE Capital in 1999. Later he worked for Lakshmi Mittal, running a private-equity portfolio focused on fintech. In that job he became an early investor in PayPal. He had seen, from the inside, both how banks decide who gets money and how technology could change that. What almost nobody outside finance knew was how thin the field was. In March 2015 Khosla and co-founder Joel Perlman, who had earlier built a research firm called Copal Amba, got a full banking licence from the PRA and FCA, making OakNorth the third new British bank in more than 150 years. Products launched in September that year. Then they did something unusual. In May 2016 OakNorth became the first bank fully hosted on the cloud, working with Amazon Web Services. The bet was that software could assess a growing firm's future, not just its past, so a bank without branches could lend with confidence. The road was not free of friction. Khosla has lobbied against capital rules that favour large banks. In March 2023 OakNorth submitted a rescue bid for the UK arm of Silicon Valley Bank. It did not win it, yet the move put the name in front of the whole City. The numbers say the model works. Net income was GBP 159.3M in 2024, and revenue reached GBP 309.9M in 2025. The Financial Times had ranked it first on its FT 1000 list of Europe's fastest-growing companies in 2020, the year Khosla received an OBE. In 2019 SoftBank's Vision Fund and Clermont Group invested $440M. The insight a founder can take is the simple one buried in his biography. A son of a travelling salesman learned young that opportunity depends on who is willing to look closely at you. Then he built a bank that looks closely. ## PhysicsX (United Kingdom) URL: https://startup.talk/uk/physicsx/ Founders: Jacomo Corbo, Robin Tuluie Founded: ; Based in: London, England; Sector: AI-native engineering simulation software (aerospace, defence, automotive, semiconductors, energy) Jacomo Corbo was in the middle of a computer science PhD at Harvard when he stepped out of it. The reason was not a lab. It was a Formula 1 team. That detour matters more than it looks. On the grid, engineers live and die by simulation: how air bends over a wing, how a tyre heats. Every run costs hours of computing, and the season does not wait. Somewhere in that world Corbo met Robin Tuluie, a veteran who had led research and development at two F1 teams, Renault's Alpine and Mercedes, and before that run vehicle technology at Bentley. Corbo's own path then turned in two directions at once. He became Chief Scientist at QuantumBlack, the data consultancy, and he also worked as Chief Race Strategist at Renault's F1 team. Numbers by day, split-second calls by weekend. Sources do not say what he felt in those years, and nothing here pretends to. What the two men saw was a gap. Engineers in aerospace, defence and cars were still waiting on slow solvers from the likes of ANSYS and Dassault Systemes. Machine learning could learn the physics from past simulations and answer in seconds. The idea was simple to say and hard to sell, because the customers build things that must not fail. The company took shape in London. In November 2023 it raised a Series A, and then the quiet part ended. By June 2025 it had closed a EUR 117.3 million Series B led by Atomico, with Temasek, Siemens and Applied Materials joining. The firm said revenue had more than quadrupled since the Series A, and headcount passed 150. Nvidia's venture arm added $20 million that November, bringing the Series B past $155 million at a valuation near $1 billion. Then, in June 2026, secondary reports said a $300 million round led by Temasek had pushed it to $2.4 billion. Treat that last figure with care; the coverage is thin. There is a price to being early in industrial AI. Contracts are slow, trust is earned model by model, and the incumbents are not standing still. The founders have said little publicly about the lows, and the record on their private lives is thin. What the story does show is where the idea came from: not from a pitch deck but from the pit wall, where simulation either saves a race or wastes a night. For a first-time founder, the lesson is plain. The unfair advantage is often a problem you watched up close for years before anyone paid you to solve it. ## Lighthouse (United Kingdom) URL: https://startup.talk/uk/lighthouse/ Founders: Gino Engels, Matthias Geeroms, Adriaan Coppens Founded: 2012 (as OTA Insight; rebranded Lighthouse 2023); Based in: London, England (founded in Ghent, Belgium); Sector: Commercial intelligence software for hotels and travel The idea reportedly arrived on a trip to the London Olympics in 2012. Two Belgians were watching a city fill with visitors and, presumably, watching hotel prices move. Sources give just that one line, so the rest is the company's record, not a scene. Gino Engels and Matthias Geeroms started in Ghent. Neither came from the hotel trade as far as profiles show. Engels had spent about three years at AB InBev's UK arm. Geeroms had spent six years as a transfer pricing specialist at the Belgian tax authority, a job built on spotting numbers that did not add up. A third name, Adriaan Coppens, appears as co-founder in some profiles. That oddity is the hook. Tax inspectors read prices for a living, and a hotel's room rate is one of the most volatile prices anywhere. The company they built, OTA Insight, collected rates from booking sites and showed hoteliers what rivals were charging, hour by hour. The name pointed at online travel agents, the middlemen every hotel had to read. For years it grew quietly. By November 2021 it had raised $100 million in total, after an $80 million Series B led by Spectrum Equity. The company moved its headquarters to London. Then it started buying: Transparent in Madrid, Kriya RevGen in Dallas, and Stardekk, a Belgian channel manager, in 2024. In 2023 it dropped the OTA Insight name. The old one described a tool; the new one, Lighthouse, described a platform. In November 2024 KKR led a $370 million Series C, taking total funding to about $470 million. What it has become is an everyday tool for more than 70,000 hospitality businesses, with over 700 staff. It has won the HotelTechAwards five years running. What remains is a pattern worth copying. Neither founder was a hotelier. They arrived with a skill, reading prices critically, and aimed it at an industry that mostly guessed. A first-time founder does not need the industry's résumé. Sometimes the outsider's habit is the product. ## Teya (United Kingdom) URL: https://startup.talk/uk/teya/ Founders: Eduardo Pontes, Ali Mazanderani Founded: 2019 (as SaltPay, London; renamed Teya April 2023); Based in: London, England; Sector: Fintech: card payments, business accounts and credit for small merchants In 2018 Eduardo Pontes left Rio de Janeiro for London. He told a Brazilian magazine he went there "to learn and try to do something new." He was 40 and already rich, yet he turned up to a Central Bank event in Brazil in jeans and a T-shirt. Pontes had spent fifteen years building payments businesses with his partner André Street. They built Braspag in 2004 and sold it to Grupo Silvio Santos. In 2013 they launched Stone, where Pontes was the technologist. He worked out that Brazil's patchy data networks were the real enemy of the card machine, and he pushed for a terminal that used as little data as possible. Colleagues described him as stubborn and obsessive about detail, and as someone who avoided the spotlight. In 2019 he started again in London. SaltPay was set up with Ali Mazanderani, who became chairman, and it was meant to be a Stone for Europe: card terminals, software and customer tools for small shops. It stayed independent of Stone. By August 2020 it had about 500 staff and offices from England to Croatia, Iceland and South Africa. The first big move was odd. In July 2020 SaltPay, Street's investment vehicle and the Lemann-backed LTS Investments bought 96 percent of Borgun, an Icelandic card acquirer with roots in the country's first payment card company, for 27 million euros. Borgun's chief executive and some directors left straight after. Pontes stepped in to help lead it, with Marcos Nunes, a former Bambora executive. An investor described the plan as a turnaround. Then the money arrived. In November 2021, with Tiger Global, Ribbit Capital, GIC and Michael Spencer among the backers, SaltPay raised a Series C reported at roughly 400 to 700 million dollars, depending on the account, at a valuation that reports put near 9 billion dollars. It bought businesses across Europe with it, including Paymentology, Storyous and Yoyo. The reset came with a new name. On 4 April 2023 SaltPay became Teya, promising next-day settlement, simple pricing and no long contracts. The pitch has widened since to business accounts, credit and Tap to Pay on iPhone, and in February 2026 the company launched in Spain and Italy taking it to nine European countries. It has about 75,000 customers and more than 200 staff in its UK arm. Turnover in its latest UK accounts was 19 million pounds for 2024, more than double the year before, while the pre-tax loss shrank from 110 million to 64 million pounds. It was valued at 2.4 billion dollars in November 2026. The lesson in Pontes's story is that a second company is rarely a copy of the first. He took a playbook from Brazil, bought a struggling acquirer in a small country, and spent his pile of capital on making it big enough to matter. ## ManyPets (United Kingdom) URL: https://startup.talk/uk/manypets/ Founders: Steven Mendel, Guy Farley Founded: 2012 (first pet insurance policy launched 2017); Based in: London, England; Sector: Pet insurance (insurtech) ManyPets did not sell a single policy for its first five years. It was founded in 2012. Its first pet insurance went on sale in 2017. That gap is the story. Steven Mendel and Guy Farley, who run the company as CEO and CTO, did not begin with a product. By their own account in an Octopus Ventures profile, they began by asking thousands of pet owners what was wrong with the insurance they already had. The answer shaped what they built. The company was then called Bought By Many, a name that told you the method: group people with the same need, such as owners of pets that mainstream insurers shied away from, and bargain on their behalf. The research pass found no source describing the founders' early lives, schools or birthplaces, so none are claimed here. Money arrived before revenue. Octopus Ventures invested for the first time in 2016, a year before the first policy, and later led the Series A. Reports put that round at about $9 million in January 2017. Many more followed. By June 2021 a $350 million Series D valued the company at more than $2 billion. The trouble with insurance is that growth costs cash. Claims come before profit. Revenue fell 16 percent to GBP 29.6 million for 2024. Still, the company says it covers more than half a million pets worldwide and has added extras: vet video calls, discounts on flea and worming treatment. It was valued at $2.4 billion in November 2026. What stands out is patience. Five years of listening before selling, and an investor willing to wait through them. For a first-time founder the takeaway is practical. Regulated products are slow. If you are going to spend years before the first sale, spend them asking customers what they wish existed, and find the backer who will still be there when you finally launch. ## Multiverse (United Kingdom) URL: https://startup.talk/uk/multiverse/ Founders: Euan Blair, Sophie Adelman Founded: 2016 (as WhiteHat; rebranded Multiverse 2021); Based in: London, England; Sector: Edtech: apprenticeships and workforce upskilling in tech and AI In 2001 Euan Blair applied to Trinity College, Oxford. The papers followed it closely. His parents, Tony and Cherie Blair, complained to the Press Complaints Commission about the coverage, and the complaint was upheld. A 2002 Telegraph headline put the outcome plainly: he misses out. He was seventeen, the prime minister's eldest son, and one of four deputy head pupils at the London Oratory School. A year earlier, at sixteen, police had found him drunk in Leicester Square after his last GCSE. He apologised and was not prosecuted. Those are the moments the public remembers. He went to Bristol instead and read ancient history, graduating in 2005 with a 2:1. Then a Yale master's in international relations, paid in part by a GBP 50,000 scholarship. The early jobs read like a politician's son trying doors: an unpaid internship in the US House, another with a Democrat he left after two weeks in 2006, a short stint at a PR firm, a runner's job on a film. None of it stuck. What almost nobody connects is the pattern. The person who did not get the famous university place went on, in 2016, to found WhiteHat with Sophie Adelman, a company that offers school leavers a different road into work. It launched with a GBP 3 million seed round from Lightspeed in 2018. The road was not smooth. Losses ran to GBP 40.5 million on GBP 45.2 million of revenue in the year to March 2023, and GBP 60.6 million on GBP 58.4 million the year after. Fortune reported layoffs and a pivot to upskilling in 2024. In February 2026 Ofsted rated the firm 'Needs Attention'. Yet the company, renamed Multiverse in 2021, hit a GBP 1.2 billion valuation in June 2022, bought SearchLight in 2024 and reported its first cash-flow-positive quarter in early 2026. Revenue reached GBP 79.6 million for the year to March 2025. Blair owns about 20 percent and has appeared on the Sunday Times Rich List at roughly GBP 350 million. He has not, in anything I read, tied the Oxford rejection to the company. That link is mine to notice, not his to confirm. Still, a founder with an unusually public early failure can build for exactly the people that failure lets you see. The lesson is quieter than redemption: your first rejection can tell you who the product is for. ## Zilch (United Kingdom) URL: https://startup.talk/uk/zilch/ Founders: Philip Belamant, Serge Belamant, Sean O'Connor Founded: 2018; Based in: London, England; Sector: Fintech, buy now pay later, consumer credit A man who had built payment companies on two continents could not get a credit card. Philip Belamant moved to London in 2015. On paper he was a fintech founder. In the eyes of UK lenders he was nobody: no credit history, no score, no file. A Times headline captured the sting: he had founded a unicorn but could not get a credit card. He grew up in Johannesburg, the son of Serge Belamant, an engineer born in Tulle, France, who had arrived in South Africa at 14 and went on to build payment systems. Philip studied IT at Rand Afrikaans University, finishing with a BSc and an honours degree. While still a student he was experimenting with mobile solutions for people the banks ignored. He started early. His prepaid and virtual card company PBel ran from 2006 to 2012. Moola Mobile, begun in Namibia, let people earn airtime by watching adverts, and reached Botswana and Nigeria. Later he ran Net1 Mobile Solutions. Each one was a different answer to the same problem: how do you serve people who do not fit a lender's boxes? What almost nobody saw coming was that London would hand him that problem personally. Out of the frustration came Zilch, founded in 2018 with his father as chairman and Sean O'Connor as co-founder. The idea was to make credit work without charging shoppers interest, with merchants paying instead. The FCA took it into its Regulatory Sandbox in 2019, and the app opened to UK consumers in September 2020. Growth was quick and expensive. A GBP 83M Series B from Goldman Sachs and DMG Ventures came in 2021, then a Series C led by Ventura Capital and Gauss Ventures that took the valuation to $2bn that November. A US push followed in Miami in 2022, and in March 2024 the company paused it to chase profit. That bet paid off. By September 2024 Zilch said it was profitable on a $130M annual run rate, and by November 2025 it counted more than 5.3 million customers, had raised over $175M led by KKCG, and had become Arsenal's official way to pay. The lesson is uncomfortable and useful. Belamant did not find his company in a market report. He found it at a lender's counter, being told he did not exist. The best problems for a founder are often the ones that happen to them first. ## Copper (United Kingdom) URL: https://startup.talk/uk/copper/ Founders: Dmitry Tokarev Founded: 2018; Based in: London, England; Sector: Crypto custody, prime brokerage and settlement Dmitry Tokarev was a quant before he was a crypto founder, and that explains almost everything about Copper. He came to London to study risk management and financial engineering at Imperial College, graduating with distinction, and started as a quantitative analyst. Later he moved to Dolfin Asset Management in Mayfair, where he became chief technology officer and a partner. His job was to think about what could go wrong with money. Then crypto arrived. Institutions wanted exposure, yet the tools for holding digital assets were crude. He saw that investors had no workable way to store them with the control and speed they needed. He founded Copper in London in 2018 to close that gap. The company offered custody, prime broking and settlement for institutions in one place. For a risk person it was the obvious design. For a market built on exchanges that sometimes vanished, it turned out to be well timed. What almost nobody saw coming was the speed of the climb. In 2021 the company went looking for up to $500M at about $2.5bn. The price settled lower, near $2bn, when the Series C closed in October 2022 with around $196M including convertible notes, and with Barclays on the register. That deal closed in a crypto winter. Money was available. Confidence was not. The cost showed up in the second act. In October 2024 Tokarev handed the chief executive role to Amar Kuchinad, a former Goldman Sachs executive, and stayed on as a founder director and shareholder. Copper began applying its collateral technology to traditional finance. Then, in May 2026, Cantor Fitzgerald was hired to run a sale process marketed at $500M, a quarter of the Series C figure. Kuchinad said he would leave and work on a succession plan for 2027. It is a sobering arc: the founder who built a risk business stepped back before his company's own valuation risk arrived. The takeaway for founders is blunt. A headline valuation is a promise about the future, not a fact about the present. What you control is whether the product solves a problem people will still have when the cycle turns. ## Marshmallow (United Kingdom) URL: https://startup.talk/uk/marshmallow/ Founders: Oliver Kent-Braham, Alexander Kent-Braham, David Goate Founded: 2017 (incorporated 2016 as Twin Thinking); Based in: London, England; Sector: Insurtech, car insurance For ten months, the office was a gym lobby. Oliver and Alexander Kent-Braham, twins from south-west London, had a GBP 80 Virgin Active membership and a laptop each. That was the headquarters of what would become a unicorn. Oliver had quit a banking job at Investec after about six months. He describes the pair as having 'boring business type' backgrounds. Nobody was going to hand them credit for a story. The itch came from a friend. A Canadian who had just arrived in Britain kept getting outrageous car-insurance quotes, not because he was a bad driver but because he had no UK history. The brothers realised that deeper data might show the opposite of what a thin file suggested. So they built for the people the market overlooked. With software engineer David Goate, they incorporated the business in 2016 as Twin Thinking and renamed it Marshmallow in 2017. They liked the idea of something soft and friendly, and wanted a name that stood out from insurers whose names began with 'A'. The choice provoked a legal fight: Marsh McLennan challenged the name in 2018. Marshmallow won. The money was small at first. In 2018 Investec, Oliver's former employer, and Passion Capital put in a $1.2M seed to build an insurer aimed mainly at migrants. Then came the jump. In 2021 an $85M round valued the company at $1.25bn, making it a unicorn. The twins split the job: Oliver takes the outside world and fundraising, Alexander culture and product. Along the way came another legal tussle, with Mulsanne Insurance over intellectual property in 2021, settled out of court. The growth was real. The Financial Times ranked Marshmallow the second-fastest-growing company in Europe in 2023, at a 660% compound annual growth rate, and a fresh raise of about $90M followed in 2025 at around $2bn. What the story teaches is easy to miss. The insight was not born in the gym. It came from a friend's complaint, taken seriously, and then acted on for ten months with a very small budget. ## Starling Bank (United Kingdom) URL: https://startup.talk/uk/starling-bank/ Founders: Anne Boden, Tom Blomfield Founded: 2014 (founded as Possible Financial Services; renamed Starling Bank 2016); Based in: London, England; Sector: Fintech, digital bank Anne Boden sold her house in Swansea to pay for staff. She had grown up in Bon-y-maen, a suburb of the city, the child of a steelworker and a department-store worker, and spent much of her childhood reading. At Swansea University she studied chemistry and computer science and graduated in 1981. She planned to work in IT. Instead she took a graduate trainee job at Lloyds, and a career in banking opened. It was a long one: Standard Chartered, where she earned a Middlesex MBA in 1990, then UBS, Aon, ABN AMRO and RBS, where she ran payments across 34 countries. In 2012 she became chief operating officer of Allied Irish Banks, helping it recover from the financial crisis. She knew the inside of a big bank better than almost anyone trying to replace one. In 2014 she founded Possible Financial Services, tagline 'Bank Possible'. The plan was a bank that lived on a phone. After the crisis she sold the Swansea house to fund hiring. The blow came in February 2015. Her co-founder and CTO, Tom Blomfield, left with four other directors and went on to start Monzo, a direct rival. A start-up that loses its tech chief and most of its board rarely survives. She rebuilt, renamed it Starling Bank, and won a banking licence in 2016. The account opened in beta in March 2017 and business accounts followed in 2018, the year she was made an MBE. Then came the price of growth. In February 2019 a GBP 75M round gave Harald McPike 60% of the company. Boden, who had put her own money in to start it, no longer held the majority. The bank itself thrived. By June 2022 it had three million accounts, including over 460,000 for businesses, and a mortgage book above GBP 2bn. For the year to March 2026 it reported revenue of GBP 887M and pre-tax profit of GBP 217M. The regulator fined it GBP 29M in October 2024 over financial-crime controls, after Boden had stepped aside as chief executive in June 2023 following a clash with investors. She kept a 4.9% stake. The lesson is double-edged. She built the thing she had imagined, and she gave up control of it to get it built. The number on the share register matters as much as the number on the balance sheet. ## Skyscanner (United Kingdom) URL: https://startup.talk/uk/skyscanner/ Founders: Gareth Williams, Barry Smith, Bonamy Grimes Founded: 2002 (launched 2003); Based in: Edinburgh, Scotland; Sector: Online travel search For two years, nobody at Skyscanner was paid. Gareth Williams, Barry Smith and Bonamy Grimes were IT contractors with day jobs. They built the company in spare hours, financing it themselves, mostly with unpaid labour. It was not until 2004 that all three had quit to work on it full time. The idea began with irritation. Williams wanted to plan trips, including visits to his brother in France, and every airline had its own website with its own prices. He wrote software to search many of them at once. He was born in Norwich and moved around in childhood, through Sussex, Switzerland and Canada. At Manchester he studied mathematics and computing, and in his first days there he met Grimes. Skyscanner launched in 2003, and the first office, in Edinburgh, opened in 2004. It was a long, thin stretch. Money was tight and the product was unfashionable. What almost nobody realised is how long they held out before taking outside money. Only in December 2007 did Scottish Equity Partners put in GBP 2.5M for a 40% stake. The founders had carried the company for most of five years. In 2009 it reported its first profit. Then it scaled. Offices opened in Singapore and Beijing, and in 2013 Sequoia Capital invested at an $800M valuation. In January 2016 a GBP 128M round put the value at $1.6bn. In November that year, Ctrip, now Trip.com Group, bought the company for about $1.75bn. The cost arrived in 2020, when COVID wiped out travel. Skyscanner cut 300 jobs, about 20% of staff, and closed its Budapest and Sofia offices. By 2025 it had recovered to revenue of GBP 540.6M and net income of GBP 56.9M, under a new chief executive, Bryan Batista. The lesson is in the sequence. The three held on to their time and their ownership while the product was still small. By the time investors arrived, they were asking to join something that already worked. ## Raspberry Pi (United Kingdom) URL: https://startup.talk/uk/raspberry-pi/ Founders: Eben Upton, Liz Upton Founded: 2008 (Foundation); 2012 (trading company); Based in: Cambridge, England; Sector: Single-board computers and embedded hardware Eben Upton spent his days deciding who got into Cambridge to study computer science. He was good at it, and it worried him. Upton was director of studies in computer science at St John's College, the person who reads the applications. Year after year the pile was getting thinner, and the applicants who did arrive had less hands-on skill than the ones a decade earlier. The kids who once learned to program on the BBC Micro in their bedrooms now had laptops that were hard to tinker with. He came to this problem from an odd angle. Born in Griffithstown, near Pontypool, he lived in Lae, Papua New Guinea, as a small child, then grew up in Leeds, Birmingham and Ilkley. His father, Clive Upton, is a linguist, and the two later wrote the Oxford Rhyming Dictionary together. Eben went to St John's himself, took physics and engineering, then a PhD in the Computer Laboratory. He met his future wife, Liz, at university. She would become a co-founder. In 2008 a group around him set up a charity, the Raspberry Pi Foundation, to bring back the spirit of the BBC Micro. The "Model A" and "Model B" names were a homage to it. The first prototypes looked like USB sticks. By August 2011 they had fifty alpha boards, and early the next year ten of the beta boards were auctioned on eBay. What almost nobody outside the project expected was the response. On 29 February 2012 the Model B went on sale at $35. The websites of the two launch distributors, Premier Farnell and RS Components, crashed. One reported over 100,000 pre-orders on the first day. A charity for Cambridge admissions had accidentally started a global hardware business. Upton left the Foundation's board in December 2012 to run a new trading company that could build and sell the boards at scale. Production moved from China to a Sony plant in Pencoed, Wales. There was a quiet irony in it: the machine meant for classrooms was soon wanted by factories. About 70% of 2024 sales went to industrial customers. The company listed on the London Stock Exchange in June 2024, and the Foundation's stake dropped from 77.31% to 49.08%. It reported 2025 revenue of US$323.2 million and net income of US$21.7 million. More than 75 million boards had sold by March 2026. The lesson sits in the first paragraph. Upton did not set out to build a company. He had a worry that he could not stop seeing in his own inbox every admissions season, and he built the smallest thing that might fix it. The business came after the worry. ## Blue Prism (United Kingdom) URL: https://startup.talk/uk/blue-prism/ Founders: Alastair Bathgate, David Moss Founded: 2001; Based in: Warrington, England; Sector: Robotic process automation / enterprise software In March 2016 Blue Prism rang the bell on London's AIM market at a value of £48.5 million. Its shares jumped 44% on the first day. For Alastair Bathgate and David Moss it was fifteen years in the making. The two had founded the company in 2001 in Warrington, far from the glass towers of the City. In the early years it built software for the back offices of financial firms, work like debt recovery and settlement, where people sat in front of screens copying data between systems. The first commercial product, Automate, arrived in 2003. Nobody was yet calling it a category. That is the long, quiet middle of this story. For years the company's pitch had no ready-made phrase to hang on. Blue Prism's own history page says it coined the term robotic process automation in 2012, a decade after founding. Once the label existed, buyers could finally search for it, and the firm that had been building it all along was suddenly early rather than strange. The listing gave it money and visibility. By November 2016 it had offices in Chicago, Miami and the UK, and in January 2017 it opened in Austin, Texas with 86 staff worldwide. In 2019 it raised new capital and bought Thoughtonomy for up to $100 million, adding an AI engine to the platform. Then the founders' era ended. Jason Kingdon, who had returned as chairman in 2019, took over as CEO in April 2020 as Bathgate stepped down. The company reported 2020 revenue of £141.4 million and 1,005 employees. What almost nobody forecast was the price war for the company itself. In September 2021 Vista Equity Partners agreed to buy it for £1.095 billion, planning to fold it into TIBCO. SS&C Technologies then bid against Vista, and shareholders backed SS&C. The deal closed on 16 March 2022 at about £1.25 billion ($1.6 billion), more than twenty times the value at listing. It now runs as SS&C Blue Prism, and in April 2026 it launched WorkHQ, a platform that coordinates AI agents, software robots and people. The founders' public record is thin on feelings, so the shape of the story has to speak. Two men in Warrington kept building through years when the market had no name for what they made. The lesson for a first-time founder is about patience: the category you are early to may take a decade to be called something, and the companies still standing when it is named get to define it. ## BVNK (United Kingdom) URL: https://startup.talk/uk/bvnk/ Founders: Jesse Hemson-Struthers, Donald Jackson, Chris Harmse Founded: 2021; Based in: London, England; Sector: Stablecoin payments infrastructure (fintech) On 12 November 2025 a deal worth about $2 billion fell apart. Five months later the same company was sold for nearly as much. BVNK is the stablecoin payments firm that Coinbase had been negotiating to buy. In October 2025 Fortune reported that both Coinbase and Mastercard were in advanced talks to acquire it. BVNK entered exclusive negotiations with Coinbase. Then Coinbase ended them. The founders are three South Africans who built the company in London: Jesse Hemson-Struthers, the chief executive, a serial entrepreneur who had run online businesses at home; Donald Jackson, the technologist, who had founded Cue, a bootstrapped customer engagement platform, and Verity, a fraud reduction service; and Chris Harmse, a chartered financial analyst and FX trader who had been a partner at a macro and crypto fund. One built things, one wrote the code, one understood money. They started BVNK in 2021 with a plain observation. Companies in Africa, Europe and elsewhere wanted to settle payments in stablecoins, which move around the clock and cost far less than the card and bank rails. But few could handle the wallets, compliance and conversion between currencies on their own. BVNK offered that as infrastructure. Investors noticed. In December 2024 it raised a Series B at a valuation of about $750 million, with Haun Ventures among the backers. A year later the valuation conversation had moved to $1.5 to $2.5 billion. The Coinbase collapse could have been the end of the momentum. Instead, on 17 March 2026, Mastercard announced it would acquire BVNK for up to $1.8 billion, with $300 million of that tied to contingent payments. The deal is expected to close by the end of 2026. Harmse's reaction, reported at the time: 'Needless to say, there's a big smile on my face.' The cost of that smile is not small. A failed exclusive deal tells every competitor and customer that a company is for sale, and it leaves a young firm exposed. The founders have said little about it in public. What they did show is that a deal's death is not a company's death. If you build the thing the biggest players need, a second bidder is already in the room. The work that matters is the work done before the offer arrives. ## Granola (United Kingdom) URL: https://startup.talk/uk/granola/ Founders: Chris Pedregal, Sam Stephenson Founded: 2023; Based in: London, England; Sector: AI meeting notes / productivity software For a full year, Chris Pedregal sat next to strangers and watched them install his software. There were 150 of them. He would sit beside each user while they set up Granola, note where they hesitated, go home, fix the problem, and come back the next day to watch another one try. He told the Silicon Valley Girl podcast that this was how he found out what was wrong. Pedregal had done the founder thing before. He built Socratic, an education app that Google acquired in 2018, and stayed on inside Google. By the time he started Granola with Sam Stephenson in 2023, he knew what it was like to ship a product with enormous reach. He also knew what a crowded field looked like, because the meeting-notes market was already full of bots that joined calls and unsettled the people on them. His choice was to build something quieter. Granola listens from the user's own device, so no bot appears in the meeting. It lets you type rough notes and then merges them with an AI-cleaned transcript. He later told Bloomberg the market is commoditised and oversaturated, and that he would never have entered it if the goal were just to generate meeting notes. The patience is the unusual part. Pedregal delayed the public launch until the product was, in his words, 'meaningfully better than the competition.' When it opened, it drew 500 installs on the first day with no advertising. Enterprise customers followed the organic growth. The money came fast once it did. The Series B in May 2025 raised $43 million at a $250 million valuation. In March 2026 a Series C of $125 million, led by Index Ventures with Kleiner Perkins, set the valuation at $1.5 billion, roughly six times what it was ten months earlier. Pedregal said revenue had grown 2.5 times since January, though he did not give a dollar figure. The risk is plain: giants like Zoom and Google can bundle note-taking into the tools people already use. Granola's answer is to turn notes into a base for agent-style features that act on what was said in the meeting. What Pedregal's year of watching suggests is simple. Behind a billion-dollar price tag is a founder who was willing to sit in a stranger's chair and be confronted, a hundred and fifty times, with the fact that his product still wasn't good enough. ## Tide (United Kingdom) URL: https://startup.talk/uk/tide/ Founders: George Bevis, Oliver Prill Founded: 2015; Based in: London, England; Sector: Business banking and fintech for small businesses In July 2020 Tide shut the door on small businesses that wanted government-backed loans. The money had run out. The company had lent more than £50 million to nearly 2,000 firms under the Bounce Back Loan scheme, during the first months of the pandemic. Because Tide was not a bank, it could not use deposits to lend and did not qualify for the Bank of England's term funding scheme. It relied on investors' money, and when that was used up, it stopped taking applications. It kept asking the government for access. That constraint tells you what Tide is. It started as Starfish Platform Limited, renamed and registered as Tide Platform Limited on 18 May 2015. Its founder was George Bevis, a former banker who knew what small-business customers disliked about branches: queues, forms and the wait. Tide won FCA permission in February 2016, raised a $2 million seed that July from Passion Capital, LocalGlobe and angels including Zoopla's Alex Chesterman and Wonga's Errol Damelin, and launched its app in January 2017. The product was built to sit inside the tools sole traders already used. It was one of five fintech firms to launch Xero's bank-feed integration, and it partnered with PayPoint and the Post Office so cash could be deposited. Tide Credit, with iwoca, offered lending up to £150,000 from October 2017. In March 2018 Bevis said he would step down as CEO. In August Oliver Prill, formerly COO at Kreditech, took over, and Eileen Burbidge became chair. The company had to prove it could grow without its founder at the front. It also had to fight for fuel. In 2019 it received £60 million from the RBS remedies fund, through a partnership with ClearBank, which competitors criticised. A £44.1 million Series B led by SBI Group and Augmentum Fintech followed in October. A $100 million Series C in 2021 valued it at $650 million. In September 2025 a $120 million share sale involving TPG and Apax put it at $1.5 billion, with reports of around 1.6 million members across the UK, India and Germany, and CB Insights listing FY2024 revenue of $222 million. The lesson is in the gap between 2020 and now. A company that was a brake-pedal away from the edge of a government scheme learned what it is to build on other people's balance sheets. For a founder, the question is not just what your customers want, but whose money you will be allowed to lend. ## Matillion (United Kingdom) URL: https://startup.talk/uk/matillion/ Founders: Matthew Scullion, Ed Thompson Founded: 2011; Based in: Altrincham, Greater Manchester, England; Sector: Cloud data integration, data pipelines Before Matillion had a customer, it had one engineer and a decision. In 2011 Matthew Scullion and Ed Thompson quit their jobs. Thompson went to work in a rural corner outside Manchester and wrote the first version of the product himself. Scullion set about writing down the principles the company would live by. There was no office tower, no Silicon Valley sponsor, just two people betting that data was about to move to the cloud. Scullion was not new to the bet. By his own account he had co-founded his first startup at 18, which meant that by the time Matillion began he had already been through the cycle of building something, running out of road and starting again. Thompson was the quiet half: the builder, and later the company's chief technology officer. What almost nobody outside the data world noticed was how awkward the job was. Companies were piling information into cloud warehouses, and then discovering that getting it in, cleaning it and shaping it into something useful took armies of people. Matillion's answer was software that did that work inside the warehouse itself. Growth was patient. A first outside round of about $5M arrived in 2016. In 2018 came $20M, with a Silicon Valley investor, Scale Venture Partners, joining the Manchester-area YFM. A further $35M or so followed in 2019. Offices opened in Denver, New York and Seattle, because most of the buyers were American. Then, in September 2021, General Atlantic led a $150M Series E with Battery Ventures and others, valuing the company at $1.5bn. It made Matillion one of the UK's unicorns, and one headquartered in Altrincham rather than London. Scullion's reaction, in a later interview, was telling. The founders, he said, had never set out to become a unicorn. They had set out to do business with integrity and quality, and to take some risks. The headline on that interview quoted the line that they never had a vision this big. That is the quiet lesson of the story. Scullion and Thompson did not start with a billion-dollar ambition. They started with a principle, a working product and a market that was moving their way. The valuation was what happened after fifteen years of not wandering off. For a founder reading this, the point is to write the first version yourself, write the rules down early, and let the scale come later. ## PPRO (United Kingdom) URL: https://startup.talk/uk/ppro/ Founders: Philipp Nieland, Tobias Schreyer Founded: 2006; Based in: London, England; Sector: Fintech, cross-border and local payment methods infrastructure Most people have used PPRO without knowing it. When a shopper in Germany pays by bank transfer on a US website, or someone in the Netherlands picks iDEAL at a checkout, the plumbing underneath is often the kind of thing PPRO built. It does not have an app or a famous logo. That is exactly the design, and it is also why the company is so little known. The company dates to 2006. Philipp Nieland and Tobias Schreyer are the founders named in company records and profiles, and they started from an observation that looks obvious now: outside the English-speaking world, people often do not pay by credit card. A merchant who accepts only cards loses customers in country after country. Each local method, though, needs its own bank links, rules and licences. PPRO's job was to take that mess and offer it as a single connection. For years the company grew quietly. Then the early online-payments world changed shape, and the investors arrived. HPE Growth was first, backing the company before 2015. That year a new chief executive came in from outside the founding team: Simon Black, who had spent about a decade at Sage and had run Sage Pay, the UK's biggest payment gateway for small business. What the founders had built and what Black added were different skills. One was a product that worked across borders. The other was a way to scale it. In 2018 PPRO raised $50M with PayPal and Citi Ventures involved. In January 2021, in the middle of the pandemic e-commerce boom, it raised $180M from Eurazeo and Wellington, with PayPal, Citi and JP Morgan alongside, and crossed into unicorn territory at a $1bn valuation. Black had by then led rounds totalling over $300M. Then the leadership changed again. Black announced he would step down at the end of 2023, after nearly nine years. A further round of about $93M was reported in 2024. PPRO's story is a reminder that not every great company has a single heroic founder at the centre. Two founders found the problem, an outsider helped industrialise it, and the rest was patience across borders. Few consumers will ever see its name, and that suits it. If you are building infrastructure, the lesson is simple: the less the customer sees you, the more carefully you have to be trusted. ## Paddle (United Kingdom) URL: https://startup.talk/uk/paddle/ Founders: Christian Owens, Harrison Rose Founded: 2012; Based in: London, England; Sector: SaaS payments, merchant of record, billing Harrison Rose started his first business at 18. Paddle was not it. By the time he and Christian Owens incorporated Paddle in August 2012, Rose had already learned the unglamorous parts of selling things online. That head start shaped what they built: a way for software makers to sell their products without becoming experts in payments. The early Paddle was a marketplace for software. Developers could list a product and Paddle would do the selling and the paperwork. It worked as an idea, but it was a narrow one. Rose has said that in those early years the company raised funds without really understanding how the venture world worked, and that they learned by making mistakes. What almost nobody noticed was the quiet pivot. Paddle moved from being a marketplace to being payments infrastructure, a seller of record that takes legal responsibility for sales tax, VAT, fraud and refunds. A developer in London could sell to a customer in Brazil, and Paddle would handle the compliance. It also changed how it sold. Early growth leaned heavily on outbound sales. Over time the founders put money into brand, and the message shifted from the founders' own story to what the service meant for enterprise buyers. In May 2022, a $200M Series D led by KKR, with FTV Capital, Silicon Valley Bank, Kindred, Notion and 83North, valued the business at $1.4bn. Within weeks Paddle used part of the money to buy ProfitWell, the subscription-metrics firm, for more than $200M in cash and equity. By March 2024 the company had about 300 employees and 3,000 customers worldwide, and had raised roughly $300M in total. There was a human cost to that growth. Rose later stepped back from the company he co-founded and went on to start another business, Good Fit. Owens, the other founder, is listed as a former chief executive. Money was only part of it. A team of about 300 had to run tax and compliance for thousands of customers in many countries, which is a different job from writing the first version. The pattern is familiar and useful. Founders who start young often build the first company to learn how companies work, and the second to put that knowledge to use. Paddle's earliest model was not its final one. What mattered was that the founders were willing to change it. ## Depop (United Kingdom) URL: https://startup.talk/uk/depop/ Founders: Simon Beckerman, Maria Raga Founded: 2011; Based in: London, England; Sector: Social marketplace, resale fashion Depop began as a shopping page for a magazine. Simon Beckerman, born in Milan, had spent years publishing PIG, a quarterly about emerging people in fashion, music and design. Readers kept asking where they could buy the clothes in its pages. His first idea, as he later described it, was a shop that would sell everything the magazine featured. Beckerman was no stranger to starting things. In 2006 he and his brother launched a glasses brand. Before that he had immersed himself in industrial design and the arts. None of it made him rich, but each project taught him a skill he would later need. Publishing, he told an interviewer in 2019, taught him how to tell stories, talk to an audience and use influencers. In 2011 the idea took shape at H-Farm, an Italian incubator near Venice. He saw that eBay, Etsy and Gumtree served everyday utility, not the underground communities he knew. He wanted a place for like-minded people, and described Depop as a rebellious little kid always looking for what is exciting. Money came in small pieces, and none of it arrived all at once. A seed from Paolo Barberis and Nana Bianca in 2012, then EUR 1M in 2013 from Red Circle, which came with a new chief executive, Runar Reistrup. In 2015 Balderton and HV put in $8M. Maria Raga, who had joined as operations chief in 2014, took over as CEO in 2016. Growth hit its stride in 2019, when General Atlantic led a $62M Series C and a Depop store opened inside Selfridges in London. By 2020, gross sales more than doubled to $650M and revenue to $70M. The same year brought ugly problems. Scammers pushed buyers to pay by direct bank transfer, outside the platform's protection, and some buyers were tricked in October 2020 and had their details sold on the dark web. Depop promised mandatory password updates and multi-factor authentication. Etsy bought Depop in 2021 for $1.6bn. In February 2026 Etsy agreed to sell it to eBay for $1.2bn, and the deal closed in July 2026. Depop sits today inside a larger marketplace group, but it still keeps its own name and identity, which says something about how strongly its community felt about it. Beckerman's lesson is that a side project to serve readers can become a business. He just had to notice what his audience kept asking for. ## Trainline (United Kingdom) URL: https://startup.talk/uk/trainline/ Founders: Virgin Group, Clare Gilmartin, Jody Ford Founded: 1997; Based in: London, England; Sector: Rail and coach ticketing, travel technology In the year to February 2021, Trainline sold GBP 783M of tickets. The year before, it had sold GBP 3.7bn. That collapse is one of the most vivid moments in the company's life, but it is not where the story starts. Trainline began in 1997, set up by Virgin Group, with online sales opening in 1999 under contract with Capgemini. Stagecoach later bought a 49% stake. The next move was blunt. In February 2004 Trainline merged with its main rival, Qjump. Stagecoach sold out, leaving Virgin with 86% and National Express with 14%. The idea survived the merger: one site that could sell tickets from every train company. Then the owners changed. Exponent Private Equity bought in July 2006. In 2007 it acquired Advanced Smartcard Technologies, including Ecebs, which was sold in 2012. In January 2015 KKR bought it, scrapping a planned IPO, and the business shed the name thetrainline.com for plain Trainline. A year later it bought Captain Train in France and became Trainline EU. In June 2019 the company listed in London at 350p a share, valuing it at nearly GBP 2bn on the first day. What almost nobody predicted was the speed of the fall. COVID-19 emptied the stations, and the company recorded an operating loss of about GBP 100M. In 2021 Jody Ford took over as chief executive from Clare Gilmartin, who had led it through the listing. The recovery was real, and it was built on the same basic promise: one screen, every operator. For the year to 28 February 2026, revenue was GBP 452.7M, operating income GBP 122.4M and net income GBP 79.8M, with 986 staff. The company now aggregates tickets from more than 270 operators in over 40 countries, and in April 2025 bought Spanish ticketing company Trenes.com. New trouble has not gone away. In June 2025 its AI assistant was criticised for giving wrong information in beta, and in August 2026 the Competition and Markets Authority opened an investigation into alleged drip pricing. Behind the app sits the platform Platform One, which also supports booking and distribution for other rail and coach operators, so the company sells technology as well as tickets. Trainline shows that persistence can matter more than inspiration. It took four owners, a pandemic and a name change to become the default way to buy a train ticket. ## Humanoid (United Kingdom) URL: https://startup.talk/uk/humanoid/ Founders: Artem Sokolov Founded: 2024; Based in: London, England; Sector: Humanoid robotics and embodied AI In August 2025 Artem Sokolov sold the thing his family had spent three decades building. The jewellery brand that carried his surname, and the chain of shops behind it, went to the Russian investor Anton Pak, according to the newspaper Kommersant. The terms were not disclosed. He had grown up inside that business. Born in Russia, he was the son of the couple who started the Sokolov jewellery brand in 1993, and he later became its chief executive. Four years before the sale he had even discussed a joint New York and Moscow listing aimed at raising US$500 million, Reuters reported. A public offering was the obvious next chapter for a family company like his. Instead he bet on something with no jewellery in it at all. In 2024 he founded a robotics company in London, known by its legal name SKL Robotics and by its brand, Humanoid. He is listed as its sole owner, and Forbes Argentina reported that he put in about US$30 million of his own money before anyone else wrote a cheque. That is a founder choosing to be the first and only investor in a field where rivals such as Tesla and Figure AI had a long head start and far deeper pockets. What the sale of the family firm bought him was time. The early months were visible mostly through videos. In January 2026 a Humanoid robot tried, a little clumsily, to move a box for Siemens, and the clip made the rounds. The company built two bodies for its robots, a wheeled one for sorting and carrying, and a bipedal one for cramped, cluttered spaces, together called HMND 01. The part that ties them together is KinetIQ, software that lets several robots share one job and hand work to each other. The turn came in May 2026, when Humanoid signed a binding deal with the German industrial group Schaeffler to deploy between 1,000 and 2,000 wheeled units across its plants by 2032. A real customer with real volume changed the conversation with investors. In July, Prime Movers Lab led a US$152 million Series A at a US$1.35 billion post-money valuation, with Schaeffler, Bosch, Fubon Financial Holding Venture Capital and Aglae Ventures joining. Total funding stood near US$270 million. Bosch is to act as contract manufacturer, which matters because the hard problem now is building thousands of robots, not one. Sokolov called the company Europe's first pure-play humanoid unicorn, reached in about two years. The human lesson is in the order of events. He sold the safe, inherited asset first, and only then asked others to back the new one. Investors tend to believe a founder more when the founder has already paid for the idea himself. ## Gymshark (United Kingdom) URL: https://startup.talk/uk/gymshark/ Founders: Ben Francis, Lewis Morgan Founded: 2012; Based in: Solihull, England; Sector: Fitness apparel and direct-to-consumer retail Ben Francis spent his evenings between five and ten delivering pizza for Pizza Hut at £5 an hour. After his shift he opened a laptop and worked on a gym clothing brand. By day he was a student at Aston University. That schedule is the part of the Gymshark story that rarely makes the posters. Born in the West Midlands and raised in Bromsgrove, he found computers at school and the gym as a teenager. Before any of it, he had already run a small online business selling car number plates. He enrolled at Aston in 2010 to study international business and management. The first version of Gymshark did not even make clothes. He and his school friend Lewis Morgan sold bodybuilding supplements, dropshipped from others. It took about six weeks to get the first sale. What bothered Francis was what he could not buy: gym wear that fitted the way lifters wanted it to. So he made it himself. He took £1,000 of savings and bought a sewing machine and a screen printer, and set up in his parents' garage. He reportedly learned to sew from his grandmother. The first year, the brand turned over about £500 a day. What almost nobody saw coming was how fast the garage outgrew itself. At the 2013 BodyPower show the stock sold out on day one. The Luxe tracksuit later brought in £30,000 within thirty minutes. When revenue passed £250,000, both founders walked away from university, and Francis soon quit the pizza rounds too, about six months after launch. The cost showed later. In 2016 Morgan partly stepped back, holding on to a 20% stake. Francis kept the controlling share and the weight of the company. A brand that started with a male focus found that by 2020 about two thirds of sales were to women. In August 2020 General Atlantic paid £275 million for 21%, valuing Gymshark at £1.25 billion. It opened its first permanent store on London's Regent Street in 2022. Revenue reached £646 million in 2025, but growth had a price: in April 2025 a restructure put almost a third of the workforce at risk of redundancy, and legal fights with influencers added noise. Francis, now an MBE, was reported in July 2026 to be in talks to buy back some of the shares sold in 2020. The lesson is in the evening shifts. Francis did not bet everything on an idea. He funded it with a delivery job and £1,000, and he left both only when the sales proved him right. ## 9fin (United Kingdom) URL: https://startup.talk/uk/9fin/ Founders: Steven Hunter, Huss El-Sheikh Founded: 2016; Based in: London, England; Sector: Debt capital markets intelligence and AI analytics Steven Hunter did not need the money. When 9fin raised US$170 million in March 2026, he said it was an opportunistic raise and that the company had not needed the capital. It is an unusual sentence from a founder at a unicorn, and it makes sense once you see how the company began. Hunter studied law at the University of Bristol and graduated with first class honours. He did not become a lawyer. He went into leveraged finance at J.P. Morgan in London, then moved to Babson Capital Management, where he covered the European high yield bond and leveraged loan markets. His working day was spent chasing information about companies in debt. That is where the itch came from. He and Huss El-Sheikh, a technologist who had worked in engineering at a large bank, saw the same problem from different sides. The people who decided where billions in credit went relied on emails, PDFs and phone calls to learn what was happening to a borrower. The data existed, but it was scattered. In 2016 the two started 9fin to fix that. The early product gathered and organised information on bonds, loans and distressed companies for the credit professionals who needed it faster than the incumbent services offered. Backers included Seedcamp, Redalpine and Spark Capital, and a small Series A of US$8 million funded the first move into the United States. What almost nobody outside finance noticed was how much of the work could be automated. Hunter's team built tools that read filings and summarise them, and he has said AI will redefine the credit markets. Growth followed: the company reports several years in a row of 100% annual recurring revenue growth, and more than 300 banks, asset managers, law firms and advisers now use the platform. In December 2024 Highland Europe led a US$50 million Series B, and its co-founder Fergal Mullen joined the board. Kerry McClelland was promoted to chief financial officer that month, a sign the company was being built for scale. The Series C in March 2026 was led by HarbourVest, with CPP Investments and earlier investors joining, at a valuation reported as about US$1.3 billion. Total funding passed US$250 million. Coverage noted that 9fin is among the few European unicorns with Black leadership. Hunter's route holds a quiet lesson. He was a lawyer by training and a banker by trade, and neither title became the company. The company came from sitting at a desk, noticing how clumsy the work was, and finding the one person who could help him build the fix. When the money finally arrived, it was a choice and not a rescue. ## Nothing (United Kingdom) URL: https://startup.talk/uk/nothing/ Founders: Carl Pei, Akis Evangelidis, David Sanmartin Garcia, Steven Gao, Levi Li Founded: 2020; Based in: London, England; Sector: Consumer electronics (smartphones, audio) Carl Pei built a fan site for a Chinese phone maker he had no connection to. He was a student in Sweden and the company was thousands of miles away. The people at Meizu noticed, and in 2011 they offered him a job in marketing. Pei was born in Beijing in 1989. His family moved to the United States and then to Sweden, where he grew up and became a Swedish citizen. In 2008 he enrolled in a Bachelor of Science program at the Stockholm School of Economics. In 2010 he interned at Nokia. A year later he left university and moved to China to work full time in smartphones, because the fan site had shown him where the energy was. In November 2012 he joined Oppo as its international markets manager, reporting directly to Pete Lau. A year later they started OnePlus together in Shenzhen. The OnePlus One reportedly sold close to a million units in 2014 against a target of 50,000. In 2015 Pei presented the OnePlus 2 in a virtual reality video, billed as the first product launch of its kind. He described the global team as a startup inside a larger firm. He left OnePlus in October 2020. By January 2021 he had announced Nothing, a London company co-founded with Akis Evangelidis, David Sanmartin Garcia, Steven Gao and Levi Li. Seed money of US$7 million came from names such as Tony Fadell, Kevin Lin, Steve Huffman and Casey Neistat. The first product, the Ear (1) earbuds, arrived in July 2021 at US$99. The first phone followed in 2022 with Glyph lights on its transparent back. It had software problems at launch, and the update policy drew criticism. In November 2023 an iMessage-style app, Nothing Chats, was pulled from Google Play within a day after security flaws were found. In August 2025 demo images on Phone (3) display units turned out to be licensed stock photos, which Evangelidis called an unfortunate oversight. What almost nobody expected was the strength of the cheaper models and of India. By 2024 annual revenue had roughly doubled to over US$500 million and more than seven million devices had sold. Tiger Global led a US$200 million Series C in September 2025 at a valuation of US$1.3 billion. By May 2026 lifetime revenue had passed US$2 billion, with about 900 staff. The first spark was never a company at all. A student made a website because he loved a product, and someone paid him for it. Pei's career has kept returning to that instinct, to the pull of a fan. ## ASOS (United Kingdom) URL: https://startup.talk/uk/asos/ Founders: Nick Robertson, Andrew Regan, Quentin Griffiths, Deborah Thorpe Founded: 2000 (incorporated 3 June 2000 as Winsupply plc); Based in: London, England; Sector: Online fashion retail The first product that ASOS tried to sell was a jacket from a film. Brad Pitt's red leather jacket in Fight Club had been a talking point in 1999, and an online shop that would sell you the look was an idea that fit the internet at the time. The company did not begin as ASOS. On 3 June 2000 it was registered in London as Winsupply Public Limited Company. Within two weeks it became AsSeenOnScreen Holdings, and the name was the pitch: buy what you see on film and TV. Four people are named as its founders: Nick Robertson, Andrew Regan, Quentin Griffiths and Deborah Thorpe. Robertson, who went to Canford School in Dorset, became chief executive and held the job until 2015. The first model was close to a fan business. The site sold imitations of the clothes celebrities and characters wore, and shoppers could browse by film or show. It reached the Alternative Investment Market in 2001, which put the young firm in front of public investors. In 2003 shareholders approved a shorter name, ASOS, and the film-and-TV gimmick faded as the catalogue grew. The turn came in 2004, when ASOS reported its first profit and launched its own women's label. Selling its own designs was the step from imitation to brand. It also set a pattern: ASOS kept adding third-party labels on one site while building its own. It was not smooth. In 2005 the Buncefield fuel depot explosion shut the business for six weeks. In 2010 it dropped kidswear when branded sales declined. In 2013 a batch of belts was recalled after contamination with radioactive cobalt-60. A warehouse fire in 2014 stopped order-taking for almost three days. A 2016 report by BuzzFeed News alleged poor warehouse conditions, which the company disputed. The business also sponsored the McLaren Formula One team in 2014 and launched a visual search tool, Style Match, in 2017. Robertson's wealth became public in 2016 when a High Court judge ordered him to pay his ex-wife £70 million, about a third of a fortune put at £220 million. The hard years came later. The share value fell 70% during 2023, and a £3.95 return fee angered customers in 2024. For FY2025 ASOS reported revenue of £2,464.8 million, an operating loss of £32.2 million and a net loss of £298.4 million. Frasers Group holds 29.26% and Bestseller 26%. The founding lesson sits in the first name. Winsupply, then AsSeenOnScreen, then ASOS: the company changed its idea twice before it found one that scaled. A first idea is a place to start, not a place to stay. ## Castore (United Kingdom) URL: https://startup.talk/uk/castore/ Founders: Tom Beahon, Phil Beahon Founded: 2015; Based in: Manchester, England; Sector: Sportswear / team kit In 2013 two brothers from Merseyside did something most young athletes never do. They walked away from the one thing they had trained for since childhood. Tom Beahon had spent four years in Tranmere Rovers' youth system, then played in Spain, then drifted through the Glenn Hoddle Academy. Phil, three years younger, had played semi-professional cricket for clubs in Cheshire and Lancashire. Neither was going to be a star. So they quit, and took the dullest jobs they could find. Tom went to Lloyds Bank in London. Phil went to Deloitte. This is the detail that rarely makes the headlines: the brothers who would one day sell kit to Rangers and England cricket began as two junior finance staff, whose real job was to collect money and contacts for an idea. The idea was simple, and a little stubborn. Sportswear they wore, they felt, was either cheap-looking or built for pros and priced like a fashion label. Before launching, they interviewed people in gyms and went looking for investors from both fashion and sport. In July 2015, aged 25 and 22, they set up Castore in Bebington, on the Wirral, and launched online in 2016. What nobody on the outside saw were the small years. Sportswear is a brutal trade, where Nike and Adidas own the shelf space and the stadium tunnels. Castore had to make clothes people believed in before it could afford sponsorships. The turn came through athletes. Andy Murray became a shareholder in March 2019 and the first major face of the brand. Forbes put the brothers on its 30 Under 30 list that year. In May 2020 came Rangers, a five-year deal reported at around £25m, and Castore stopped being a gym label and became a kit maker. Rich backers followed, among them the Issa brothers' Monte Group, Ineos and the founders of Redrice. Each brother held 18.59 per cent in 2021. The price of growth was public. Fans complained about Rangers kit confusion. Aston Villa players said their shirts grew heavy when wet, and Villa left early for Adidas in January 2024. A company that sells on credibility lives or dies by the feel of a fabric. By the 18 months to August 2025 Castore reported revenue of about £334.6m and a loss of £40.3m, and in autumn 2026 it was reported to be weighing a sale with JPMorgan advising. Tom has framed the losses as the cost of investing for scale. The lesson is in the opening move. The Beahons did not romanticise the dream of playing. They treated finishing it as the first piece of fundraising, and borrowed two years of bank and accountancy discipline before they spent a pound of anyone else's money. ## GoCardless (United Kingdom) URL: https://startup.talk/uk/gocardless/ Founders: Tom Blomfield, Hiroki Takeuchi, Matt Robinson Founded: 2011; Based in: London, England; Sector: Fintech / bank payments The first time Tom Blomfield met Silicon Valley's best-known startup school, he was not in the room. His partners in an Oxford student venture, Boso.com, went through Y Combinator. He stayed behind to sit his exams. Blomfield was born in Hong Kong in 1985 and grew up in Buckinghamshire, attending Dr Challoner's Grammar School in Amersham. Even as a schoolboy he was selling: he built a website for a local estate agent and charged £250 plus £3 for each house listed. Then he went to Oxford to read law. Boso.com was meant to be an eBay for students. It struggled to raise money, and the YC trip happened without him. It was eventually sold to a Canadian company for what he has described as several million dollars, and became Auctomatic. A success, on paper. But the wound is easy to see: the programme he skipped for a degree became the thing he later chased. So in January 2011, with Hiroki Takeuchi and Matt Robinson, also Oxford students, he started again with a duller problem. Collecting money from customers every month is a pain for small businesses. Cards fail, invoices are chased, direct debit is available mostly to large firms. GoCardless would let anyone do it through code. This time he did go. Y Combinator invested £125,000, and the doors opened: about £1.5m in 2012 from YC, Google Ventures, Salesforce and Accel. What almost nobody knows is how slow the early years were compared with the later glory. Blomfield left after about three years, having raised roughly £35m and hired around 100 people, to start Monzo in 2015. GoCardless kept compounding without him: UK launch in 2013, Europe in 2014, the US in 2018, a unicorn round in 2020, and $312m at a $2.1bn valuation in 2022. The ending was quieter than the rise. In December 2025 Dutch payments firm Mollie agreed to buy GoCardless for about €1.05bn, mostly in stock, with closing expected by mid-2026. The company then served around 100,000 businesses. For a first-time founder the lesson is in the exam he chose. Missing the programme once did not close the door. The second company worked partly because he had already failed to get through the first. ## Allica Bank (United Kingdom) URL: https://startup.talk/uk/allica-bank/ Founders: Richard Davies Founded: 2019 (banking licence; incorporated earlier as CivilisedBank); Based in: London, England; Sector: SME banking When Richard Davies arrived at Allica Bank in July 2020, he has said, the bank had made a couple of loans. That was it. A full banking licence, which takes most founders years of regulatory pain to win, and almost nothing behind the door. Davies had just left Revolut, where he was chief operating officer, and before that he had been the first CEO of OakNorth, where he helped shape the business model and licence application. Moving from the most famous consumer fintech in Europe to a quiet licensed shell looked, from the outside, like a step down. But the licence was the prize. Allica had received its PRA banking licence in 2019. Some profiles say the company began life years earlier under a different name, CivilisedBank, though that detail is thinly sourced. What mattered was that someone had cleared the regulatory hurdle, and that nobody had yet built a business on top of it. Davies chose the narrow target: established small and medium-sized firms. Not startups, not sole traders, but companies a bank branch used to know. Sept 2020 brought £26m led by Warwick Capital Partners. In November 2021, Allica announced the purchase of Allied Irish Banks' British SME book, about £600m in loans to roughly 2,000 customers, completed in August 2022. A bank that had made a couple of loans now had a book of lending. More money followed: £110m in 2021, £55m and £100m in 2022 with TCV joining. The business current account arrived in January 2023, and by year end had about 2,000 customers. Then the rankings began to arrive: the UK's fastest-growing fintech in Deloitte's Fast 50 in 2023, the fastest-growing private company in The Sunday Times 100 in 2024. In 2025 it reported revenue of £371.3m and net income of £36.9m, with over 30,000 business customers by early 2026, when a $155m round made it a unicorn at $1.2bn. The quiet closing lesson: a licence is not a company. Davies took the boring, regulated, unglamorous asset that others had walked past, and spent the five years proving that the first loan matters less than the second thousandth. ## Charlotte Tilbury Beauty (United Kingdom) URL: https://startup.talk/uk/charlotte-tilbury-beauty/ Founders: Charlotte Tilbury Founded: 2013; Based in: London, England; Sector: Beauty / cosmetics At thirteen, Charlotte Tilbury was sent back to England alone. She had lived nearly all her life in Ibiza, where her parents, an artist father and a production-manager mother, had taken the family when she was nine months old. Now she was a boarding pupil at a Rudolf Steiner school in Sussex. That is where the makeup began. In a school with few mirrors to hide behind, she started experimenting with colours and advising friends on what suited them. She had been born in Kensington, in February 1973, but the island had shaped her first. At eleven, while in the US, she had met Mary Greenwell, a makeup artist who would later become her mentor. Tilbury has called her incredible. After the Glauca Rossi School of Makeup she began assisting Greenwell, and then worked her way up in fashion: Vogue, Vanity Fair, show makeup director at Fashion Week. Backstage was the laboratory. During years of shows she formulated the cream that would become Magic Cream, and she learned what models and designers wanted in the ninety seconds between a look and the runway. What almost nobody saw was how long that apprenticeship lasted. She was not an overnight founder. By 2013 she was forty, a working artist with a career, when she launched Charlotte Tilbury Beauty at Selfridges in London. A YouTube channel gave her an audience before she had a store; her 'makeup wardrobe' idea sold shoppers the whole look instead of one lipstick. The price of brand success is imitation. In 2019 she won a lawsuit against Aldi over a copied product. A year later, in June 2020, Spanish group Puig bought a majority stake. Tilbury kept a significant minority share and stayed on as chair, president and chief creative officer, and joined Puig's executive committee. Her net worth was estimated at £350m as of March 2025. The quiet takeaway is that the fashion world's backstage had been her business school. She spent decades learning exactly what women wanted to feel in the mirror, and only then did she put a name on a jar. ## Oxford Nanopore Technologies (United Kingdom) URL: https://startup.talk/uk/oxford-nanopore-technologies/ Founders: Gordon Sanghera, Hagan Bayley, Spike Willcocks Founded: 2005; Based in: Oxford, England; Sector: Genomics / DNA sequencing Gordon Sanghera has described himself, in a Mishcon de Reya interview, as the eldest child of a first-generation Punjabi family. Anyone from an Indian household will recognise what that means: the eldest carries the plan. His own plan began in science. He studied chemistry, took a PhD in bio-electronic technology, and then joined MediSense, a spin-out in Oxford that built glucose sensors. It taught him how a lab idea becomes a product people carry in their pocket. In 2005 he was appointed chief executive of another Oxford spin-out, this one built on the work of professor Hagan Bayley. The idea was a nanopore: a hole so tiny that single strands of DNA could be threaded through it, and each base read as it passed. IP Group provided the seed. Spike Willcocks was co-founder. What almost nobody outside the lab appreciated was how long nothing happened. Ten years separated the founding in 2005 and the commercial MinION in May 2015, a sequencer small enough to hold in one hand when rivals needed rooms. A £70m fundraising in 2015 helped cover the gap. Then the product started to travel. In August 2016 NASA astronaut Kate Rubins used a MinION aboard the International Space Station. GridION arrived in 2017; PromethION, with 144,000 channels against MinION's 512, in 2019. The company listed on the London Stock Exchange on 30 September 2021 under the ticker ONT. The cost of building a platform shows in the accounts: for 2025, revenue of £223.9m against an operating loss of £155.3m. In December 2025 the company said Francis Van Parys, former CEO of Radiometer, would succeed Sanghera, who had led it for around 20 years. The founder's lesson is patience with the eldest-child kind of discipline. A decade of silence is survivable if you keep the first product small enough to ship and bold enough to take to space. ## The Hut Group (United Kingdom) URL: https://startup.talk/uk/the-hut-group/ Founders: Matthew Moulding, John Gallemore Founded: 2004 (Manchester; Moulding's co-founding sometimes dated 2003); Based in: Manchester, England; Sector: E-commerce, beauty and nutrition brands, retail technology Before the billions and the boardroom rows, there was a Burnley boy whose father laid tarmac and whose mother traded antiques. Matthew Moulding was born in 1972 and went to a Catholic high school in Colne. Two trades, both about buying something and making it worth more. He studied industrial economics at Nottingham, then qualified as a chartered accountant, the safest road a clever lad from Lancashire could take. He took it. He ended up as finance chief of the distribution business at the Caudwell Group, working close to John Caudwell, the mobile phone tycoon. A salaried seat beside a founder is a good place to learn what founding costs. Then he left. In Manchester in 2004, with a £500,000 investment and a partner, John Gallemore, he began selling CDs and DVDs online. Soon the pair were building web stores for others, including Asda, Tesco and WHSmith. It looked like a business. Then the iPhone arrived in 2007 and the ground shifted. What nobody outside the company noticed was the pivot hiding in the wreckage. Instead of renting out their software, they started buying the shops. Zavvi in 2009. Lookfantastic in 2010. Myprotein in 2011. Each brand ran on the platform they had built, and the platform got better with every one. The price came later. THG sued Myprotein's former owner, Oliver Cookson, and lost, with the Court of Appeal dismissing its challenge in March 2016. The flotation on 16 September 2020 was the peak. It was the largest London listing since 2013, and shares rose 25% on day one. Moulding received at least £830m in shares. Then the market turned. Shares fell over 60% by October 2021, investors questioned property deals in which he sold assets to himself and leased them back, and Charles Allen replaced him as chairman in March 2022. He stayed as chief executive. The company has since shrunk to fit its shape: luxury sold to Frasers in 2024, the technology arm demerged in January 2025. Revenue was £1,717.9m for FY2025, with net income of £54.1m. The lesson is not about scale. It is that the boy from a tarmac family learned the accountant's rule early, that you can survive almost anything except running out of cash, and then he spent two decades testing how far a founder can bend it before the board bends back. ## Huel (United Kingdom) URL: https://startup.talk/uk/huel/ Founders: Julian Hearn, James Collier Founded: 2014 (first product released 2015); Based in: Aylesbury, England; Sector: Nutrition, complete meal powders and ready-to-drink The newspaper headline that followed the biggest day of Julian Hearn's life did not call him a food tycoon. It called him an ex-labourer. That is the detail that stuck. Reports say Hearn left school at 16 with few qualifications, and that he worked in retail and manual labour before drifting into marketing. These accounts come from press summaries, and the details deserve care. But the outline is consistent: he started with no head start. Marketing was where he found his footing. He is reported to have held senior roles at big names in retail and food, and to have sold an earlier company in 2011. Then came the itch that every busy person knows. Why does eating well take so long? In 2014, in Aylesbury, he started Huel. He did not formulate it himself. The recipe came from nutritionist James Collier, and the first powder reached customers in 2015. A simple idea, a plain bag, a clear promise: a full meal you could make in a shaker. What almost nobody expected was how far the plain bag would travel. Shipping across Europe began in 2016 and to the United States by 2017. That same year James McMaster became chief executive, and Hearn stepped toward the marketing side. A founder who knows his limits is rarer than a founder who knows his product. It was not smooth. The Advertising Standards Authority banned two ads in February 2023, and a further one featuring Steven Bartlett in August 2024 for undisclosed affiliations. In 2025 a Consumer Reports study reported elevated lead and cadmium in some products, and in September 2026 another ad was banned for implying Huel could replace all conventional food. Trust, for a food company, is the product. The money did arrive. Revenue was £184.5m in the year to July 2023 and passed £200m the next year, with profit tripling. In March 2026 Danone agreed to buy Huel for around €1bn, and the Competition and Markets Authority cleared it in August 2026. Press reports say Hearn will take about £400m. For a founder with no degree and no inheritance, the lesson is quiet. You do not need to invent the thing yourself. Hearn found the expert, trusted the recipe, and put his own energy where he was strongest: telling people it existed. ## Zopa (United Kingdom) URL: https://startup.talk/uk/zopa/ Founders: Richard Duvall, James Alexander, Giles Andrews, David Nicholson, Tim Parlett Founded: 2005 (launched Mar 2005; formed 2004); Based in: London (Canary Wharf), England; Sector: Digital banking, lending, fintech In Buckinghamshire in 2004, a small team who had once helped build Britain's first internet bank decided that the next idea was to leave out the bank. They had met at Egg, the online bank that launched in 1998 and showed that British customers would trust a website with their money. The five who would become Zopa's founders, Richard Duvall, James Alexander, Giles Andrews, David Nicholson and Tim Parlett, had seen the inside of that machine. They had also seen eBay take off. If strangers could trade a teapot online, why couldn't they lend each other money? The company needed money first. Early in 2004 it had to raise funds, and Giles Andrews is said to have had experience raising angel finance. The plan sounded, to many, like a hobby for people who had read too much about the internet. Lenders pick borrowers, banks take a cut and pay savers less, the founders argued. Zopa would simply arrange the match. It launched in March 2005. For years it grew quietly. Through the 2008 financial crisis its investors suffered only a slight dip in returns, a record the founders could point to when banks were collapsing. In January 2017 it became the first British P2P lender to pass £2bn in loans, and in May that year it became fully FCA regulated. But what almost nobody outside fintech noticed was that the lender who had made peer-to-peer famous was already planning to leave it. In November 2016 Zopa announced it wanted a banking licence. It got interim licences in December 2018 and a full one in June 2020. In December 2021 it closed the P2P business entirely and bought back investors' loans. The price was the dream that started it. A company founded to bypass banks had become one. The money followed. SoftBank and others put in $300m in October 2021 at about a $1bn valuation. A 2022 IPO plan never happened. In December 2024 Zopa raised $87m privately instead, citing a sluggish tech IPO market. Revenue reached £303.4m in 2024, with net income of £34.2m. For a founder, the story is not that the first idea was wrong. It is that Zopa was willing to treat its own founding idea as a stepping stone, and keep the team, the brand and the customers when it changed. ## Moneybox (United Kingdom) URL: https://startup.talk/uk/moneybox/ Founders: Ben Stanway, Charlie Mortimer Founded: 2015 (app launched about 2016); Based in: London, England; Sector: Savings, investing, pensions, wealth management app In 2015 two founders began building an app on a simple observation: most people do not save because saving feels like a decision, and decisions are tiring. Ben Stanway and Charlie Mortimer built Moneybox around that. Round up each card purchase to the next pound, drop the change into a savings or investment pot, and the user hardly notices. What is clear is the stakes. A fintech aimed at beginners has to win trust from people who are nervous about money, with a team that is still tiny. It launched around 2016 as a small savings and investing app, and by the Series D in March 2022 it had attracted Fidelity International Strategic Ventures, Oxford Capital, Breega, Burda and CNP. Then came the quiet test. In the year to May 2023 it posted a loss of £4.1m. A year later it posted a pre-tax profit of £26.5m on revenue that rose from £28.7m to £77.2m. That swing is the part that is rarely told. Growth had not stalled. The product had finally started earning more per customer than it cost to serve them. In October 2024 about £70m of secondary investment valued the business at £550m, up 84% from the Series D. Apis put in around £60m and Amundi £8m. These were purchases of existing shares, not new money. The next two years brought scale and a trade-off. In 2025 revenue passed £100m for the first time, at £115.6m, with 1.71m customers and £19bn under administration. Profit before tax slipped to £14.9m from £19.4m as the company reinvested in technology. In 2026 a share sale of up to £45m for long-serving staff set a value near £800m. What a first-time founder can take from this is a plain one. Stanway has called the latest valuation recognition of progress. The more useful number is what came before it: a loss, then a profit, in two years. Habits first, margin second. ## Oxford Ionics (United Kingdom) URL: https://startup.talk/uk/oxford-ionics/ Founders: Chris Ballance, Tom Harty Founded: 2019; Based in: Begbroke / Oxford, England; Sector: Quantum computing (trapped-ion chips) Two doctoral students at Oxford stood in front of the same stubborn machine more than ten years before either of them ran a company. Chris Ballance and Tom Harty met during their PhDs in trapped-ion physics. The ions themselves were beautiful, the best qubits anyone had. The problem was everything around them. The control systems were bulky, hand-built and wildly hard to scale. A machine that worked in a lab could not become a product. They did not leave. Both stayed in the field and, in 2019, founded Oxford Ionics. Ballance later described them as two physicists with a strong conviction that the technology had untapped potential. Conviction is what a founder has before there is evidence. The stakes were plain. A quantum startup spends years burning money on hardware before it can show a customer anything, and bigger names like IBM and Google had a head start. Early backing came from Oxford Science Enterprises, followed by external rounds. What almost nobody outside the field understood was the idea itself. Rather than taming the control hardware, they chose to replace it, using conventional silicon chip technology to run the ions. If it worked, a quantum computer could be built with the same factories that make ordinary chips. It worked well enough to attract a buyer. On 9 June 2025, US-listed IonQ announced a $1.075bn deal for Oxford Ionics, paid as $1.065bn in IonQ shares and $10m in cash. The deal completed in September 2025 and the University of Oxford calls it the highest-value acquisition of any of its quantum spinouts. The team is expected to keep working in the UK. There is a cost hiding in that price. A British deep-tech company, built over six years on a university's research, now sits inside an American one. The quiet lesson is in how it started. The founders did not meet at an accelerator. They met at a lab bench, spent a decade learning what was broken, and only then decided it was fixable. ## Faculty (United Kingdom) URL: https://startup.talk/uk/faculty/ Founders: Marc Warner, Angie Ma, Andy Brookes Founded: 2014; Based in: London, England; Sector: Applied AI software and consulting Saul Klein says he decided to back them after about five minutes. What Marc Warner remembers saying in that room is stranger than any pitch. He told one of Britain's best-known seed investors that most PhDs were not suited to commercial work. That was the point. In 2014 Warner, a physicist who had worked on quantum sensing as a Marie Curie Fellow at Harvard, set up something with a clumsy name: the Advanced Skills Initiative. It was not an AI company. It was a fellowship that took academics with doctorates and gave them about two months of real-world training, because, as the founders saw it, you cannot get a job without experience and you cannot get experience without a job. He built it with Angie Ma, a UCL physicist who had worked in applied optics, and Andy Brookes, the one founder without a PhD, who became the technical chief. The five-person founding team was heavy with physicists and computer scientists. Warner has said little in public about his childhood, and the sources I found offer no family story, so the human detail sits in the founding choice itself: they started by solving their own world's problem, the academic who cannot get hired. What almost nobody connects to the Faculty of today is that origin. The fellows were not a side project. They were the product. Companies and government departments paid to take on fellows, and the fellows turned out to be good at it. In 2016 the business raised GBP1.5m, and ASI Data Science grew from training people into building models for clients. Then came the part that made headlines for the wrong reasons. In 2019 the company was renamed Faculty, and in the same period it faced scrutiny over links to the Vote Leave campaign and payments from a company linked to Dominic Cummings. Critics questioned a non-competitive COVID contract with the Department of Health. The National Audit Office found no evidence that ministers influenced procurement, and Faculty has said it follows conflict-of-interest best practice. Even so, the early work with the Home Office on a terrorist-content blocking tool and at least GBP3m in government contracts between 2018 and 2020 left the firm explaining itself for years. It kept building. In 2021 Apax Digital put in GBP30m, taking total funding to about GBP40m. In January 2026 Accenture agreed to buy Faculty for over GBP740m according to the Financial Times, and Warner was named Accenture's chief technology officer. The lesson is easy to miss. Faculty's first customer was the problem its founders knew best. They did not start with a clever model. They started with a gap they had personally watched people fall into, and they charged to close it. ## Beamery (United Kingdom) URL: https://startup.talk/uk/beamery/ Founders: Abakar Saidov, Sultan Murad Saidov, Michael Paterson Founded: 2013; Based in: London, England; Sector: Talent acquisition and talent management software (HR tech) The family lived on GBP28 a week for food. That is the number Abakar Saidov chose to give when he later described arriving in Britain in the mid-1990s. His father, a theoretical physicist, had retrained for a desk job and sometimes walked three hours to get to work. His mother, a doctor and neuropathologist, was not allowed to practise and ran a shop instead. He had spent his first ten years moving between Moscow and Dagestan, and has said he never felt Russian in Moscow. In England he did not speak the language, and he has said he was bullied for it. That, he says, made him study harder. He was also, from early on, selling things. With his brother Sultan he ran a paper round and a sweet shop and imported goods such as electronics and perfume. At 18 he went to Beijing to learn Mandarin, taught golf, and tried to export Chinese cement to Europe. It failed. What almost nobody knows is how close he came to a different life. At Warwick he studied maths and economics on a Royal Air Force sponsorship, hoping to fly. The rules changed to exclude naturalised citizens, and he has said it came just after the 7/7 bombings. The door closed. He went to Goldman Sachs, where his internship fell in the 2008 crash. He has described gaining about 10 kg and sleeping two or three hours a night, and he spent his student loan on Apple shares. Then he moved to the buyout firm Francisco Partners and decided he wanted to build companies rather than fund them. In 2013 he, Sultan and Mike Paterson founded Beamery, first called Seed.Jobs, working from the family home. Candidates from abroad sometimes slept there. They were accepted into the AngelPad accelerator in New York, and at demo day in San Francisco they held more than 80 meetings and raised nothing. Back in London they stayed in a crowded hostel. Saidov called it a depressing low point, and has said he sold his car just to make payroll. The turn was a decision to leave money on the table. In 2015 small recruitment firms became their first paying clients; the brothers dropped them to aim at big enterprises. VMware signed in 2016, then Meta. Index Ventures led a $5m Series A in 2017, and by 2022 a $50m Series D valued the company at $1bn, bringing total funding to about $228m. The price came later: a 12% staff cut in January 2023 and around 25% that November. In April 2025 Abakar became executive chairman and Sultan took over as CEO. The lesson is in the budget. A family that counted pounds a week raised a founder who could count what he was willing to give up, including the clients who paid first. ## Cleo (United Kingdom) URL: https://startup.talk/uk/cleo/ Founders: Barney Hussey-Yeo Founded: 2016; Based in: London, England (most users in the US); Sector: Consumer fintech / AI money assistant There is a stretch of 2016 in which the founder of what would become a nine-figure revenue business could not get his own family to download it. One profile says Barney Hussey-Yeo struggled to get even his partner or his mother to sign up. Cleo was a chatbot, and it lived, in its first form, in his bedroom. He came to it by a plain route. He took a computer science degree at Manchester Metropolitan, a masters in machine learning at Bristol, and went to work as a data scientist at Wonga, the payday lender, for about 13 months. A podcast blurb describes state-school beginnings in Sheffield, though I could not find a fuller account of his childhood. The product he built next was a Facebook Messenger assistant that read your bank data and nagged you, with a joke, about spending. He joined Entrepreneur First and, by March 2016, had already launched. The wound was ordinary and the stakes were small. The first users were hard to find. The first money, a GBP2m round led by LocalGlobe in 2017, came with backers such as Taavet Hinrikus and Ian Hogarth. What nobody expected came from across the Atlantic. When Cleo launched in the US, daily signups jumped from around 1,000 in the UK to around 10,000 within a week, by Hussey-Yeo's account, because American bank APIs, especially Plaid, made connecting an account easy. By the December 2020 Series B, led by EQT Ventures with $44m, Cleo had 4 million registered users, 96% of them in the US, and over $10m in annual recurring revenue. A free tier sat beside a $5.99 premium plan and a salary advance of up to $100. The price was a business that looked more like American fintech than a British chatbot. Reported 2024 revenue rose from $65.9m to $135.7m with a small pre-tax loss, and in 2025 the company launched a voice-and-memory version of its coach. In February 2026 a press release claimed $185m in annual recurring revenue; Hussey-Yeo has also cited $280m. The figures differ by source and are the company's own. The lesson is in the chatbot's first users. If a product only works when strangers like it, you will not learn much from your mother. Hussey-Yeo kept going until strangers did. ## Stability AI (United Kingdom) URL: https://startup.talk/uk/stability-ai/ Founders: Emad Mostaque, Cyrus Hodes Founded: 2019 (company founded with Hodes; public breakthrough 2022); Based in: London, England; Sector: Generative AI (image, video, audio models) The failure came first. In 2019 Emad Mostaque started a company called Symmitree to cut the cost of technology for people living in poverty. It did not work, and the two co-founders later gave conflicting accounts of why. It is almost never mentioned alongside the billion-dollar valuation that followed. He was born in Jordan in 1983 to a Bengali Muslim family, moved as an infant to Dhaka and came to the UK at seven. He studied mathematics and computer science at Oxford, met his future wife at 19 on a student trip to the US, and has said he was diagnosed with Asperger's and ADHD. In his twenties he built online forums for Muslim communities and floated an idea of an Islamic AI to guide religious life. He worked as a hedge fund manager in crude oil, and advised governments on Middle East affairs. Then, in Covid, he and Cyrus Hodes led a big-data effort to support government decisions. They co-founded Stability AI, and Mostaque took the chief executive job. What almost nobody expected was how fast it would blow up. In August 2022 the company released Stable Diffusion, a text-to-image model that grew from research by groups at LMU Munich and Heidelberg. It was open for anyone to use. That year Stability raised $101m at a $1bn valuation, led by Coatue and Lightspeed. The cost arrived within a year. In June 2023 Forbes, citing more than 30 sources, reported that Mostaque had misled people about his education, an AWS partnership and his role in Stable Diffusion; he answered on his blog. In July 2023 Hodes sued, claiming he had been tricked into selling a 15% stake for $100. The company called it seller's remorse. Artists sued. Getty sued. In October 2023 Lightspeed wrote that his mismanagement had severely undermined its confidence. On 23 March 2024 Mostaque resigned. He said he wanted to pursue decentralised AI; reports said investors pushed him out. Prem Akkaraju, formerly of Weta Digital, became CEO in June 2024, with Sean Parker as executive chairman, and James Cameron joined the board in September. In November 2025 the UK High Court ruled that Stability AI did not infringe copyright in the Getty case, and a $76m Series B followed in 2026, while US suits continue. The lesson is in the order of events. A failed small company and an enormous public one share a founder. The second did not erase the first; it raised the cost of every claim he made. ## Cera (United Kingdom) URL: https://startup.talk/uk/cera/ Founders: Ben Maruthappu, Marek Sacha Founded: 2016; Based in: London, England; Sector: Home healthcare / health tech The reason Ben Maruthappu gives for starting Cera is not a market gap. It is a family emergency. He was a doctor. The documented fact is the contrast. Here was an NHS doctor who knew how care worked from the inside, and still found it hard to arrange for someone he loved. The system was not built to talk to itself. He launched Cera in November 2016 with co-founder Marek Sacha, aiming to put technology between a vulnerable person at home and the hospital. The pitch was practical: carers and nurses would visit homes, and software would predict trouble before it landed in an ambulance. The company claims its AI predicts over 80% of falls and 83% of hospitalisations a week ahead, and that hospitalisations fall by up to 70%. These are the company's own numbers, not audited ones. What almost nobody sees in the funding story is how it was paid for. In January 2025 Cera announced $150m, mostly debt with some equity, led by BDT & MSD Partners and Schroders Capital. That valued it above $1bn and made it a unicorn. Debt is a particular vote of confidence: lenders expect a steady, contracted business, and Cera sells largely to councils and the NHS. The company says it works with more than 150 local governments and two-thirds of NHS Integrated Care Systems. Scale followed. Cera said it delivers over 60,000 home visits a day, and a data provider listed more than 10,000 carers and nurses in April 2026. Maruthappu has said the company is getting to profitability, though no audited profit figure appears in what I could read. The price of this model is not dramatic, but it is real. Home care is a regulated, low-margin trade where staff retention and council budgets decide the year. Cera has had to grow into Germany while still proving its numbers. The lesson is almost banal. The founders of the large things often begin with a phone call to arrange something for someone they love, and a system that made that call hard. ## Motorway (United Kingdom) URL: https://startup.talk/uk/motorway/ Founders: Tom Leathes, Harry Jones, Alex Buttle Founded: 2017; Based in: London, England; Sector: Online used-car marketplace Seven years before Motorway existed, Tom Leathes, Harry Jones and Alex Buttle already knew what it felt like to sell a company. In 2010 uSwitch bought Top10, the broadband and mobile comparison site the three had built. That first exit is the quiet fact under everything that followed. Top10 was a lesson in a particular kind of market: one where the customer is confused, the middlemen are comfortable and the price is hidden. The founders had spent years inside that confusion. When they looked for the next one, they landed on something far larger and more emotional than a mobile tariff. The family car. Selling a used car in Britain was, and often still is, a small humiliation. You type your registration into a website, get a number, drive to a branch, and watch the offer shrink. Motorway's own research later found online offers ran on average about 6 per cent above what sellers actually received. So in July 2017 the three started again. They did not rush to the big cheques. A GBP 500,000 angel round came first, then GBP 2.75m of seed money from LocalGlobe and Marchmont Ventures in 2018, and GBP 11m in 2019. The idea was a flip: the seller photographs the car once, and thousands of verified dealers bid against each other in a private auction. The dealer, not the seller, takes on the hassle. What the outside world did not see was how fragile a marketplace is at the start. A buyer-side network of dealers means nothing without cars, and cars mean nothing without dealers. Then, in March 2020, the UK locked down and Motorway stopped trading altogether. It restarted in May with contact-free collection. The pandemic then helped. A chip shortage pushed used-car prices up, sellers wanted a safe, quick route, and by the third quarter of 2021 sales had risen 300 per cent to $411m. Index Ventures led a GBP 48m round in June 2021, and in November came a $190m round with ICONIQ Growth at a valuation above $1bn. About 4,000 dealers were on the platform. Leathes told TechCrunch that eight thousand cars a month was still under one per cent of the UK market. The price of that speed has been money. Motorway reported revenue of about GBP 66m for 2023/24 against a loss of GBP 37.3m for 2024, and has said it expects to turn profitable in 2026. For a first-time founder the interesting part is the order of events. The three did not begin with a vision of a unicorn. They began with a product that had worked once, a sale that gave them credibility, and a market where people were quietly being short-changed. ## Tractable (United Kingdom) URL: https://startup.talk/uk/tractable/ Founders: Alex Dalyac, Razvan Ranca, Adrien Cohen Founded: 2014; Based in: London, England; Sector: AI for accident and disaster recovery (insurance claims) Alex Dalyac was turned down. The programme was Entrepreneur First, the London talent-investor that backs people before they have an idea, and the reason was blunt: he did not know enough about technology. He had been drawn to the field by a Coursera course. Six months before deep learning became the thing everyone talked about in 2013, he took Geoffrey Hinton's neural network lessons and, as he later put it, fell in love. At Imperial College London he and teammates built a plant-recognition app, and he spent spare hours on image classification. So he did the unglamorous thing. He went away, learned, and reapplied. This time he was accepted. What almost nobody realises about Tractable is where its famous insurance product did not come from. At an EF hackathon Dalyac met Razvan Ranca, a machine-learning researcher out of Cambridge, and the two coded through the night. In 2014 they started the company. The first paying customer was not a car insurer, or even a car park. They were a business that used the software to inspect plastic pipe welds. That customer, Dalyac wrote, stopped working with them around the time of their first funding round. Investors, for their part, thought pipe inspection was too small a market. The founders found themselves with a technology that could see and no clear thing to look at. They tried utilities, geology, dermatology and medical imaging. Help came from a third founder. Adrien Cohen, who had co-founded Lazada before it was sold to Alibaba for $1.5bn, joined in 2015 and taught the two engineers how to build a business and hire. The company that came out of that search appraises damage on cars and homes from photographs, so that an insurer can settle a claim in hours rather than weeks. In June 2021 a venture round valued it at $1bn, making it Britain's hundredth unicorn, with more than 300 staff. In July 2023 SoftBank's Vision Fund 2 put in $65m. There are limits that Tractable itself would admit: the AI sees only visible damage, not what has twisted inside a chassis or behind a wall. For a founder, the quiet lesson is in the first two chapters. One rejection was treated as homework, and the first customer was treated as a clue rather than a destination. The business that made them rich came from the years they spent being wrong about where to point the same idea. ## Flo Health (United Kingdom) URL: https://startup.talk/uk/flo-health/ Founders: Dmitry Gurski, Yuri Gurski Founded: 2015; Based in: London, England; Sector: Femtech / women's health app The man who helped build one of the world's most-used period trackers began his career selling books. In 2003, aged about twenty, Yuri Gurski became CEO of the Minsk representative office of Piter, a publishing house. He ran it for six years. He was born in Minsk in January 1983, when Belarus was still part of the Soviet Union, and he has a twin brother, Dmitry. He studied journalism at Belarusian State University, finishing in 2008, then took an Executive MBA with the Warsaw-based IPM Business School in 2010. Books did something quietly useful. Yuri went on to write around thirty books about IT, among them a Photoshop title that became a bestseller. He had learned how to explain software to ordinary people, and how to sell it. What the story of Flo often leaves out is how long the brothers practised before it. In 2009 Yuri co-founded Viaden Mobile, which made fitness and gaming apps; it was bought in 2012 by Teddy Sagi. He then ran Sport.com, mentored a mapping startup that Mail.ru later acquired, and became a Mail.ru vice president in 2016. Flo started in 2015, in Minsk. It was a cycle-tracking app with a deceptively plain goal. The money came in small cheques: $1m in 2016 from Flint Capital and Haxus, $5m in 2017 with the model Natalia Vodianova among the backers, and $6m from Mangrove in 2018, which valued the company at $200m. Founders Fund joined with $7.5m in 2019. The headquarters later moved to London, with engineers in Amsterdam and Vilnius. The cost of holding intimate data came in public. In 2020 the US Federal Trade Commission alleged Flo had told users their health data was private while sharing it with outside firms, including Facebook and Google. Flo settled in 2021 and later launched an Anonymous Mode. In 2025 it paid $8m to settle a US class action, while Meta, which did not settle, was found liable by a jury. The company has kept growing. In July 2024 General Atlantic led a round of more than $200m at a valuation above $1bn. Dmitry, the CEO, said the company already generated positive cash flow, and Flo says it reaches more than 70 million people a month. For a founder the thread is practice. Flo looked like an overnight idea to the outside world. Inside, it came at the end of fifteen years in which two brothers wrote books, built apps, sold one company and learned what users would trust them with. ## Fresha (United Kingdom) URL: https://startup.talk/uk/fresha/ Founders: William Zeqiri, Nick Miller Founded: 2015; Based in: London, England; Sector: Salon and spa booking software and marketplace Before Fresha was a company with a billion-dollar valuation, William Zeqiri was a student in Auckland who was already being paid to write software. From 1999 to 2002 he studied Information Systems at the University of Auckland, and during those years he worked as a contracted software engineer for several companies. That is the sort of detail that never reaches a funding announcement, but it explains the founder: someone who learned the trade by being useful to paying customers before he had a degree. When he graduated, he did not rush off to a big firm. He co-founded MacroWeb, a web development and hosting company, and became its chief technology officer. In between he consulted on IT for large organisations including Air New Zealand and Auckland Hospital. He also earned the Project Management Professional credential in 2014, a quiet sign that he cared about how things get delivered as well as built. What almost nobody remembers is that Fresha did not begin under that name. Zeqiri started the company in April 2015 with Nick Miller, who became chief of product, and it traded as Shedul until it rebranded as Fresha in 2020. The business was aimed at the least glamorous customers in tech: independent hairdressers, barbers, nail bars and spas who ran their diaries on paper. The model was bold. The software was free. Fresha made its money from payments and from a consumer marketplace, so a salon owner could start without a contract or a card. It is a model that depends on volume, and volume depended on being better than the incumbents' paid tools. By the time of its latest reported raise, in May 2026, a KKR-led investment valued Fresha at roughly $1bn and brought reported total capital to about $285m. The company has said it is profitable, with a revenue run-rate above $140m, and by earlier counts it hosted over 50,000 businesses and handled over 20 million bookings a month, headquartered in London. For a founder the lesson sits in how Zeqiri's early years prepared him. He was a builder who worked for clients before graduation, ran a service business before a product one, and picked a customer base that bigger software firms had underserved. ## Currencycloud (United Kingdom) URL: https://startup.talk/uk/currencycloud/ Founders: Nigel Verdon, Mike Laven Founded: 2007 (as FX Capital; rebranded Currencycloud around 2012); Based in: London, England; Sector: Cross-border payments APIs (Visa-owned) The introduction happened on the way to a brewery. In 2011, at a founders' event in Dublin run by Paddy Cosgrave, the investor Fred Destin mentioned an opportunity to Jos White of Notion Capital as they headed for the Guinness factory. White later recalled it as the first he heard of Currencycloud. By then the company was four years old, and it did not have that name. Nigel Verdon founded it in 2007 as FX Capital, and he was one of the early people to see that moving money across borders could be a piece of software rather than a bank counter. The pain was easy to feel. Cross-border payments were slow, expensive and full of hidden fees, controlled by banks that had no reason to change. Verdon's idea was transparent, low-cost payments delivered through the cloud. What the story rarely mentions is how small it still was when real money arrived. In 2012 Notion led a Series A alongside Fred Destin at Atlas and Sean Park at Anthemis. The company had just hired Mike Laven as CEO and rebranded as Currencycloud. It was making about GBP 1m in revenue and moving about GBP 25m a month. Modest, but the shape of the business was clear: sell the plumbing, not the storefront. Over the next decade the customers who plugged into that plumbing were the fintechs that would reshape British banking, among them Monzo, Starling and Revolut. By June 2021 the platform had processed more than GBP 2bn in transactions for nearly 500 customers in 180 countries. The ending was neither an IPO nor a failure. Visa, which had invested in an $80m round in early 2020 that sources put at around a $500m valuation, agreed in July 2021 to buy the whole company in a deal valued at $963m (GBP 700m). Visa already owned a stake, which reduced what it actually paid. For Notion's White it closed roughly nine years of involvement. The quiet insight for a founder is how little glamour built it. Currencycloud made its fortune by being the dull infrastructure other people's products stood on, and its journey from FX Capital to a near-billion-dollar exit took fourteen years. ## Moonpig (United Kingdom) URL: https://startup.talk/uk/moonpig/ Founders: Nick Jenkins Founded: 2000; Based in: London, England (head office also in Guernsey); Sector: Online personalised greeting cards, gifts and flowers It took Nick Jenkins four days to find the name. He sat checking domains on Network Solutions, looking for something short and easy to say over the phone, and landed on a word that had once been aimed at him in a school corridor. Moonpig was his nickname. That detail says more than the usual startup tale. Jenkins did not name his company after a clever brand idea. He named it after the thing other children called him, and bet his working life on it. Before that he had spent about eight years in Moscow as a commodities trader, a job about as far from greetings cards as you can get. He was a man who left a trading career for a website that put a customer's own photo on a card and posted it. He launched in July 2000. That was the worst month imaginable: the dot-com bubble was bursting and online retail was a joke at dinner parties. He raised more money from private investors and venture capital, and then waited. It took five years to turn a profit. On a podcast episode his own words were reported as: we came close to failing several times. What almost nobody expected was who would carry the company through. Not advertising, at first. Broadband and digital cameras spread, people discovered they could make a card with their dog on it, and customers told friends. The first profit arrived in 2005. A television campaign followed in November 2006. By 2007 Moonpig reportedly held about 90% of the UK online card market, and by summer 2009 it had sold to 2.57 million people. In July 2011 Photobox bought it for GBP120m in cash and shares. The price of success was not only commercial. In 2015 the company disabled its API and apps after a developer's public warning; the flaw, first found in August 2013, had exposed names, addresses and partial card details for about three million users. The business outlived it. Moonpig floated in London on 2 February 2021 at over GBP1.2bn on day one, and for the year to April 2026 reported GBP373.0m in revenue and GBP51.7m in net income. The lesson is small and odd. The name that might have embarrassed him became the asset. Jenkins did not hide the school nickname; he built a brand on it and made strangers remember it. ## Funding Circle (United Kingdom) URL: https://startup.talk/uk/funding-circle/ Founders: Samir Desai, James Meekings, Andrew Mullinger Founded: 2010 (launched Aug 2010; idea 2008, founders quit jobs 2009); Based in: London, England; Sector: Small business lending marketplace (fintech) The idea did not arrive in a boardroom. In 2008, as the banking system shook, three friends from Oxford talked about a peer-to-peer lending marketplace in a pub. Samir Desai, James Meekings and Andrew Mullinger were not obvious revolutionaries. Desai had been a consultant at BCG and then an investment executive at Olivant. I could not find where Desai grew up or what his parents did, and I will not invent it. What the record gives is the stakes. In 2009, with the financial crisis still biting, the three quit their jobs. The platform did not launch until August 2010, so for a year there was a plan, a team and no revenue. That gap is the hidden part of the story. It is easy to treat founding as a launch date. For these three it was a year of walking away from salaries to build something banks had made urgent. Small firms were being squeezed, and the founders bet that savers could lend to them directly. The turn came in 2012, when the British government used the platform to lend GBP20m to small businesses. A start-up that had only just launched suddenly had the state as a customer. In 2013 it raised $37m led by Accel Partners and went into the United States. In October 2015 it bought Zencap and reached Germany, Spain and the Netherlands. Then came the price of getting big. Funding Circle listed in London in September 2018, raising about GBP300m at GBP1.5bn, with the Danish investor Anders Povlsen anchoring the offering. Shares fell 24% on the first day. In 2019 pre-tax losses grew to GBP50.7m, and lenders reportedly waited almost 100 days to sell unwanted loans. In April 2020 the secondary market was suspended during COVID. The company changed shape. In March 2022 it left peer-to-peer lending for an institutional model, and in July 2024 it sold its US operations. Desai left the board in October 2024, the last original founder to go. For 2025 it reported revenue of GBP222.5m and net income of GBP46.0m. The lesson is in the pub. The best ideas often begin as complaints between friends, but the work is a year of quiet leaving before anyone else notices. ## Boohoo (United Kingdom) URL: https://startup.talk/uk/boohoo/ Founders: Mahmud Kamani, Carol Kane Founded: 2006; Based in: Manchester, England; Sector: Online fashion retail / marketplace Before there was a billionaire, there was a market stall. Mahmud Kamani's father, Abdullah, escaped war in Kenya in the 1960s and sold handbags in Manchester to feed his family. One account says he borrowed GBP500 from a friend in 1968 to start over, though that comes from a single newspaper profile and I would treat it with care. Another says six relatives shared one rented room. The son grew up inside the trade. Kamani worked the stall, then the family wholesale firm Pinstripe, delivering goods in a van for his father. It was at Pinstripe, which supplied chains such as Primark and New Look, that he met Carol Kane, who joined in 1993. In 2006 the two founded Boohoo. They knew what clothes cost to make and how fast a factory could turn a design around, and they put that knowledge on a website. The turn came in March 2014, when Boohoo listed on AIM at 50p. Shares traded well above the float price. The company was valued at roughly GBP560m to GBP600m; Kamani made GBP135m and Kane GBP25m. By April 2017 it was worth about GBP2bn. Then it went shopping. Karen Millen and Coast in 2019, Oasis and Warehouse in 2020, the Debenhams name in January 2021 for GBP55m, and Burton, Wallis and Dorothy Perkins the next month. The price arrived in July 2020. The Sunday Times reported that workers at a Leicester supplier were paid GBP3.50 an hour. A review by Allison Levitt QC found the allegations of poor working practices substantially true and called Boohoo's monitoring of factories inadequate. The share price fell 46% that month. Sir Brian Leveson was brought in for ethical oversight. A man whose father sold bags from a stall now faced a question about who was making his clothes. The company has since changed. Since March 2025 it trades as Debenhams Group, though Frasers Group blocked the legal name change. In the year to February 2026, revenue fell to GBP917.0m from GBP1,217.9m, partly because of a marketplace shift, and the loss before tax narrowed to GBP108.6m. The lesson is in the van. Kamani learned supply by delivering it. The advantage of knowing the factory floor is real, and so is the responsibility when someone else runs it. ## AO World (United Kingdom) URL: https://startup.talk/uk/ao-world/ Founders: John Roberts, Alan Latchford Founded: 2000; Based in: Bolton, England; Sector: Online electrical appliance retail A pound coin bought the whole thing. On Christmas Eve 1999 John Roberts had a drink with his friend Alan Latchford, and the friend bet him GBP1 that he could not start a business selling white goods online. Roberts took the bet. By 2000 the company existed. He was not new to work. His first job, at 14, was as a waiter at the Last Drop Hotel in Bolton. The first version was called Appliances Online. It sold fridges and washing machines direct and also for other retailers including Next, House of Fraser and B&Q. Latchford joined as a co-founder, but the friendship was shorter than the company: he left in 2006 for health reasons, and he died in 2020. Roberts has carried the pound-coin story since. The turn was not a website. It was control of the delivery. In 2009 the firm bought Expert Logistics, so its own vans handled the item at the door, and in 2013 it rebranded as AO.com. In February 2014 it floated in London to fund European expansion. Then came the cost of ambition. A Dutch site was announced in 2015 and a German headquarters in Bergheim opened in 2016. Roberts stepped down as chief executive in 2017, and when Steve Caunce left in January 2019 he came back. In November 2019 AO said it would close the Dutch business, and in June 2022 it shut Germany, which was about 10% of group revenue. The same year it raised GBP40m and cut GBP30m of costs, after a full-year loss. It has since done what it said: focus on cash. For FY2026 it reported revenue of GBP1,266.6m and net income of GBP35.9m. Roberts still owns about 18.3%, and Frasers Group holds about 25%. The lesson lies in the smallness of the start. A bet of GBP1 does not need to be big; it needs to be taken. And the founder who walks away from the controls is sometimes the one who has to come back for them. ## ITM Power (United Kingdom) URL: https://startup.talk/uk/itm-power/ Founders: Graham Cooley Founded: 2000; Based in: Sheffield, England; Sector: Green hydrogen PEM electrolysers A fuel-cell company in a small Essex town is not the picture most people have of ITM Power. It started in June 2000 in Saffron Walden, making fuel cells. Only later did it move into electrolysers, the machines that split water into hydrogen. The personal side of this company is thin. What this story offers instead is a company's life and the man who led it longest. In 2004 ITM floated on AIM and raised GBP10m, becoming the first hydrogen company listed on the London market. Then it waited. Hydrogen was a story that kept being told and not bought. In 2009 Graham Cooley took over as chief executive. He had run Sensortec and founded Metalysis and Antenova, and a podcast listing frames his time at ITM as one that began almost out of cash and ended with a GBP250m raise. That is a headline, not a source I could read in full, so I use it only as a signal. The turn came in 2019. Linde plc invested GBP38m as part of a GBP52m raise and a joint venture, ITM Linde Electrolysis, followed in January 2020. A GBP172m raise in 2020, including GBP30m from Snam, and GBP250m in October 2021 followed, with plans for 5 GW of annual capacity by 2024. The price came next. In 2022 the pre-tax loss almost doubled to GBP46.7m on revenue of just GBP5.6m. The company dropped its 5 GW target, cancelled a planned second Sheffield factory and said Cooley, after 13 years, would leave. Dennis Schulz became chief executive in December 2022. Revenue reached GBP16.5m in 2024, still small beside the money raised. The industry that attracted the cash is still waiting for the customers. The lesson is patience of a particular kind. ITM survived two decades because it had a technology and a listing, and then it learned that raising money for capacity is easier than selling to fill it. ## Google DeepMind (United Kingdom) URL: https://startup.talk/uk/google-deepmind/ Founders: Demis Hassabis, Shane Legg, Mustafa Suleyman Founded: 2010; Based in: London, England; Sector: Artificial intelligence research The first computer Demis Hassabis ever owned was paid for with a chessboard. In 1984, aged about eight, he bought a ZX Spectrum 48K with his tournament winnings. He was born in London in 1976 and grew up in the north of the city, the son of a Greek Cypriot father and a Chinese Singaporean mother. The family name had been spelled Hassapis before he changed it to Hassabis. Chess came first. By thirteen he was a master with a rating of 2300 and captaining English junior teams. The Spectrum arrived and the board lost some of its hold. He taught himself to program from books and later wrote a Reversi-playing program on an Amiga, his first taste of making a machine decide. What almost nobody pictures when they see him on a Nobel stage is a teenager in a games studio. After winning a magazine competition he joined Bullfrog Productions and, at 17, co-designed and lead-programmed Theme Park with Peter Molyneux. He is reported to have turned down a seven-figure offer around then, and went to Cambridge to study computer science, graduating in 1997 with a double first. Games still pulled at him. He worked on Black & White at Lionhead, then in 1998 started his own studio, Elixir. It closed in 2005. A founder who has watched a company he built fold is not the same founder as one who has only risen. He went back to school, this time for neuroscience. At UCL he took a PhD under Eleanor Maguire, studying memory and imagination, the machinery he wanted AI to borrow from. In 2010 he co-founded DeepMind with Shane Legg and Mustafa Suleyman, and in 2014 Google bought it for around GBP 400m. The payoff came on a different scale. AlphaFold, the system for predicting protein structures, brought him a share of the 2024 Nobel Prize in Chemistry, with John Jumper, and a knighthood in the same year. The quiet lesson is how little of the arc was a straight line. A boy with a chessboard, a teenage game designer, a founder whose first company closed, a neuroscience student. DeepMind drew on all of them, and none of it looked like preparation at the time. ## Babylon Health (United Kingdom) URL: https://startup.talk/uk/babylon-health/ Founders: Ali Parsa Founded: 2013 (launched 2014); Based in: London, England; Sector: Digital health / virtual GP Ali Parsa was sixteen when he left Iran. It was 1981, two years after the revolution, and he came to Britain without the O and A-levels he would need for university. So he taught himself the material. He was born in January 1965 in Rasht, in Iran's Gilan province, to a middle-class Persian family. The ability to start over, with nothing but a plan, sits under everything he did afterwards. He got into UCL, finished a civil and environmental engineering degree in 1987, and much later a PhD in engineering physics. In the middle he did something unusual for a student. He co-founded Victoria and Gilan, a media promotion company, and in 1993 won the Royal Award for Young Entrepreneur of the Year. He sold it in 1995. Money followed. He worked in investment banking at Credit Suisse First Boston, Merrill Lynch and, from 1999, Goldman Sachs. Then he turned to something else, and in 2004 founded Circle, the first private company to run an NHS hospital. He stepped down as chief executive in 2012, and Circle lost its contract at Hinchingbrooke in 2015 after severe criticism from the health regulator. Babylon came next, in 2014: a subscription app that let patients see a doctor by video, with an AI tool in front. Critics worried. In 2019 the Times reported a misleading AI diagnosis for a woman with chest pain. Parsa kept arguing for more private-sector involvement in the NHS. What almost nobody expected was how quickly it would end. In June 2021 Babylon announced a New York listing at a $4.2bn valuation. It went public that October, and the shares fell from $272.50 at flotation to below one dollar. Parsa later called the listing an 'unbelievable, unmitigated disaster.' A planned handover to MindMaze fell apart in August 2023. The US business entered Chapter 7, Rwanda was wound down, and in September 2023 the UK arm was sold to eMed for about GBP 500,000. For a founder the lesson is brutal and simple. A man who rebuilt his life alone at sixteen could not rebuild a company that the market had priced for perfection. Valuation is a promise you must then keep. ## Cazoo (United Kingdom) URL: https://startup.talk/uk/cazoo/ Founders: Alex Chesterman Founded: 2018; Based in: London, England; Sector: Online used-car retailer Alex Chesterman had already sold two companies when he started selling cars online. Neither of them had anything to do with cars. He went to St Paul's School in London and then studied economics at University College London. He was born on 9 January 1970. His first big job was at Planet Hollywood, where he became an executive vice president and stayed until 1998. That was the moment he decided to start something of his own. That something was ScreenSelect, a DVD-by-post service that became LoveFilm. Amazon bought it for GBP 200m. He was not finished. In 2007 he founded Zoopla, a property website, floated it in 2014 at a valuation close to GBP 1bn, and sold it to Silver Lake in 2018 for GBP 2.2bn. What almost nobody expected was where he would go with that money and name. He picked used cars, a market of shaky trust and heavy cost. In December 2018 Cazoo announced a round of more than GBP 30m, then raised another GBP 50m before launch. By 2020 it had raised GBP 180m in total. The pitch was simple: buy the car yourself, check it, deliver it. But owning stock turns a website into a car dealer. A marketplace carries other people's risk. Cazoo carried its own. Chesterman stepped aside as chief executive in January 2023, and left the company that December. In May 2024 Cazoo entered administration. His own wealth had been put at GBP 750m in the 2021 Sunday Times Rich List. He was also a visible political figure, giving GBP 300,000 to the Liberal Democrats for the 2019 election. For a founder the lesson is about capital intensity. Two asset-light digital businesses made him rich. The third needed warehouses, reconditioning and vast inventory, and it asked a different skill. A founder's track record transfers, but it does not guarantee the economics. ## Farfetch (United Kingdom) URL: https://startup.talk/uk/farfetch/ Founders: Jose Neves Founded: 2007 (launched 2008); Based in: London, England; Sector: Luxury fashion marketplace Jose Neves started out as a boy from a family of shoemakers. His grandfather owned a shoe factory in Porto, the Portuguese city where Neves was born in June 1974. He studied economics at the University of Porto, but he was not waiting for a degree. While still at university he founded Grey Matter, software for clothing manufacturers. He knew the factory floor from the family, and he saw how clumsily it connected to the world. That double life, software and shoes, shaped what came next. He also set up Platforme, software for small fashion brands, and around 1996 launched a footwear label called Swear and opened a shop in London. In 2001 he opened bstore on Savile Row, a boutique that put him on the other side of the counter. What almost nobody noticed was that this was the education. Running a physical shop, Neves learned how a small boutique has the best taste in the world and no reach beyond its street. In 2008 he launched Farfetch to fix exactly that. The idea took off. In June 2017 JD.com took a minority stake for $397m, before Farfetch had even launched in China. In September 2018 it listed on the New York Stock Exchange, and Neves's stake was valued at $1.4bn. Then the ground gave way. In January 2024 Farfetch went through a pre-pack administration and was bought by Coupang of South Korea for $500m. The deal wiped out existing shareholders, and Neves stepped down. Investors criticised the sale and sued. He has since moved to Brazil, with a wife and five children. For a first-time founder the quiet insight sits in the start. Neves built Farfetch on two things he had earned before the idea: a factory family's feel for product, and a software maker's patience with systems. His early businesses were not detours, they were the raw material. ## Arrival (United Kingdom) URL: https://startup.talk/uk/arrival/ Founders: Denis Sverdlov Founded: 2015; Based in: London, England; Sector: Electric vans and buses Denis Sverdlov's parents made fan blades for electrical turbines. They were engineers at a state factory in Saint Petersburg, and that is the home he grew up in, after being born in 1978 in the Georgian village of Agara. He studied economics rather than engineering, graduating from the Saint Petersburg State University of Engineering and Economics. In 2000 he started an IT consulting firm, IT Vision, then co-founded Korus Consulting from 2003 to 2007. Then came Yota, a mobile broadband company he co-founded. It reached operating profit within five months of launch, and MegaFon bought it in 2012 for $1.2bn. What almost nobody remembers is the next move. Sverdlov joined the Russian government as deputy minister of communications and stayed about fifteen months. He left in 2013 after a law stopped officials from holding foreign assets. He had to choose, and chose his own business. He moved his family from France to London in September 2015 and founded Arrival the same year. He ran it as CEO from March 2016 to November 2022. The idea was radical: electric vans and buses built in small, low-cost microfactories. He also started Roborace, a driverless electric racing series that folded in 2022. The peak came in March 2021, when Arrival went public by merging with the SPAC CIIG Merger Corp. It debuted at a valuation of roughly $13.6bn to $13.8bn, the largest listing in British startup history. By April 2021 he was worth $11.7bn on Bloomberg's index. Then the price turned. By April 2022 he had lost billionaire status, and his 94 per cent fall was called the largest wealth loss outside China in 2021. He stepped down in November 2022. In February 2024 Arrival was delisted and the UK arm went into administration. The lesson for a founder is how the public market treats a promise. A son of factory engineers tried to reinvent how factories work, but a valuation set on a story is a bill that comes due before the product does. ## Made.com (United Kingdom) URL: https://startup.talk/uk/made-com/ Founders: Ning Li, Brent Hoberman, Julien Callède, Chloe Macintosh Founded: 2010 (March); Based in: London, England; Sector: Online furniture and homeware retail (direct-from-manufacturer) Ning Li was in his mid-twenties and working as an investment banker when a sofa changed his life. Not a grand dream, just the plain irritation of trying to buy one. Years later he would tell interviewers that this small, annoying errand was where Made.com began. That is the quiet fact behind a company that once carried a £775 million price tag. Li was not a furniture man. He was a young finance graduate who noticed that a sofa in a shop cost far more than the wood, foam and fabric inside it. Sofas, he decided, were being sold by the wrong people in the wrong buildings. Before Made, he co-founded MyFab.com in Paris. It was an earlier attempt at the same itch: design for people who could not afford a showroom. What came next depended on a conversation. Brent Hoberman, who had built lastminute.com, told Li to settle in the UK, where he would help him launch a new business idea. Li has said that was the boost he needed. In 2010, with £2.5 million, Made.com opened its virtual doors, with Julien Callède and Chloe Macintosh as co-founders. The model was clever. Show a design online. Collect orders. Only then ask the factory to build it. No stockrooms full of unsold chairs. Funding followed: £6 million in 2012, £38 million in 2015, £40 million in 2018. Li even joined David Cameron's 2013 trade mission to China. What almost nobody outside the company remembers is that Li did not stay to see the end. He had left as CEO long before the crash, and in October 2022 he publicly accused the bosses who followed of losing focus as the business teetered. The floatation in June 2021 was the peak. The company listed at £775 million and posted £371 million in sales. But the model that avoided stock also meant customers waited. Shipping delays stretched for months. Freight costs soared. Losses more than doubled to £31.4 million. In February 2022 the CEO stepped down, citing family reasons. Sales fell. Talks with a buyer failed. In October it stopped taking orders. On 9 November 2022 Made.com went into administration, and about 500 people lost their jobs. Next bought the name and the website, not the staff, and relaunched it in August 2023. The lesson for a young founder is uncomfortable. A clean idea born from a personal annoyance can raise a fortune, but a promise to the customer, here a delivery date, is the real product. When the promise breaks, no clever model can save it. And a founder who has left still watches what happens to the thing they named. ## Bulb Energy (United Kingdom) URL: https://startup.talk/uk/bulb-energy/ Founders: Hayden Wood, Amit Gudka Founded: 2015 (launched to customers 2016; predecessor company incorporated earlier); Based in: London, England; Sector: Green energy retail Before he was an energy mogul, Amit Gudka ran a club night. He was a Cambridge maths graduate who spent about seven years trading gas and electricity at Barclays, and in his spare hours he DJed and ran a record label called Man Make Music. Few of Bulb's million customers knew that. It matters, because it shows the two lives he was trying to square. The trader saw a market built by insiders for insiders. The music man saw people who just wanted something that sounded right. Together with Hayden Wood, a management consultant who had advised the big suppliers, he decided that bills should be simple, green and kind. The two pooled savings and asked friends for money, then took funding from JamJar Investments. In 2015 they founded Bulb. They called customers members. One flat tariff. Emails in plain English. A referral scheme that paid both sides £50. There was almost no advertising budget. People simply told people. The curve was steep. About 15,000 customers in January 2017. Around 1.13 million by March 2019. In August 2018, DST Global and Magnetar put in £60 million, reportedly at a valuation of £400-500 million. That same year both founders cashed out £4 million of shares, a quiet detail that would later look very different. Then the market turned on them. In 2021 wholesale gas prices surged, and Bulb had promised fixed prices it could not cover. In February Gudka stepped back from day-to-day work to start a battery company, saying the business was stable when he left. By September Bulb was asking the government for a bailout. In November Ofgem put it into special administration, a first for a UK energy supplier. The government set aside £1.7 billion, and the Office for Budget Responsibility later put the possible liability at £6.5 billion. Octopus completed its takeover in December 2022. Years later Gudka admitted it was hard. He called the collapse pretty devastating and said that if the last fundraise had started about three months earlier, Bulb would probably have survived. He and Wood, he said, have had little contact since. He went on to launch Field, a battery storage business. The insight is not about bad luck. It is about timing and runway. A company that grows by promising customers a stable price is, underneath, a bet on its own balance sheet. The founders who survive start the next raise while they still look strong. ## Graphcore (United Kingdom) URL: https://startup.talk/uk/graphcore/ Founders: Nigel Toon, Simon Knowles Founded: 2016; Based in: Bristol, England; Sector: AI semiconductors (Intelligence Processing Units) Nigel Toon has sold a company before, and watched a chip company get absorbed by a bigger one. That history is the quiet fact beneath Graphcore. He began in chips at Altera, where he stayed about 13 years. In 2002 he co-founded Icera, a wireless chip firm that NVIDIA bought in 2011; reports put the price at $367 million or $435 million, depending on the source. He then ran Picochip and XMOS, and it was inside XMOS that the seed of Graphcore was incubated before the company stood alone in 2016. So when Toon and CTO Simon Knowles started Graphcore in Bristol, they knew the rules. Build a chip designed for one job, machine learning. Skip the cache hierarchy that traditional processors lean on and give the chip its own large memory. Name the ambition after statistician I.J. Good's idea of an ultra-intelligent machine. The money came quickly. A round led by Robert Bosch Venture Capital in autumn 2016. Atomico in July 2017, then $50 million from Sequoia. By December 2018 a $200 million round with Microsoft, Samsung and Dell Technologies valued it at $1.7 billion. In late 2020, a further $222 million lifted it to $2.8 billion. What the headlines of those years hid was the revenue line. Graphcore sold chips, but it was fighting NVIDIA, whose software ecosystem held customers tight. In 2022 revenue was $2.7 million against a net loss of $205 million. Meta took Graphcore's AI networking team in early 2023. Then in July 2024 SoftBank agreed to buy the whole company for around $500 million, a fraction of the 2020 valuation. About 450 people worked there by 2024. Toon is not a stranger to this part of the cycle. His earlier exits taught him that a deep-tech company can have the better engineering and still lose the market to the better ecosystem. Honoured with an honorary Doctor of Science by Bristol in 2023, he remains the face of British chip ambition. The lesson is blunt. In hardware, a billion-dollar valuation is a promise, not a business. Customers buy software and habit as much as silicon, and a founder who has exited once should plan the second company around the thing that locked them out the first time. ## Dyson (United Kingdom) URL: https://startup.talk/uk/dyson/ Founders: James Dyson Founded: 1991 (G-Force vacuum first sold in Japan 1986); Based in: Singapore (HQ since 2019); original base Malmesbury, Wiltshire, England; Sector: Consumer technology and appliances James Dyson was nine when his father Alec died of prostate cancer. The family home, he later said, was penniless. What nobody in Norfolk knew was that a headmaster would decide the rest of his life. Born in Cromer in 1947, one of three children, he was sent to Gresham's School, a boarding school in Holt. The headmaster agreed to pay his fees. The boy found something there: long-distance running. He has credited it with teaching him determination, the habit of going on when everything hurts. After a year at Byam Shaw School of Art he went to the Royal College of Art, first furniture and interior design, then industrial design. While still a student he worked on the Sea Truck. His first invention of his own was the Ballbarrow, a wheelbarrow with a ball instead of a wheel, and it was featured on the BBC's Tomorrow's World. The vacuum cleaner began in annoyance. His Hoover Junior clogged and lost suction. He borrowed an idea from a sawmill, cyclonic separation, and then spent five years and about 5,127 prototypes making it work. His wife Deirdre, an art teacher, paid the bills. 'There were 5,126 failures,' he said in a 2007 interview. 'But I learned from each one.' Then came the part that hurt most. Major manufacturers turned him down. Nobody in Britain would touch a vacuum without a bag, because bag sales were their profit. He took it instead to Japan, where the G-Force, bright pink, sold by catalogue for the equivalent of about $2,000. Only then did he found his own company, Dyson, and the Dual Cyclone became the UK's fastest-selling vacuum. Success brought wars. In 1999 he sued Hoover and won, with the court finding they had deliberately copied a fundamental part of his patent; Hoover paid £4 million. In 2017 he committed £2 billion of his own money to an electric car, only to scrap it in October 2019. In January 2019 he moved headquarters to Singapore. Today Dyson is owned entirely by the family, with 2025 revenue of £6.13 billion. Forbes puts his own net worth at $15.1 billion as of September 2026. The quiet lesson is that a founder's stubbornness is usually built young, from loss and from one person who paid the fees. Persistence is rarely a trait. It is a response to being given a chance and refusing to waste it. ## BrewDog (United Kingdom) URL: https://startup.talk/uk/brewdog/ Founders: James Watt, Martin Dickie Founded: 2007 (April); Based in: Ellon, Aberdeenshire, Scotland; Sector: Craft beer, bars and brewing James Watt once spent most of his early twenties at sea. He held a deep-sea captain's licence. At 24 he gave it up for a hut in Fraserburgh. Born there in 1982 and raised in the fishing village of Gardenstown, he studied law and economics at the University of Edinburgh from 2000 to 2004. Then he went to sea. Sir Tom Hunter later told the Herald that Watt worked on a trawler, studied law, and then started the craft beer company with Martin in a hut. The partner was Martin Dickie, a brewer. They were disillusioned with British beer. In April 2007 they launched BrewDog. Nobody had heard of it, and they had no marketing budget. What they had was noise. They made beers with absurd strength, 32 per cent for Tactical Nuclear Penguin and 55 per cent for The End of History, and let the press argue about them. They avoided traditional advertising altogether. In 2011 they offered shares to the public, raising about £2 million for around 8 per cent. In 2017 TSG Consumer Partners bought 22 per cent for about £213 million. A hut had become a beer empire. What was less visible was the cost of the noise. A prize of 'solid gold' cans turned out to be gold-plated brass, and Watt bought back 40 cans at a cost of £470,000. Then in 2021 former staff signed an open letter describing a culture of fear. A BBC Disclosure documentary followed in 2022, and BrewDog lost its B Corp certification. Watt stepped down as CEO in May 2024, staying as Captain and Co-Founder. That year net revenue was £280 million, and adjusted earnings before tax were £7.5 million, the first profit since 2021. But decline followed. Ten UK bars closed in July 2025. The Ellon distillery shut in January 2026. In March 2026 Tilray bought the UK and Ireland operations for £33 million, with 38 bars to close and nearly 500 jobs lost. It is hard to be fair to a story like this. The fisherman's son did what few people do: he built a global brand from nothing, in the open, on provocation. The lesson for a young founder is the other half of the same talent. Noise raises a company fast, but a team and a culture are what stay when the noise stops, and no captain can steer a ship on spectacle alone. ## Fever-Tree (United Kingdom) URL: https://startup.talk/uk/fever-tree/ Founders: Charles Rolls, Tim Warrillow Founded: 2004; Based in: London, England; Sector: Premium mixers and soft drinks Before there was a share price, there was a trip to the Democratic Republic of the Congo and a question that sounds almost trivial: why does the tonic in a gin and tonic taste so bad? Charles Rolls had spent his working life around gin. Tim Warrillow had spent his in advertising. They are the kind of pairing that makes a story, though what each gave up to join the other is not something their public record spells out. What it does show is the itch. A gin drinker could spend a fortune on a spirit and then drown it in a mixer full of preservatives and artificial sweeteners. The pair wanted a tonic made with natural ingredients. That meant going to the source. The tonic they were chasing depends on quinine, and the tree it comes from has a name that became their brand. In India, the cinchona tree was known as the fever tree, and quinine mixed with soda water and sugar was the earliest tonic. The founders travelled to the Congo and then spent months experimenting with botanicals before they launched Fever-Tree Indian Tonic Water in 2005. What almost nobody outside the drinks trade grasps is how small this was. The company was founded in 2004, and the first money came in 2006, when London & Lochside invested. There was no brand, no distribution, and a product that cost more than anything beside it on the shelf. The bet was that bartenders would care about the mixer once someone gave them a reason. They did. Fever-Tree launched in the US and Spain in 2007 and got onto Sainsbury's and Tesco shelves in 2008. By 2011 it was in the Sunday Times Fast Track 100. In 2013 the founders sold 25% to Lloyds Development Capital and London & Lochside walked away. In November 2014 the company listed in London at 134p, valuing it at about £154m. The years after were not a straight line. Reuters ran a headline in 2020 saying the shares were losing their sparkle as British tastes for premium tonic cooled, and the company has spent the years since leaning on America, where a distribution deal with Southern Glazer's put it in 29 states in 2018. In 2024 it even had to correct a news report that wrongly said it sponsored the Republican National Convention; it had only supplied free product. Today Fever-Tree reports revenue of £228.5m (2024), and a market value of roughly £932m in May 2026, a long way from a premium tonic pitched at gin drinkers. The lesson is easy to miss. They did not invent a new drink. They picked the least loved ingredient in a famous one and treated it as the product. Sometimes the opening is not a new category but the part everyone else decided was too small to bother with. ## Innocent Drinks (United Kingdom) URL: https://startup.talk/uk/innocent-drinks/ Founders: Richard Reed, Adam Balon, Jon Wright Founded: 1998 (company launched 1999); Based in: London, England; Sector: Smoothies and fruit drinks Three friends stood at a London music festival with a stall and two bins. One said YES. The other said NO. The question on the sign was whether they should quit their jobs to make smoothies. They had spent six months working out recipes and about £500 on fruit. Richard Reed, Adam Balon and Jon Wright had met at St John's College, Cambridge. By then they worked in consulting and advertising, steady jobs. The festival was a cheap experiment. Visitors who liked the drinks would drop their empty bottle in the answer they wanted. At the end of the weekend the YES bin was full. Only 3 bottles sat in NO. The next day, the three resigned. But a bin of bottles is not a bank. What came after is the part that rarely makes the poster. Finding money was hard. They had a product, a jar of encouragement and no revenue. The turn came with Maurice Pinto, a wealthy American businessman, who put in £250,000. It took about fifteen months from idea to market, a slow start for a company that later told its story as a joke. The business grew. By the mid-2000s Innocent was in a McDonald's Happy Meal trial that would run for five years, and had launched The Big Knit, where people knitted tiny hats for bottles in return for money to charity. But growth had a price. In 2007 the company recalled about 100,000 bottles after reports of smoothie bottles exploding, which it put down to natural fermentation. The Advertising Standards Authority also censured it that year over health claims on its superfoods range. Then the crash. In 2009 revenue fell and the company lost £8.6m. That year Coca-Cola agreed to buy a stake for £30m, with the founders keeping operational control. The stake rose to 58% in 2010 and past 90% in 2013, leaving the founders with a small minority holding. Ethical Consumer cut the company's rating from 12.5 out of 20 to 6.5 after the sale. By 2016 revenue was £247.4m, and the company sold more than two million smoothies a week. What stays with me is the bins. They did not ask a bank, a mentor or a survey panel. They asked strangers who had just tasted the product to do something small and physical. The quit-your-job decision came from evidence rather than courage, and that is easier to repeat than most founders think. ## Gousto (United Kingdom) URL: https://startup.talk/uk/gousto/ Founders: Timo Boldt, James Carter Founded: 2012; Based in: London, England; Sector: Recipe box / meal kits The Dragons said no. In 2013 Timo Boldt and James Carter stood in front of the investors on the BBC's Dragons' Den and left without a deal. Gousto was a year old. Today it makes more than £340m a year. Boldt had not planned this life. He grew up with entrepreneurship at the edges of his family, because his godparents had started several companies and he admired the lives they led. He studied between Germany and the US, then moved to London in 2008. He worked as an analyst at Rothschild, then as a vice president at the hedge fund Petrus Advisers. Four years into finance, he left. What he wanted was different. He told startups.co.uk he wanted to do something that had a positive impact on people and the planet, and he tied food waste, health and convenience together as the idea. Meal kits were new in the UK, and HelloFresh was the main competitor. Gousto launched on 12 April 2012. The television rejection did not stop it. In December 2015 the company raised £9m from the Business Growth Fund, MMC Ventures, Unilever and the Angel Co-Fund. Another £10m followed in 2016, the year Boldt was named IGD's Young Entrepreneur of the Year. In 2019 came £18m with fitness coach Joe Wicks among the backers, then £30m from Perwyn, and £33m in April 2020 as lockdown made cooking at home a national habit. The bill arrived in 2022. Post-pandemic inflation hit sales, and a business that had been valued at £1.4bn that year raised money in February 2023 at under £250m, about 80% lower. Sky News described a governance storm. The company cut jobs and aimed for profit. It got there. For FY2025 Gousto reported revenue of £342m and adjusted EBITDA of £45m, its third consecutive year of underlying profitability. Then in September 2026 came the news that Boldt would step down as CEO after 15 years. The quiet lesson is in the order of things. The investors who said no on television were not wrong about the difficulty; the business needed years and several rounds before it paid. Boldt kept the idea he started with, a better dinner at home, and let the money catch up with it. Rejection in public is rarely the end; it is just one sample, like a bin with three bottles in it. ## PolyAI (United Kingdom) URL: https://startup.talk/uk/polyai/ Founders: Nikola Mrkšić, Tsung-Hsien (Shawn) Wen, Pei-Hao Su Founded: 2017; Based in: London, England; Sector: Conversational / voice AI for customer service Nikola Mrkšić was born in Serbia. Years later he would run a company valued at $750m that teaches machines to talk to people on the phone. That path began with a scholarship. He won a place to study computer science at the University of Cambridge on a scholarship. That detail matters because most of the public story of PolyAI starts in a lab. He did not arrive there as an insider. He arrived as a student from Serbia. At Cambridge's Machine Intelligence Lab he met Tsung-Hsien Wen, who also goes by Shawn, and Pei-Hao Su. The three worked on how computers hold conversations. In 2017 they started PolyAI. The idea was practical. Companies spend huge sums on call centres, and customers hate being stuck in menus. A voice agent that could handle an ordinary conversation would be worth a lot, if it did not collapse the first time someone interrupted it. Money came in stages. A $40m Series B in 2022 was led by Georgian. A $50m Series C followed in 2024, led by Hedosophia with NVIDIA's NVentures, pushing the valuation to about $500m. Then in December 2025 came an $86m Series D at $750m, led by Georgian, Hedosophia and Khosla Ventures, with Citi Ventures, Zendesk Ventures and NVIDIA Ventures also reported. Total funding is cited at over $200m. That sounds smooth. The harder reality is the market. Cloud giants sell contact-centre tools, and a wave of startups is chasing the same budgets. The company has not published revenue or profit. What PolyAI shows is how much of a company's strength can be the team it forms before there is a company. Three people who understood a narrow problem for years in a university lab built something the market now values highly. A founder who arrives from a small country on a scholarship does not need a famous network on day one; the room they get into can supply it. ## Atom Bank (United Kingdom) URL: https://startup.talk/uk/atom-bank/ Founders: Anthony Thomson, Mark Mullen Founded: 2013 (incorporated July 2013; launched 2016); Based in: Durham, England; Sector: Digital bank / fintech Anthony Thomson had co-founded Metro Bank. Then he left it, after, as press accounts put it, opposing its commitment to costly branches. That disagreement is where Atom starts. Thomson has said the idea for a bank with no branches goes back to early 2012. Mark Mullen, who had been chief executive of First Direct, the original telephone bank, left his job in 2014 to join him. The company was incorporated in July 2013. The stakes were unusual. This was not an app that moved money around. It was a bank that needed a licence, capital and regulators who had never granted one to a business without a single branch. The licence arrived in June 2015. In November 2015, Spain's BBVA agreed to invest a substantial stake before launch. Atom opened in April 2016 to people who had registered an interest, and widened out that October. It chose Durham for its headquarters, away from London. By 2019 Beauhurst ranked it second on its list of 50 top UK fintech scale-ups, the only company outside London in the top twenty. In February 2022 it raised £75m at a £435m valuation and talked about an IPO. It also moved all 430 staff to a four-day week without cutting salaries. The price came later. In November 2023 shareholders including BBVA and Toscafund put in £100m at a £362m valuation, below the 2022 mark. Then the turn: in June 2024 Atom reported a £27m operating profit, about seven times the previous year and its first annual pre-tax profit since launch. Borrowing rose 39% to £4.1bn. Revenue for 2024 was £99.1m. FY2025 brought an operating profit of £25m, and Mullen announced his departure after more than ten years as CEO. Thomson, who stepped down in 2018, reportedly plans a new bank for the very wealthy. The human lesson in this story is how long a stubborn idea can take. Thomson walked away from a successful bank because he disagreed with its model, and then spent about a decade proving the opposite. A founder's strongest asset may be knowing exactly what he refuses to build. ## Shopify (Canada) URL: https://startup.talk/canada/shopify/ Founders: Tobias Lutke, Daniel Weinand, Scott Lake Founded: 2004 (Snowdevil); Shopify launched June 2006; Based in: Ottawa, Ontario; Sector: E-commerce software For about ten years, the founder of what would become one of Canada's most valuable companies slept in his girlfriend's parents' house in Ottawa. Tobias Lutke did not plan it that way. He was born in 1980 in Koblenz, in what was then West Germany, and his parents gave him a Schneider CPC computer when he was six. School, it turned out, mattered less to him than the machine. After tenth grade he left to apprentice as a programmer at the Carl-Benz-School in Koblenz. He has said that John Carmack's .plan files, the plain-text updates of the Quake programmer, left a huge impression on him as a teenager. In 2003 he moved to Canada, following love, and settled in Ottawa. A year later, with Daniel Weinand and Scott Lake, he started Snowdevil, an online store for snowboards. The shop was the idea. The trouble was the software: every e-commerce package he tried was clumsy. So Lutke, a programmer who had never finished school, wrote his own, in Ruby on Rails, a young framework most people had not heard of. What almost nobody saw at the time was where the money was coming from. Lutke took no salary while the platform was being turned into a product. He lived with Fiona McKean's parents, in their home, for roughly a decade, which kept his costs close to zero. Her father, Bruce McKean, supplied early seed capital. A founder's courage is easy to admire; a family that quietly pays the rent is easier to overlook. Shopify launched as its own product in June 2006. The snowboards faded; the software stayed. Bessemer, FirstMark and Felicis led a US$7 million Series A in December 2010, followed by US$15 million in 2011 and US$100 million in 2013. The company went public in May 2015, raising more than US$131 million. The years after were not smooth. In 2017 a short-seller attack and a boycott campaign knocked more than 11 percent off the stock in a day. In 2022 Lutke said the company's growth forecasts had not paid off and cut about 10 percent of staff; in 2023 another 20 percent went, and the logistics arm was sold to Flexport. A founder who had been cheered as a prophet had to stand in front of his people and say he had been wrong. The recovery was fast. In 2025 Shopify reported revenue of about US$11.6 billion, up roughly 30 percent, with net income near US$1.2 billion, and Forbes put Lutke's fortune at US$12.3 billion that August. The lesson is less about code than about runway. Lutke's real early investor was a household that gave him time. Before you look for venture money, look at who is already, quietly, buying you years. ## Constellation Software (Canada) URL: https://startup.talk/canada/constellation-software/ Founders: Mark Leonard Founded: 1995; Based in: Toronto, Ontario; Sector: Vertical market software holding company At an age when most venture capitalists are settling into their careers, Mark Leonard walked away from his. He was 39, and by his own account the fund he had spent eleven years inside, Ventures West, had delivered mediocre returns. The odd part is how he got there. A bio prepared for a speaking event at Ivey, where he did his MBA, reportedly lists gravedigger, bouncer, mason and dog handler among his early jobs. Profiles add draftsman and furniture mover, and a student stint as a stonemason. As an intern at Barclays, a senior colleague told him he would make a poor banker. A managing director gave him other advice: study the bank's clients. That nudge pointed him toward venture capital in Canada. Then came the itch. Inside the portfolio, Leonard noticed something the industry ignored. The steady winners were not the glamorous bets. They were small software companies serving one narrow trade, too tiny to move a big fund's returns, so nobody wanted them. What almost nobody knew was that the idea began with a modest name. He proposed a company he called "Software Co.", and in 1995 raised C$25 million, reportedly from OMERS, the Ontario municipal pension fund, and former colleagues at Ventures West. The rule was simple. Buy these businesses, never sell them, and put the cash into the next one. It worked slowly, then enormously. The company listed on the TSX in 2006. In 2013 it bought Total Specific Solutions for US$360 million, which later became part of Topicus.com. Leonard wrote shareholder letters that investors treated like scripture. "Hurdle rates are magnetic in nature," he wrote in one. He avoided interviews so completely that the one he is known to have given was reportedly pulled offline. The price was a certain opacity. In 2018 the company stopped holding quarterly earnings calls, which analysts called highly unusual. In 2016 the founder of Innoprise sued over a Harris unit's free software. Leonard himself stopped drawing a salary in 2016. In September 2025 he stepped down as president for health reasons, staying on the board, with Mark Miller taking over. Revenue reached US$10.07 billion in 2024. He has said his biggest regret is not spending more time with his family. For a first-time founder the takeaway is quiet. Leonard did not invent a product. He looked at a fund's weakest-looking holdings, realised the dull ones were the best, and spent the next thirty years refusing to be talked out of it. ## Cohere (Canada) URL: https://startup.talk/canada/cohere/ Founders: Aidan Gomez, Nick Frosst, Ivan Zhang Founded: 2019; Based in: Toronto, Ontario; Sector: Enterprise generative AI / large language models At twenty, Aidan Gomez was an intern. His name was on one of the most consequential papers in modern computing. In 2017 he was at Google Brain, one of eight authors of 'Attention Is All You Need', the paper that introduced the transformer, the architecture underneath today's chatbots. He was the youngest in the room by a distance, and probably the only one who still had undergraduate exams in his recent past. He came from small-town Ontario. Born in Toronto, he grew up in Brighton, then returned to study computer science and mathematics at the University of Toronto. He later went to Oxford to work on a PhD. At Toronto he worked with Ivan Zhang at FOR.ai, a research collective, and he knew Nick Frosst, a Google Brain researcher, from the same Toronto circle. The pull of a startup was strong, and the cost was real. In 2019 the three founded Cohere, and Gomez paused the Oxford doctorate to do it. He finished the degree only in 2024, five years into running a company, something few chief executives can say. What almost nobody anticipated was how hard it would be to raise money at home. The New York Times reported that Cohere struggled to attract US investors, even as capital poured into rivals. Early backers, though, believed: Index Ventures led a US$40 million Series A in September 2021, and Tiger Global led US$125 million in 2022. By June 2023 the valuation had reached US$2.2 billion, with Oracle, Salesforce and Nvidia on the cap table. It has not been smooth. In February 2025, 14 publishers, including The Atlantic, Conde Nast and Forbes, sued for copyright infringement over alleged unauthorised use of their content. The company has also been pushed toward a deal with Germany's Aleph Alpha, backed by US$600 million from Schwarz Gruppe, which the Times said would value the combined business near US$20 billion. Revenue has grown from an annualised US$22 million in March 2024 to a reported US$240 million in February 2026. The quiet lesson in Gomez's path is that timing can turn a junior into a founder. He did not wait for a title. He simply happened to be in the room, did the work, and then, unusually, built on it himself. ## Lululemon (Canada) URL: https://startup.talk/canada/lululemon/ Founders: Chip Wilson (Dennis J. Wilson) Founded: 1998 (12 January); Based in: Vancouver, British Columbia; Sector: Athletic apparel retail Chip Wilson was twelve when his parents divorced. He later chose a job in Fairbanks over a university desk. Born in Los Angeles in 1955, he moved to Calgary at five. His father had been a Canadian hockey and football player, his mother an American gymnast. Wilson swam competitively and planned to play football at the University of Alberta. Then he put school off, and went to work in the oil industry in Fairbanks, Alaska. He eventually took an economics degree at the University of Calgary in 1980. He had already started his first company, in 1979. Westbeach Snowboard Ltd made clothes for surf, skate and snowboard kids. He sold it in 1997. A year later he started something else. Lululemon Athletica opened in Vancouver in January 1998 as a maker of yoga clothes for women. The first standalone store followed in November 2000. Wilson picked a name with three Ls, partly because he thought it would seem exotic to Japanese buyers, who often struggle with the letter. It is the kind of small, odd bet that nobody could have defended in a boardroom. What the public rarely connects is the personal weight he carried while building it. At 32 he was diagnosed with facioscapulohumeral muscular dystrophy, and he has since donated C$100 million toward research. A company built on the body's strength was founded by a man whose own body would be tested. The company sold 48 percent to Advent International in 2005 and raised US$327.6 million in a Nasdaq IPO in July 2007. Then came the damage. In 2013 black pants turned out to be too sheer, and a recall hit about 17 percent of women's pants sold. The chief executive and product chief left. Wilson's remarks about customers' bodies drew heavy criticism, and he stepped down as chairman that year. He did not fade quietly. In December 2025 he launched a proxy fight over the board, ending in a cooperation agreement in May 2026. Today Lululemon reports revenue of about US$11.1 billion and net income of US$1.58 billion (2025). Wilson owns 8.4 percent. The lesson is one of attachment: a founder can build a brand and still never fully let go of it, and the board may remember that long after he steps aside. ## Hut 8 (Canada) URL: https://startup.talk/canada/hut-8/ Founders: Sean Clark, Andrew Kiguel, Harris Fricker, Marc van der Chijs, Bill Tai, Asher Genoot Founded: 2017; Based in: Toronto, Ontario; company is U.S.-domiciled and listed on Nasdaq and TSX; Sector: Bitcoin mining, energy infrastructure and AI compute The man who became the face of a bitcoin miner spent almost two decades selling shares, not mining coins. Andrew Kiguel joined Hut 8 as President and CEO on 2 April 2018. Before that he had spent more than 18 years at GMP Securities in Toronto, mostly in real estate and retail capital markets, and had raised over C$3 billion for clients in his last five years there. By the end he sat in GMP's blockchain investment banking group, one of the few bankers in the country with a crypto desk. That detail matters because of what Hut 8 was built to be. In October 2017, with bitcoin in a frenzy, Sean Clark, Harris Fricker, Kiguel and Marc van der Chijs conceived a company to give ordinary investors exposure to bitcoin without buying the coin. Their vehicle was a listed miner. Hut 8 was incorporated on 15 November 2017, with Clark as interim CEO. The name reportedly nods to Hut 8 at Bletchley Park, where Alan Turing's team worked on the German naval Enigma. What almost nobody noticed was how thin the founders' mining credentials were. The machines came through a partnership with Bitfury Group, which supplied data centres, and the first site went up in Drumheller, Alberta. Michael Novogratz, through Galaxy Digital, contributed mining assets for equity. Bill Tai, a venture investor, became chairman and said the aim was to be "the leader in North American mining." Kiguel took over when Clark stepped back at the end of March 2018, and said he wanted the company to be the world's leading crypto-currency miner. The timing was brutal: bitcoin fell through 2018. He stayed until 2020, when the board began a search for his successor. The harder years came later. Bitcoin's 2022 slump pushed the company toward a merger of equals with US Bitcoin Corp, signed on 6 February 2023 and closed on 30 November 2023. Asher Genoot, from the USBTC side, took over as chief executive in early 2024. The merged company, listed on Nasdaq and the TSX, began describing itself as an energy infrastructure platform, adding high-performance computing and AI data centres to mining. Its value was near US$13 billion in November 2026. The lesson is plain. Hut 8 began as a financial product wrapped around a technology, and its early founders understood the wrapper. When bitcoin turned, it was the power contracts and sites underneath that carried the company into a second life. ## OpenText (Canada) URL: https://startup.talk/canada/opentext/ Founders: Anouar Namouh, Frank Tompa, Gaston Gonnet Founded: 1991 (OpenText Systems Inc. founded 1989); Based in: Waterloo, Ontario; Sector: Enterprise information management software Before OpenText was a software giant with more than twenty thousand employees, it was a puzzle about a dictionary. The Oxford English Dictionary is not a tidy book. It is a sprawling, layered record of words and their histories, and in the 1980s someone at the University of Waterloo had to work out how to make all of it searchable. Two professors, Frank Tompa and Gaston Gonnet, were part of that effort. Nobody was thinking about enterprise software. They were thinking about how to find a word inside a mountain of text. That is the quiet fact at the start of OpenText. In 1991 an entrepreneur, Anouar Namouh, founded OpenText Corporation alongside the two professors, spinning the dictionary technology out of the university. A related company, OpenText Systems, had been set up two years earlier. The founders do not appear to have left much public record of their childhoods or schooling, so the story stays with the idea: if you can index a dictionary, you can index anything a company owns. The early company was a search business, and search was not yet a business that paid. What changed things was people. Tom Jenkins joined as chief operating officer in 1994 and later became president and chief executive. In 1996 the product known as Livelink became OpenText's, and the company went public that year. Livelink pointed the technology at a more profitable problem: how big organisations store, find and share their documents. What almost nobody outside the industry noticed was the second act. OpenText started buying. Hummingbird in 2006 for US$489M. Vignette in 2009. Then larger: Dell EMC's Documentum division for US$1.6B in 2016, Carbonite and its Webroot and Mozy units for about US$1.45B in 2019, and Micro Focus, a deal of roughly US$6B, completed in January 2023. The pattern was consistent. Buy mature software with loyal customers, fold it in, cut cost. The approach has a price. In 2022 the company suspended its Russian business after donating to Ukrainian humanitarian efforts, and said it would resume only when the war ended and sanctions lifted. In 2025 Russia declared it an undesirable organisation. That same year the chief executive of thirteen years, Mark Barrenechea, was replaced; James McGourlay stepped in on an interim basis in August, and Ayman Antoun was named CEO in April 2026. Today OpenText reported revenue of US$5.17B for fiscal 2025, with net income of US$436M, both down on the year, and its value was near US$11B as of November 2026. The lesson for a first-time founder is a small one. The company's biggest assets, a dictionary project and a knack for absorbing other firms' software, were not glamorous. OpenText was built by being useful to boring customers for thirty years, and by taking an academic curiosity seriously enough to turn it into a company. ## Wealthsimple (Canada) URL: https://startup.talk/canada/wealthsimple/ Founders: Michael Katchen, Brett Huneycutt, Rudy Adler Founded: 2014; Based in: Toronto, Ontario; Sector: Fintech, investing and banking The first Wealthsimple was a spreadsheet. By Michael Katchen's telling, he built it in a single night, and it was meant for ten friends. Those friends had just come into money. Katchen had moved to Silicon Valley in 2011 to join a photo-scanning startup called 1000memories, and in October 2012 Ancestry.com bought it for several million dollars. Colleagues who had cashed out asked him how they should invest the proceeds. He had a long answer. He had, he told the Globe and Mail, been investing since he was 12. That itch started early. A business degree from Ivey, graduating 2009, led to a job as a strategy and marketing consultant at McKinsey in Toronto, and then the leap to California. The spreadsheet taught his friends to buy low-cost exchange-traded funds instead of paying the high fees Canadian banks charged. "The very first idea I had was not to create a business," he told the Globe. Then he looked at what he had built. He moved back to Toronto in 2013, partly for personal reasons and partly because almost nobody offered robo-advice in Canada. Wealthsimple was founded in September 2014 with Brett Huneycutt and Rudy Adler. The first outside money was C$1.9M from three Canadian investors, and in April 2015 Power Financial put in C$10M. The early days were small. In a podcast he recalled that a good week meant winning one client, and he phoned new sign-ups himself within seconds of their signing up. By 2016 the company had about 20,000 clients and 40 staff, and was not yet profitable. What almost nobody expected was how much the product would matter more than the robo-adviser. In March 2019 Wealthsimple Trade launched, Canada's first commission-free trading app. When markets convulsed in March 2020, the app buckled: demand was so high that the company capped signups and waitlisted investors. It was an awkward way to learn that popularity is its own stress test. The cost showed up in the products themselves. Wealthsimple has changed its lineup repeatedly, turning Cash from a savings account into a peer-to-peer payments app and retiring Save and Smart Savings in 2025. The company's valuation rose through the cycle: C$5B in May 2021, and C$10B in October 2025 when it raised C$750M, led by Dragoneer and GIC. In October 2022 it became Canada's first non-bank, non-credit union approved for a Bank of Canada settlement account. By June 2026 it reported about 3.6 million clients and C$156B in assets. The company's origin is still visible in its habits. It started as a favour, a clear explanation for ten people who trusted the person explaining. A founder could do worse than ask: whom am I already helping for free, and what would happen if I charged less than anyone else? ## 1Password (Canada) URL: https://startup.talk/canada/1password/ Founders: Dave Teare, Roustem Karimov, Sara Teare, Natalia Karimova Founded: 2005; Based in: Toronto, Ontario; Sector: Cybersecurity, password management The first version of 1Password was not called 1Password. It was 1Passwd, and it was supposed to take three months. That is how the company describes it: a quick side project, built by two friends who were tired of remembering passwords. Dave Teare and Roustem Karimov were colleagues in corporate software, building websites for their day jobs. Both had recently moved to the Mac, and the tool they wrote for themselves was a Mac app. In 2005 they founded AgileBits, and on June 18, 2006 the first release went out. They are private people, and the story that survives is about what they did: a pair who left the corporate path and decided to strike out on their own. Sara Teare and Natalia Karimova are also named as co-founders, and the two women were central to running the early company. The next fourteen years are the part that nobody in venture capital tends to hear about. AgileBits did not raise a cent from outsiders. The product grew by being good: version 7 introduced a subscription in 2016-2018, Travel Mode arrived in 2017, and Jeff Shiner joined as CEO in 2012. The first outside investment came in November 2019, when Accel put in US$200M, reportedly its largest single cheque at the time. Then the numbers jumped. A US$100M round at a US$2B valuation came in 2021. In January 2022 a US$620M Series C led by Iconiq Growth set the value at US$6.8B, with Ryan Reynolds, Robert Downey Jr. and Justin Timberlake among individual backers. The trade-off of being a password company is that your mistakes are public. In 2017 Cloudflare's Cloudbleed bug raised alarms, and a German research institute found flaws in the Android app, which were fixed. Removing local vault storage in version 8 in 2021 angered loyal users. In 2023 a customer-support incident at Okta touched 1Password, which said no user data was compromised. By September 2023 annual recurring revenue passed US$250M, and in November 2025 David Faugno, the CEO, told CNBC it was above US$400M and that an IPO in 2026 or 2027 was under consideration. What a founder can take from it is that a three-month side project can be a company if it fixes something you personally hate. The two men who built it did not wait for permission or capital. They shipped for a niche, the Mac, and let customers pay for the next fourteen years of work. ## Tenstorrent (Canada) URL: https://startup.talk/canada/tenstorrent/ Founders: Ljubisa Bajic, Milos Trajkovic, Ivan Hamer Founded: 2016; Based in: Toronto, Ontario; Sector: Semiconductors, AI chips (RISC-V) Ljubisa Bajic turned up late. He was carrying a prototype in what he later called a big, brown ragged box, and the man waiting for him at Tesla was his old boss, Jim Keller. Bajic did not start out as a chip person in Toronto. He grew up in Serbia, then part of Yugoslavia, and finished high school in Moscow, drawn there by its mathematics programmes. In 1985 he moved to Canada to study electrical engineering, then went to Silicon Valley to design processors. He worked at VLSI Technology and Nvidia, and in 2014 joined AMD, where he worked under Keller. Then Keller left for Tesla. Bajic stayed with an itch of his own: a chip built for machine learning. Keller had told him that if he came up with something cool, he would fund it. The box held Bajic's answer. Keller was impressed, and he told Fortune it would be funny if Bajic and a couple of others lived in a basement for a year building the thing. They did. Tenstorrent was founded in 2016 with Milos Trajkovic and Ivan Hamer, and for about a year the team worked below ground before moving into offices in Toronto. Bajic joked to Fortune, on a Zoom call from that same basement, that he had become used to low light. What almost nobody outside chip circles knew is how long the angel stayed an angel. Keller invested early, and only joined as president and CTO in 2021, the same year Tenstorrent raised more than US$200M from Fidelity and others at a US$1B valuation. The first chip, Grayskull, slipped because the team chose to build client software before shipping it. Then the roles flipped. In January 2023 Keller became CEO and Bajic became president and CTO, with Bajic saying he wanted full focus on the technology. Four months later he stepped back from daily duties, staying on the board and as an advisor. The company gave no reason, only that he would pursue new projects. By then Tenstorrent had nearly 300 employees and two chip generations taped out. The money kept coming. In December 2024 a Series D of more than US$693M, led by Samsung Securities and AFW Partners with Bezos Expeditions in the round, valued it at about US$2.6B after the money went in. Total funding was close to US$1B, and by October 2026 the company was valued at about US$6.2B. A founder can take one thing from the box. Bajic did not pitch a deck to his old boss. He carried the thing itself into the room, late, and let a person who knew chips decide. ## PointClickCare (Canada) URL: https://startup.talk/canada/pointclickcare/ Founders: Mike Wessinger, Dave Wessinger Founded: 1995-2000 (company was more than 20 years old in 2021); Based in: Mississauga, Ontario; Sector: Healthcare software for senior care On August 10, 2021, one brother gave the top job to the other. Mike Wessinger, who had run PointClickCare as founding CEO, moved to executive chair. Dave Wessinger, his co-founder and brother, became chief executive on September 1. Dave described it as orderly succession and good governance. Mike said he was incredibly proud of what he had grown. It was a calm announcement for a company that began with something very small: a clinical documentation tool built for one long-term care facility in Mississauga, Ontario. Nobody in that building was shopping for a platform. They needed to record what was happening to patients. The brothers were not a Silicon Valley pair. Both have undergraduate degrees from the University of Western Ontario, and Mike also started a separate software firm, Wescom Solutions, in 2000. The exact founding year varies by profile, but the company was already more than twenty years old when the handover came. Over those years it chose a market most software firms avoided: care homes, where staff are stretched and records still lived on paper. Dave worked as chief operating officer and president before the top job. A board chair, Bill Dillane, stayed in the seat for more than twenty years, a rare kind of loyalty in a venture-backed company. The outside money was patient too. Dragoneer invested in 2017 and 2018. In January 2021 Hellman & Friedman took a minority stake while the founders kept control, at a valuation reported as US$4B. By the time of the handover PointClickCare served more than 21,000 care providers and over 1,300 hospitals, with about 1,500 staff, most of them in Canada. What almost nobody noticed in the announcement is how little drama it held. The founder who built the product stepped sideways into a role facing outward, and a brother who had spent decades on operations took the wheel. The company was valued at around US$5B in November 2026. The lesson for a founder is about the end of the story, not the start. The brothers had one customer with a real problem and years of patience. And when it was time to change roles, they did it on a date they chose, in public, with the board still in place. ## Hopper (Canada) URL: https://startup.talk/canada/hopper/ Founders: Frederic Lalonde, Joost Ouwerkerk Founded: 2007; Based in: Montreal, Quebec; Sector: Travel technology / fintech Two Expedia executives walked away from one of the world's biggest travel companies in April 2007 to build something that did not yet make money. It was not even about flights. Frederic Lalonde and Joost Ouwerkerk had spent years inside the machine of online travel. They knew how trips were sold, and they knew how clumsy planning still felt. So they started Hopper in Montreal as a tool for working out where to go and what to do when you got there. A planning site is a hard thing to turn into a business. People browse it, enjoy it, and book somewhere else. That is the part of the story that rarely gets told. The company everyone now knows for airfares spent its early years as something else, and the founders had to let go of their first idea to find a better one. Its effect is clear. The product that launched in January 2014 and the app that arrived in 2015 were built around one question: should I buy this ticket now, or wait? Answering it meant predicting prices. Hopper's app watched fares, sent alerts, and told people when to move. Investors followed. A US$12m Series B in 2012 from OMERS and Atlas Venture kept the lights on, and by 2016 and 2018 the rounds were US$62m and US$100m. Hotels came in 2017. Then fintech features, which let customers lock a price for a fee, turned worry into revenue. What almost nobody outside the industry noticed was how much the company leaned on the very anxiety it soothed. Capital One put in US$170m in March 2021 and Hopper became a unicorn. By February 2022 it was valued at US$5bn. The price came later. In 2018 a Secret Fares programme angered Air Canada and WestJet, who said the fares were falsely sold as exclusive. Hopper shut it down. In July 2026 it agreed to pay US$35m to settle FTC claims that it charged fees without consent and misled users about them. The business that sold certainty was being asked to explain its fine print. Today Hopper reports revenue of about US$850m (2024) and runs a B2B arm selling its travel technology to banks and airlines. For a first-time founder, the lesson is in the pivot hiding at the start. Lalonde and Ouwerkerk had the credentials to defend their first idea. They chose the one customers actually paid for. ## Clio (Canada) URL: https://startup.talk/canada/clio/ Founders: Jack Newton, Rian Gauvreau Founded: 2008; Based in: Burnaby, British Columbia; Sector: Legal technology (cloud software) Before Clio, there was a childhood friendship and a very ambitious company name: Newton-Gauvreau Aerospace. That was the first venture of Jack Newton and Rian Gauvreau, two boys who, by the company's own telling, met on a playground. It is easy to smile at the name. It is harder to ignore what it says. These two had been planning to build something together long before either knew what a law firm needed. The idea that stuck came later. In 2008 they set out to put law firms in the cloud. At the time that sounded close to reckless. Lawyers guard client files with near-religious care, and the profession was not known for rushing toward new technology. Asking a partner to move billing and case records onto servers they could not see was a hard sell. The founders have said little in public about their early money worries, and what exists online does not let us tell that story honestly. What can be said is that they stayed with the problem for years. By March 2014 they had raised US$20m, which paid for the long, unglamorous work of convincing firms one by one. What almost nobody saw coming was how far that patience would carry them. Cloud software became ordinary for lawyers, and Clio, built on a single idea from 2008, became the place many firms ran their practice. Then came the bigger bet. In November 2025 it completed a US$1bn acquisition of vLex, whose AI assistant Vincent works over a vast legal database, and raised US$500m led by NEA at a US$5bn valuation. The cost of growing that fast shows in the numbers. Joining 350-plus vLex employees to a company of about 2,000 people means two cultures sharing one product. For a first-time founder, the useful detail is the oldest one. Newton and Gauvreau did not meet through an accelerator. They were friends first, and they had already practised being partners on a project that never left the ground. The company that did take off was built by two people who knew exactly how the other worked under pressure. ## Nuvei (Canada) URL: https://startup.talk/canada/nuvei/ Founders: Philip Fayer Founded: 2003; Based in: Montreal, Quebec; Sector: Payments / fintech He was 24 when he started the company, and he had already lost his father at 16. Philip Fayer's early life, as that profile tells it, was unsettled: born in Montreal, raised partly in Belgium and Israel, then sent to a military school in the United States. It is a secondary account and should be treated with care. But it fits a pattern the later record supports. Fayer was a young man in a hurry. He started Pivotal Payments in 2003 while taking time away from Concordia University. He never went back. Payments is a brutal trade to enter: margins are thin, banks decide who gets processed, and merchants leave the moment a competitor undercuts them. He had to build trust in an industry that had little reason to give it. Goldman Sachs invested in 2006, a vote of confidence for a business still small by Wall Street standards. Years later came the decision that defined him. A rival wanted to buy the company, and Fayer turned the sale down. He took money from Novacap and CDPQ in 2017 instead and kept control. What almost nobody knew until the share price began to climb was how personal that choice was. A sale would have made him wealthy at once. Staying meant years more risk. Pivotal became Nuvei in 2018, bought SafeCharge in 2019, listed in Toronto in 2020 and Nasdaq in 2021. The fall-out arrived on the same schedule as the success. Short seller Spruce Point attacked management and accounting in December 2021 and again in April 2023. Nuvei rejected the claims, and a related class action was discontinued in February 2023. In September 2026 the FTC filed a complaint over merchant screening. In November 2024 Advent International took the company private for about US$6.3bn, or US$34 a share. Fayer kept a stake of about 24% and stayed as chair and CEO. Revenue was roughly US$1.19bn in 2023. For a first-time founder, the lesson sits in that turned-down offer. Fayer left a university without a degree, but he understood early that the real prize was owning the outcome, not cashing out of it. ## CoolIT Systems (Canada) URL: https://startup.talk/canada/coolit-systems/ Founders: Geoff Lyon Founded: 2001; Based in: Calgary, Alberta; Sector: Data-centre liquid cooling When KKR bought CoolIT Systems in 2023, the staff were handed a share of the outcome. A Calgary law firm that acted on the deal noted that employees became owners through a broad-based ownership programme. In a sale of a technology company, that is unusual enough to mention. It also hints at something about the place: a company that had spent more than twenty years in a quiet corner of computing, where people matter more than slogans. CoolIT started in Calgary in 2001. Its business was moving heat away from processors, first with liquid loops for desktop computers and then for the data centres that run the internet. What the record does show is persistence. This was not a product that seemed urgent when it began. Air cooling was cheap and familiar, and most data-centre operators saw no reason to put liquid near expensive electronics. Then the load changed. AI chips draw far more power than the servers before them, and the heat they throw off became a design problem for every operator. Direct-to-chip liquid cooling, once a specialty, started to look like infrastructure. KKR and Mubadala completed the purchase in July 2023. One outlet reported a price of US$365m, though terms were never officially published. Three years later the picture had shifted. Ecolab agreed to buy CoolIT for about US$4.75bn, with closing expected in the third quarter of 2026. A law firm involved called it one of the largest technology deals for a Calgary company. The gap between those two figures says something blunt about timing. The technology barely changed. The world around it did. For employees who held shares after 2023, the story ends with a payoff their founders could scarcely have scheduled. For anyone starting a company today, the takeaway is an uncomfortable one. Being early can look the same as being wrong for a very long time. The founders who survive are the ones who can afford to wait, and the ones who remember to share the reward with the people who waited with them. ## SSENSE (Canada) URL: https://startup.talk/canada/ssense/ Founders: Rami Atallah, Firas Atallah, Bassel Atallah Founded: 2003; Based in: Montreal, Quebec; Sector: Luxury fashion e-commerce The idea began with a pair of jeans. Three brothers sold a limited-edition pair of Diesel jeans online, and the speed of the sale told them something they could not unlearn. Rami, Firas and Bassel Atallah were born in Damascus, Syria, and came to Montreal with their family as teenagers. By Rami's own account, quoted in a Globe and Mail profile, they were immigrants with no experience of fashion who had to learn everything from scratch. That is not modesty. It is a description of the stakes: no insider contacts, no family in the trade, no safety net. They opened in 2003 and began with a physical store in Montreal. The online shop came in 2006. Early work was done by hand: Rami photographed the products for the site, and Firas carried orders to the post office. A brand known for cool restraint began in a very unglamorous place. What almost nobody noticed at the time was how much the physical store taught them. Customers told them what they wanted, and the brothers turned that into a buying instinct. SSENSE grew into a multi-brand retailer shipping to 114 countries, with a flagship in a 19th-century Montreal building designed by David Chipperfield's firm in 2018. The fall-out came in pieces. That same year it bought the community site Polyvore and shut it down, wiping users' data, and later apologised. A cyberattack in 2022 exposed employee data. Sequoia's 2021 investment had put its value above US$4bn, by one account, but layoffs followed in 2023, affecting 138 people. Then tariffs. In 2025 SSENSE sought creditor protection with about C$371m in debt and put itself up for sale. In early 2026 a court accepted a bid from the three brothers to keep ownership. That last fact is the quietest one in the story. The company they had built from a single online sale was at risk of being taken from them, and they fought to hold it. For a first-time founder, the point is that inexperience is not the same as weakness. The brothers started with nothing to protect, so they learned by doing every job, and they refused to hand the result over cheaply. ## Xanadu (Canada) URL: https://startup.talk/canada/xanadu/ Founders: Christian Weedbrook Founded: 2016; Based in: Toronto, Ontario; Sector: Photonic quantum computing The money arrived in the spring of 2026, and it arrived smaller than planned. Xanadu had told the market to expect about US$500 million from its merger with a blank-cheque company called Crane Harbor. Redemptions cut that to roughly US$302 million. On 27 March the company began trading in Toronto and New York under the ticker XNDU, valued at about US$3.1 billion. For a firm that had spent a decade selling a future, that was still a big day. It began in 2016 with Christian Weedbrook, a quantum physicist who started Xanadu in Toronto and went through the Creative Destruction Lab, the University of Toronto accelerator that has fed many Canadian science companies. What is documented is the bet. Most of the quantum field chose cold. IBM and Google cooled superconducting circuits to near absolute zero. Xanadu chose photons, particles of light moving through chips, which promised hardware that could run at room temperature and be networked with fibre. It was a contrarian choice and a slow one. In 2020 the team published a blueprint for a fault-tolerant photonic machine, a design on paper for something nobody could yet build. Proof came in pieces. In June 2022 the company reported a boson sampling experiment it said ran 50 million times faster than earlier ones. In January 2025 it published in Nature a scalable, modular way to link photonic quantum computers. Investors followed the papers: Bessemer, Tiger Global, In-Q-Tel, OMERS Ventures and Radical Ventures, plus grants from DARPA and Sustainable Development Technology Canada. Total funding stood near US$245 million before the listing. The other asset was software. Xanadu runs PennyLane, an open-source library that quantum programmers use no matter whose hardware they target. That gave a hardware startup something rare in the field: a community that already knew its name. The public debut also brought a cost. A SPAC merger exposes a company to the mood of the market, and the shortfall against the original US$500 million target showed how fast that mood can move. One report also mentioned a possible investment of up to C$390 million from the governments of Canada and Ontario, pending final agreements. The lesson for a founder is about time. Weedbrook picked a hard path in 2016 and spent ten years publishing results rather than promising products. Deep-tech patience is not a virtue until a market is ready to pay for it, and the listing was the first real test of whether the market was. ## ApplyBoard (Canada) URL: https://startup.talk/canada/applyboard/ Founders: Martin Basiri, Meti Basiri, Massi Basiri Founded: 2015; Based in: Kitchener, Ontario; Sector: Education technology, international student recruitment Before ApplyBoard was a company, it was a favour from one brother to two others. Martin Basiri, the eldest, had left Iran for Waterloo to study mechatronics engineering. When his twin brothers Meti and Massi, still at home, wanted to follow, he walked them through the Canadian application process himself. It was slow, confusing work, and he did it by hand. The family background, as described in profiles, was modest. Their parents were educators. Martin was the first to look abroad, and he pushed the twins to come. They arrived a year after him and enrolled at Conestoga College, while Martin studied at the University of Waterloo. What nobody planned was that the favour would become a business. The brothers had met every obstacle an international student meets: unfamiliar visa rules, grade conversions that nobody could explain, applications that went to one school at a time. In 2015, in Waterloo, they founded ApplyBoard to put all of it on one platform. They soon moved into the Velocity Garage, the University of Waterloo's startup space. The idea worked because it served two sides. Students got one application route. Schools got a filtered stream of candidates through recruitment partners. By 2019 the team had about 170 people, and in 2019 it launched ApplyProof, a way to verify acceptance letters and English test results. In 2020 it opened in the UK. By June 2021 headcount had reached around 1,000 and the company was valued at US$3.2 billion after raising roughly US$475 million. The price came later. Critics said the company did not supervise its agents closely enough, and that some applicants were given empty promises or biased advice. Meti Basiri said in 2022 that the company met the regulations. That November it laid off 6 percent of staff. By 2024 its valuation had fallen about 74 percent to under US$1 billion, which the company called a moment in time. In 2025, with fewer student visas being approved, it cut more than 150 jobs. The business had been built on a government dial it did not control. When Canada tightened international student numbers, the demand that carried ApplyBoard up to a unicorn valuation shrank with it. For a first-time founder the story is plain. The best product ideas can come from helping one person you love through a bad process. But a platform that sits on top of policy inherits the policy's risks, and the brothers learned that at scale. ## Waabi (Canada) URL: https://startup.talk/canada/waabi/ Founders: Raquel Urtasun Founded: 2021; Based in: Toronto, Ontario; Sector: Autonomous trucking and self-driving AI Raquel Urtasun was born on 30 January 1976 in Pamplona, Spain, and did not arrive in Canada until 2014. She was thirty-eight, a professor with a doctorate from Switzerland, two postdocs behind her and years at a Chicago institute. Seven years after landing she was running one of the best-funded startups in the country. The path was long. A degree in telecommunication engineering in Navarra in 2000. A PhD at EPFL in 2006. A postdoc with Trevor Darrell at MIT and then at Berkeley. Then Toyota Technological Institute at Chicago, where she taught from 2009 until she moved to the University of Toronto. There she co-founded the Vector Institute alongside Geoffrey Hinton, and became one of the more visible faces of Canadian AI. In 2017 Uber hired her as chief scientist and head of R&D for its self-driving unit. It was a front-row seat to how hard the problem is. What she did next is the record: she left Uber in 2021 and started Waabi. The pitch was a different recipe. Rather than collect ever more road miles, Waabi would build its driver inside a simulator and use generative AI to make the testing realistic. The first big cheque, US$83.5 million in June 2021 led by Khosla Ventures, was the largest Series A in Canadian history. In 2022 came Waabi World, the simulator itself. Partners followed the logic. Nvidia worked with it from March 2024. A US$200 million Series B that June was led by Uber and Khosla. In February 2025 Volvo Autonomous Solutions signed on, and in October 2025 the Volvo VNL autonomous truck was launched. In January 2026 came a US$750 million Series C and a separate US$250 million from Uber linked to deploying 25,000 robotaxis. The price of a big bet is time. Autonomous driving has swallowed billions and humbled many companies, and Waabi is still trying to prove at scale what its simulator promises. Urtasun has been recognised along the way: a place on Time's 2023 list of influential people in AI, the Order of Ontario, a Fellowship of the Royal Society in 2026. For a founder, the lesson is about the late start. She did not build a company in her twenties. She built one after she had spent decades learning a problem and watched a giant fail to solve it the usual way. Experience was the asset. ## LayerZero (Canada) URL: https://startup.talk/canada/layerzero/ Founders: Bryan Pellegrino, Ryan Zarick, Caleb Banister Founded: 2021; Based in: Vancouver, British Columbia; Sector: Blockchain interoperability protocol Almost every number about LayerZero arrived with a correction attached. In April 2023, when the company raised US$120 million at a US$3 billion valuation, one outlet placed it in New York. CoinDesk later corrected that: the company is Vancouver-based, a detail BetaKit also confirms. A founder can build a global protocol from a rainy Canadian city, and the press can still lose track of where. The sources available to us say little about the people. Bryan Pellegrino is co-founder and CEO, Ryan Zarick is co-founder and CTO, and Caleb Banister is the third co-founder. What is documented is how fast the money moved. LayerZero Labs was founded in 2021. By March 2022 it had raised US$135 million at a US$1 billion valuation, with Andreessen Horowitz, Sequoia and FTX Ventures among the co-leaders. Just over a year later came the Series B at three times that valuation. Reports differ on who led it: a16z crypto is named by Fortune, while others credit OpenSea Ventures and The Spartan Group. The product explains the appetite. Blockchains cannot natively talk to each other, and the early fix, bridges that locked assets in a vault on one chain and minted copies on another, was a repeated target for hackers. LayerZero pitched a lighter messaging layer that developers could plug into their own applications and set their own security rules around. The record since is thinner. A US$55 million investment was reported in May 2025, and a strategic round followed in February 2026. The lesson here is a quiet one about claims. In a field where valuations are headlines, the founders who last are the ones whose basics hold up under checking: where the company is, who led the round, what it actually does. LayerZero's technical bet made it a top-twenty Canadian-linked startup by funding. Whether it endures depends on the same boring thing every infrastructure company depends on, which is being reliable when other people build on top of it. ## Verafin (Canada) URL: https://startup.talk/canada/verafin/ Founders: Jamie King, Brendan Brothers, Raymond Pretty Founded: 2003 (incorporated March 2003); Based in: St. John's, Newfoundland and Labrador; Sector: Anti-fraud and anti-money-laundering software for banks Three months in, the students were worried. They were engineers with no business background, working on a research project paid for by a man named David Kelly, who had already built and sold a cash-processing software company. What if he simply took what they found and built the business himself? They were Jamie King, Brendan Brothers and Raymond Pretty, all Memorial University engineers in St. John's. Before any of this they had been building pattern-recognition software to help robots navigate underground mines, and had incorporated a company for it. Mining was a modest market. King, then a PhD student, had been picked by Kelly for the project over a Rhodes scholar, because Kelly liked his drive. Kelly's idea came after 9/11 and the USA PATRIOT Act, which forced banks to hunt for suspicious money. In October 2002 he proposed a research project on suspicious transactions to Memorial's business dean, with US$50,000 and equipment from his firm Jambrex. He suggested the students aim their mine-robot pattern recognition at money laundering. They saw a market far bigger than mining. Then came the fear. Kelly and the professor Bob Richards settled it in the students' favour: they offered the IP and 75 percent ownership, with 10 percent for the university and 15 for Kelly. By March 2003 Verafin Inc. was incorporated with Kelly, King and Kristine Trask as directors, working out of the university's Genesis Centre. The launch was quiet. A May 2003 presentation in New York brought no sales. After searching widely, an advisory board member pointed to the Newfoundland and Labrador Credit Union, which became customer number one and a testing ground. Killick Capital put in C$1 million. Later the Royal Bank's venture arm added US$5 million to move the company toward subscription revenue. The pay-off came in two steps. In late 2019 the company announced a C$515 million financing, described as the largest venture round in Canada at that time. In November 2020 Nasdaq agreed to buy it for US$2.75 billion. At the sale it had about 600 employees, more than 3,000 bank and credit union customers and annual revenue above US$140 million. The human lesson sits in the second quarter of 2003. The students could have walked away from a sponsor they did not fully trust, or stayed quiet. The deal they ended up with, their own IP and three quarters of the company, is the reason a Newfoundland incubator produced a Canadian unicorn. Ask for the ownership conversation early. It was awkward, and it was worth billions. ## StackAdapt (Canada) URL: https://startup.talk/canada/stackadapt/ Founders: Vitaly Pecherskiy, Ildar Shar, Yang Han Founded: 2013; Based in: Toronto, Ontario; Sector: Advertising technology (programmatic advertising platform) On the day StackAdapt's backers put US$235 million into the company in early 2025, the number that travelled furthest was not the cheque. It was the valuation, near US$2.5 billion, attached to a business that had begun as a small Toronto startup with three co-founders and a narrow idea: make buying online ads less of a black box. Vitaly Pecherskiy, Ildar Shar and Yang Han started StackAdapt in 2013. The ad industry they walked into was not short of giants. Google, Amazon and The Trade Desk already controlled the pipes through which brands bought attention, and a Toronto company had no brand, no budget and no history to lean on. What it had was software, and a belief that marketers wanted a single place to run native ads, video and, later, connected-TV campaigns without hiring an army of specialists. That bet looks obvious now. For years it was a slow grind. Programmatic advertising is a business of trust and small margins, where an agency will test a platform with a tiny budget and only scale spend if the numbers hold. Each customer had to be won campaign by campaign. The founders kept the platform self-serve while building support teams around it, a choice that let smaller advertisers in without losing the larger ones. The money followed the proof. In 2022 Summit Partners invested US$300 million, a vote of confidence from a growth investor that does not usually write cheques for ideas. Then came the Teachers' Venture Growth round in February 2025, which gave the company a war chest and, reportedly, some secondary liquidity, though StackAdapt declined to confirm that detail. The numbers behind the headline are the unusual part. Reports in early 2025 said StackAdapt was expected to pass US$500 million in revenue and about US$125 million in operating earnings that year, and the company told BetaKit those figures were within range. A profitable ad-tech platform of that size, built outside Silicon Valley and New York, is rare. There is no villain in this story and no dramatic collapse, which is itself the lesson. StackAdapt grew in a market where the largest competitors set the rules, and it did so without turning to a public listing or a rescue round. The founders chose a boring-looking niche, stayed close to the advertisers who paid the bills, and let the work compound for more than a decade. For a first-time founder the takeaway is plain. You do not need a category nobody has seen before. Sometimes you need a crowded one, a product that is easier to use than the incumbents', and enough patience to be the vendor customers still trust after the novelty has gone. The headline is still the Toronto roots: three co-founders who built one of Canada's most valuable private tech companies. ## DECIEM (Canada) URL: https://startup.talk/canada/deciem/ Founders: Brandon Truaxe (born Ali Roshan), Nicola Kilner Founded: 2013; Based in: Toronto, Ontario; Sector: Beauty and skincare Brandon Truaxe was born with a different name. Ali Roshan came into the world in Tehran in 1978, and in 1995, as a teenager, he arrived in Toronto with his family as a refugee. The name he later used on every bottle of skincare he made was one he adopted in Canada. That detail rarely comes up in the story of DECIEM, the company behind The Ordinary, which Estee Lauder would eventually value at US$2.2 billion. The usual version starts with the products. The real one begins with a newcomer in a new country, studying computer science at the University of Waterloo, graduating in 2001, and landing an internship at a large cosmetics company. What he saw there stuck with him: the markups. Creams that cost pennies to make were sold for the price of a dinner. He did not go into software. He started a software company and a supplements company, both of which were later dissolved. Then, in 2003, he launched a skincare brand called Euoko, and in 2009 another called Indeed Labs. He left both. Each one taught him something about the industry, and each one ended without him in charge. DECIEM, founded in 2013 as an umbrella for several brands, was his third attempt at a beauty company, and this time he built it around a simple argument: make the lab, the factory and the shop yourself, and you can sell serious ingredients at honest prices. In August 2016 he launched The Ordinary. Shoppers could buy a bottle of a single active, read exactly what was in it and pay a fraction of what department-store brands charged. The line sold out again and again. By the time of the Estee Lauder investment in June 2017, a 28% stake for US$50 million, the company had grown to roughly thirty stores of its own. The success was unusually loud, and so was what came next. In 2018, accounts of erratic behaviour spread. In October of that year Truaxe ordered the company's operations closed, alleging financial crimes, and was removed as CEO. A restraining order and an investigation by PricewaterhouseCoopers followed. In January 2019, he died in Toronto at 40. Estee Lauder called him a true genius. The company he left behind did not stop. Nicola Kilner, his co-founder, ran it, sales continued, and in February 2021 Estee Lauder agreed to take its stake to 76% in a deal valuing DECIEM at US$2.2 billion. For the year ended January 2021 the company reported about US$460 million in net sales. In 2024, Estee Lauder bought the rest, bringing its total outlay to roughly US$1.7 billion. For a founder, the lesson is uncomfortable. Truaxe saw clearly what the industry hid and built something customers loved. But a company that depends on one person's vision and one person's temper is fragile. DECIEM survived because others were able to keep running it. ## Alphawave IP (Canada) URL: https://startup.talk/canada/alphawave-ip/ Founders: Tony Pialis Founded: 2017; Based in: Toronto, Ontario (listed in London); Sector: Semiconductor connectivity IP and chips When Qualcomm announced in June 2025 that it would buy Alphawave for US$2.4 billion, the headline was a milestone for a company most people had never heard of. Inside the semiconductor world, however, the name had been known for years, and the man behind it was already on his second company. Tony Pialis had earlier founded V Semiconductor, a chip-design firm, which Intel went on to acquire. At Intel he served as a vice president running analog and mixed-signal intellectual property, work that sits deep inside the chips that move data at high speed. He left that role to start again, and in 2017 he co-founded Alphawave in Toronto. The idea was precise. Data centres were growing faster than the wires that connect them. Every chip, from a graphics processor to a networking switch, needs a way to send data across a board or a rack at enormous speeds. Designing those links is expert work, and few chip companies want to hire a team just to do it. Alphawave sold them the finished design to license, along with later chips and chiplets built around it. That licensing model carries a particular kind of risk. Customers can take years to design a product, and revenue lands only after they commit. A company in that position needs patience and a technical reputation strong enough to win design slots from large firms. Alphawave built that reputation quickly enough to list on the London Stock Exchange in 2021, an unusual choice for a Toronto-based technology company, since most Canadian founders aim for Toronto or New York. The real test came with the rise of artificial intelligence. Training large models depends on moving huge quantities of data between processors, and demand for exactly the kind of connectivity Alphawave made surged. That demand caught the attention of Qualcomm, which was looking to build out its data-centre business. The agreed price was US$2.48 per share in cash, or about 183 pence, and the board backed it unanimously. Qualcomm completed the purchase in December 2025, about a quarter ahead of schedule. Pialis stayed on, and was named to lead Qualcomm's data-centre business, a rare outcome in which a founder of an acquired company is handed a major operating role rather than a farewell package. The pattern in his career is worth noticing. One company sold to Intel, a second to Qualcomm, each in a niche most investors find intimidating. The lesson for an Indian founder building deep technology is that you do not need a consumer brand. You need a hard problem, a customer base that cannot easily replace you, and the stamina to solve it more than once. ## theScore (Canada) URL: https://startup.talk/canada/thescore/ Founders: John S. Levy, Benjamin (Benjie) Levy Founded: 1994 (as Sportscope Television Network); digital business spun out as theScore Inc. in 2012; Based in: Toronto, Ontario; Sector: Sports media and online sports betting In 2012, John Levy did something that looked, from the outside, like giving up. The sports television channel he had built over almost two decades, The Score, was being sold to Rogers. What he kept was the part many investors regarded as an afterthought: a website and a handful of mobile apps. Levy had founded the business in 1994 as Sportscope Television Network, with Clairvest Group as the majority shareholder. In 1997 its channel launched as Headline Sports, later renamed The Score, a scrappy Canadian rival to the established sports networks. Alliance Atlantis took a stake of nearly 48% in 1999, and the company went public through a reverse takeover in 2000 as Headline Media Group. Even then its name changed more often than its fortunes. The television business was never the sole point. By 2012 Levy and his son Benjie, the company's president and chief operating officer, had seen where audiences were going. Fans checked scores on their phones. They did not wait for a broadcaster's ticker. So when Rogers agreed to buy the channel, the company spun out its digital assets as theScore Inc., and existing shareholders received one share of the new company for each share of the old. The bet needed years to pay off. The app grew its audience, and by 2019 Penn National Gaming had signed a twenty-year market-access agreement covering 11 states, and theScore Bet launched with a New Jersey sportsbook partner. Penn took a 4.7% stake. In 2020 the company moved to the Toronto Stock Exchange, and in 2022 theScore Bet launched in Ontario. The pivotal moment arrived in October 2021, when Penn agreed to buy the whole company for about US$2 billion, paying US$17 in cash plus 0.2398 Penn shares per theScore share, which Penn valued at US$34 in total. Shares jumped sharply on the announcement. The ending was less triumphant. Within months of the deal Penn shut down theScore Bet in the United States, in June 2022, keeping the Ontario operation. John Levy left in February 2024, describing his time at Penn as tumultuous, and Benjie Levy stepped down in April alongside two other members of the family. Penn's chief executive called them pioneers of sports media and gaming. Later, in 2025, Penn's ESPN Bet customers moved over to theScore Bet, and by then it operated in 21 US states. The story is hard to reduce to one lesson, but a founder can take this one: the asset the market undervalues is sometimes the one that matters. Levy let go of the business with the brand recognition and kept the one that fit where his audience was heading. ## Clearco (Canada) URL: https://startup.talk/canada/clearco/ Founders: Michele Romanow, Andrew D'Souza Founded: 2015; Based in: Toronto, Ontario; Sector: Fintech (revenue-based financing for e-commerce) Before Michele Romanow ran a startup valued near US$2 billion, she ran a coffee shop. At Queen's University in Kingston she co-founded Tea Room, a zero-consumer-waste cafe, while studying civil engineering. In 2007 the university gave her its highest student honour, the Agnes Benidickson Tricolour Award. She was born in Calgary and raised in Regina, the daughter of Marvin Romanow, who went on to be president and chief executive of Nexen. Business was a family language, and she took to it early. After Queen's, where she also finished an MBA, she co-founded a string of companies: Evandale Caviar, Buytopia.ca and SnapSaves. In 2014 Groupon acquired SnapSaves and relaunched it in the United States as Snap by Groupon. Forbes had already named her a Millennial on a Mission in 2013, the only Canadian on that list. She was a seasoned founder by 2015, when she and Andrew D'Souza started Clearbanc. The pitch was blunt. Startups, especially online sellers, spend heavily on advertising and need cash before the sales arrive. Venture capital took equity and months to decide. Clearbanc offered a decision within 48 hours, a 20-minute term sheet, and cash repaid as a share of future revenue, with no stake given up. Funding ranged from US$10,000 to US$10 million, with fees between 6 and 12.5 percent. It worked, and loudly. By August 2019 Clearbanc had invested in 791 businesses and raised US$1 billion. Romanow joined Dragons' Den that same era, becoming its youngest cast member. In 2021 the company rebranded as Clearco, reached a valuation around US$2 billion and announced a European push in Germany with a pledge of 500 million euros. Then the market turned. Rising interest rates and falling tech valuations hit a business that lent against future sales. In 2022 Clearco cut about 125 employees, roughly a quarter of its team, then pulled back from the UK, Germany, Ireland and Australia. In January 2023 another cut of around 30% followed, and Romanow stepped down as CEO. Headcount fell to about 140 from roughly 500 the year before. Both co-founders were gone. The record shows a company that grew faster than its funding model could bear. Money lent against ad spend looked safe in a boom and fragile in a downturn. For a founder, the lesson is about speed. Raising and lending quickly created enormous momentum, but the same speed removed the margin for error. ## Photonic (Canada) URL: https://startup.talk/canada/photonic/ Founders: Stephanie Simmons, Michael Thewalt Founded: 2016; Based in: Vancouver, British Columbia; Sector: Quantum computing and networking A weekend, a microscope built by a master's student, and a stream of data. That was the moment Stephanie Simmons has described as the first real evidence that silicon could carry the weight of a quantum computer. Her lab at Simon Fraser University had just measured silicon spins for the first time, and the numbers said the idea was viable. She had been drawn to quantum technology since her teens. She trained in Canada and abroad, then settled in Vancouver, where she leads the Silicon Quantum Technology lab at SFU. For years she lived the academic life: papers, grants, students, instruments that someone had to build from scratch. Then came a harder question. Was a university lab the place where a useful quantum computer would ever be built? Simmons concluded it was not. She has said she understood that the jump to useful machines would have to happen inside the private sector. So in 2016 she and fellow Canadian physicist Michael Thewalt started Photonic in Vancouver. The bet was specific. Rather than building one giant chip, the company would use spins in silicon and link them with light, so that many smaller quantum machines could be networked together. Simmons became chief quantum officer. That is a long road for a startup. Quantum hardware does not ship in a year, and investors have to be convinced to wait. For much of its life Photonic worked quietly, the kind of company that did not need a loud launch. The turn came in November 2023. Photonic announced a strategic collaboration with Microsoft and US$100 million in new funding. Microsoft was among the investors, along with the British Columbia Investment Management Corp., the UK's National Strategic Investment Fund, Inovia Capital and Amadeus Capital Partners. Part of the deal was a roadmap for quantum networking and access to Photonic's silicon technology for Microsoft's Azure cloud. In January 2026 the company said it had closed C$180 million as the first phase of a new round, with Microsoft and BCI returning. Existing backers coming back is its own kind of proof: they had watched the machine for years and chose to put in more. Simmons also co-chairs the advisory council for Canada's National Quantum Strategy, so her work now sits between a company and a country's ambitions. For a first-time founder the lesson is in the order of events. The conviction came first, from an unglamorous weekend of data in a university lab. The company came later, once she was sure the lab could not finish the job alone. Patient belief in a measurement, not a pitch, carried it. ## Jane.app (Canada) URL: https://startup.talk/canada/jane-app/ Founders: Alison Taylor, Trevor Johnston Founded: 2012; Based in: North Vancouver, British Columbia; Sector: Practice management software for health and wellness clinics Alison Taylor was complaining about her booking system. That was all. She had just opened a clinic, and the software for scheduling patients was clumsy and ugly. The man who had designed the clinic's branding heard her out and offered to build something better. Taylor grew up in North Vancouver, where her parents were both physiotherapists, and for a time she expected to follow them. Instead she studied English at UBC with a minor in psychology. She still ended up in the family world, managing one of her parents' practices. In 2011 she opened her own place, Canopy Integrated Health, a clinic that rented space to midwives, counsellors, acupuncturists, massage therapists and others. The designer was Trevor Johnston, co-founder of the agency Thought Shop Creative. He built a first version for Canopy. For about a year and a half it was simply the clinic's own tool, nothing more. Then other clinics started asking about it. Taylor and Johnston added billing and began licensing it, around 2012, as a side business. By the time the company reached eight to ten staff, both had to leave their other jobs. What nobody planned was the name. Taylor found competitors' names generic and their screens poor. They called theirs Jane, a person's name, and made clean design the whole argument. It was a bet that a physiotherapist or a massage therapist, who did not want to think about software, would choose the one that felt friendly. The money story is unusual too. Jane took an early C$2 million loan from CIBC Innovation Banking, whose banker Joe Timlin pointed out that the company had been built for Taylor's own use rather than to chase a market. It was profitable early, spent little on marketing, and spread through the physiotherapy community largely by word of mouth. By 2021 Jane had about 50,000 clients worldwide and 217 staff in North Vancouver. Taylor has described running a small clinic as lonely, low-margin work, often without support, and says Jane tries to give owners the help she wished she had. The lesson is plain. Jane did not start with a market size. It started with one clinic owner's irritation, and a designer who treated it as a problem worth solving properly. ## Trulioo (Canada) URL: https://startup.talk/canada/trulioo/ Founders: Stephen Ufford, Tanis Jorge Founded: 2011; Based in: Vancouver, British Columbia; Sector: Identity verification Stephen Ufford decided to be an entrepreneur right after high school. Not after a degree, not after a corporate job. His first idea was online credit reports, because he had looked at how consumers got theirs and found the process outdated. That company was iQuiri, launched in 2000. It was one of the early businesses to put consumer credit reports and scores online, and it was acquired within three years. Ufford did not stop. He built two more companies in very similar markets, and both were acquired too. Trulioo, he has said, was his fourth startup. So he arrived in Vancouver in 2011, with co-founder Tanis Jorge, already a veteran of data-driven businesses and already knowing the hard part. Data was the business, and data belongs to someone else. To verify who a person is online, you need access to records held by banks, governments and credit bureaus, country by country. That was the work of the early years. Trulioo set out to offer what Ufford described as instant electronic identity and address verification, with a goal of reaching four billion people in more than 60 countries. At the time of that interview the company had raised over US$25 million. Growth meant persuading one data source after another to let a small Vancouver company plug into it. The first real validation came in 2014 with a US$6 million Series A led by Tenfore Holdings, joined by BDC Capital and Blumberg Capital. A year later American Express Ventures put money in alongside those backers. The company launched its international identity verification platform in 2014. Then the numbers grew. In 2019 Trulioo raised US$70 million, including a US$60 million Series C led by Goldman Sachs Growth Equity. In 2021 came a US$394 million Series D led by TCV at a US$1.75 billion valuation, which made it a unicorn. Later years brought acquisitions, offices in Dublin, Singapore and San Diego, a rebrand in 2023 and a new CEO, Vicky Bindra, in 2025. For founders, the quiet pattern is repetition. Ufford chose one unglamorous problem, trust between strangers online, and had already spent more than a decade learning how data markets work before he built the company that made it a unicorn. ## Paper (Canada) URL: https://startup.talk/canada/paper/ Founders: Philip Cutler, Roberto Cipriani Founded: 2014 (original name GradeSlam); Based in: Montreal, Quebec; Sector: Education technology (online tutoring) Paper began with a different name: GradeSlam. Philip Cutler and Roberto Cipriani, both McGill graduates, started it in Montreal in 2014 as a way to get students help with their schoolwork on demand. The name sounded like a student project, and for a while that is what it was. The idea was simple and appealing. Students would message a tutor through a chat window, not wait for an appointment. No travel, no scheduling. The company later became Paper and pushed that model into schools. The turning point was the pandemic. When classrooms emptied in 2020, school districts in the United States suddenly had federal relief money and a hole where after-school help used to be. Paper was ready. It ranked 21st on a list of Canada's fastest-growing companies in 2020, and by June 2021 it reported more than one million students using the platform, with about 1,000 tutors and 130 staff. Growth brought acquisitions: Figure Math in 2022, then MajorClarity and Readlee in 2023. The company had raised roughly US$389.6 million from investors by 2023. It was a long way from GradeSlam. What nobody inside the boom wanted to look at was the model underneath. Reporting later described a chat-based service in which tutors often worked with several students at once. Company records cited in that reporting showed tutors spent about 45% of their time with two or more students, and tutors were paid more for handling four or more simultaneously. That sits uneasily next to a promise of one-on-one tutoring. Then the money changed. Pandemic funding to districts shrank, and revenue and bookings fell short of expectations. In 2023 officials in New Mexico cancelled a contract after concluding the services were not giving students adequate academic help. Layoffs followed. Reports said 45% of headquarters staff were let go and Canadian tutors were dismissed. It is a hard ending to a story that began in good faith, with two classmates and a chat box. The sharpest lesson for founders is that a boom funded by a single source of public money can hide whether the product itself works. When relief spending stopped, the question that remained was whether students were being helped, and the company had to answer it in public. ## Nexii (Canada) URL: https://startup.talk/canada/nexii/ Founders: Founded: 2018; Based in: Vancouver, British Columbia; Sector: Prefabricated green construction materials Nexiite was the secret, and for a long time nobody could say exactly how to make it at scale. It is a concrete-like material developed in Moose Jaw, Saskatchewan, and the company that grew around it, Nexii, set out to turn it into prefabricated walls and roofs. The pitch was seductive: greener buildings, put up faster. Nexii was a Vancouver-based startup built on that Moose Jaw material, with plants in Moose Jaw and Squamish, British Columbia. In September 2021 it became a unicorn, valued above US$1 billion, which Canadian press called the fastest in the country to reach that mark. It gathered celebrity along the way. Actor Michael Keaton took an ownership stake in 2021 and was expected to take an active role in a planned plant near Pittsburgh. Former Vancouver mayor Gregor Robertson joined as an executive. Customers reportedly included Starbucks, Popeyes, Marriott and Walmart. A nine-storey Marriott Courtyard in Nanaimo opened in early 2023. In summer 2022 Nexii closed a US$45 million equity financing at a valuation of around C$2 billion, nearly double the year before. In total it had raised over US$250 million. What nobody outside the company saw clearly was how much the product itself still needed work. In 2021 Nexii re-engineered Nexiite, and the redesign soaked up large amounts of capital while growth lagged expectations. At the same time it pushed hard into the United States. A licensing deal for a facility in Hazleton, Pennsylvania, fell apart when the counterparty did not pay a US$5 million licensing fee and could not perform. Nexii spent more than a year on that project trying to protect its brand. The costs were higher than the revenue. In April 2023 it defaulted on interest to its senior secured lenders. Its CEO Stephen Sidwell stepped down in October, and Robertson left in November. In January 2024 a British Columbia court granted creditor protection, with more than C$109 million owed to creditors. The business was ultimately sold in a court-approved deal for C$500,000 plus the assumption of more than C$20 million in debt. For a founder, the story is a warning about sequence. A material has to be proven and repeatable before it is licensed, financed and scaled across a continent. Nexii raised the money of a finished company when the product was not yet finished. ## Tailscale (Canada) URL: https://startup.talk/canada/tailscale/ Founders: Avery Pennarun, David Carney, David Crawshaw, Brad Fitzpatrick Founded: 2019; Based in: Toronto, Ontario; Sector: Networking and security software (WireGuard-based mesh VPN) The product that now sits inside tens of thousands of company networks began as a favour to a bank. A Canadian bank had a Windows application and a problem: it needed sign-in that a phisher could not fool. The small team that would become Tailscale combined single sign-on with WireGuard, the lean modern VPN protocol, and handed it over. Avery Pennarun later described the result with a shrug. It worked so well for that 'one weird trick', he said in an interview, that they repackaged it for everyone. Pennarun did not arrive at that moment by a straight road. He grew up in Thunder Bay, Ontario, where he first got online through a dial-up connection. At the University of Waterloo, where he earned a computer engineering degree in 2001, he co-founded Net Integration Technologies, a company built to make networking just work for small businesses. IBM bought it. Then came more than seven years at Google, working on Google Wallet and Google Fiber, where the networks were enormous and the people who kept them running were few. He carried one irritation out of that world. Connecting two computers safely was still harder than it should be. In 2019, he started Tailscale with fellow Waterloo engineer David Carney, David Crawshaw and Brad Fitzpatrick, all of whom had worked at Google. The name came from a Google research paper, The Tail at Scale, and from the idea that most networking pain lives in the long tail of everyday cases. Money followed the idea. Accel led a US$12m Series A in November 2020. Remote work during the pandemic pushed monthly users up, and in May 2022 CRV and Insight Partners led a US$100m round that valued the company near US$1bn, when it had only about 35 employees. Pennarun's tone in public stayed oddly small. He wrote that the company wanted to grow at a healthy pace, at human scale, and called the team a bunch of weirdos. The hard part came after the fun part. A developer wants a tool that disappears. A security buyer wants controls, logs and policy. Pennarun has said the answer is to add only the enterprise features Tailscale truly needs and to keep the product small and coherent, never breaking what customers already rely on. By 2025 a further US$160m Series C, again led by Accel, brought total funding to roughly US$272m, and the company counted more than 20,000 paying business customers. It does not publish revenue. The lesson is in the origin story. The company did not start with a grand theory of the internet. It started with one customer who had one real problem, solved it, and noticed that everyone else had the same problem too. ## Rumble (Canada) URL: https://startup.talk/canada/rumble/ Founders: Chris Pavlovski Founded: 2013; Based in: Longboat Key, Florida (founded in Toronto, Ontario); Sector: Online video platform and cloud / AI infrastructure Rumble's founder has spent a decade being described by his critics and his fans. He has rarely described himself. Chris Pavlovski is a Canadian businessman whose family is from Macedonia. Public record says little more about his early years, and he has kept his private life out of the spotlight. What is documented is that in 2013 he started two companies at once. One was Rumble, launched in Toronto on October 30 as a place where independent creators could upload videos and earn from them. The other was Cosmic Development, an IT services firm. Years later, it would quietly supply the content moderation for the first. The early Rumble was small and unglamorous. It was a challenger to YouTube at a time when YouTube looked unbeatable. For years it grew slowly, sustained by a founder who owned most of it and by a belief that creators wanted a second home. The turn came from outside. In January 2021 Rumble sued Google for more than US$2bn, claiming its search results favoured YouTube. That May, Narya Capital, backed by Peter Thiel and the then senator JD Vance, put in US$25m at a valuation of about US$500m. Later that year Rumble bought Locals, the creator subscription site. Donald Trump joined in 2021, and in 2022 the platform began hosting Truth Social. Audiences arrived in waves, many of them people banned or frustrated elsewhere. The cost was reputational and legal. France blocked Rumble in 2022 after it declined to remove Russian state media accounts. The SEC confirmed an investigation in 2024, which the company blamed on short sellers; it was later cleared. Brazil suspended the service in February 2025 after Rumble refused to follow court orders. Critics say the site moderates less than rivals on hate speech and hosts misinformation. Through it all, the money grew. Rumble went public on Nasdaq in September 2022 through a SPAC merger brokered with Cantor Fitzgerald, and moved its headquarters to Florida in 2023. After the 2024 US election, Tether invested US$775m. Revenue was US$95.5m in 2024, though the company was still unprofitable, and the stock's rise made Pavlovski a billionaire in January 2025. In 2025 he agreed to buy Northern Data for US$767m, and in June 2026 the company renamed itself RUM Group and added an AI and GPU unit, Quake. The video site that began as a YouTube alternative is now also a bet on computing infrastructure. For a founder, the quieter lesson sits in that 2013 pairing. Pavlovski built the thing people could see and the thing nobody noticed, the back office that made it run. Controversial platforms are often decided less by the front page than by the boring machinery underneath. ## Keel Infrastructure (Canada) URL: https://startup.talk/canada/keel-infrastructure/ Founders: Emiliano Grodzki Founded: 2017; Based in: New York City, New York (formerly Toronto, Ontario); Sector: Bitcoin mining turned AI and high-performance computing data centres and energy infrastructure In the spring of 2026 a company that had spent nearly a decade mining bitcoin changed its name, its country and its answer to a simple question: what is a cheap power plug worth? Bitfarms was a bitcoin miner, built on the idea that cheap electricity could be turned into digital coins. The best-known name tied to the beginning is Emiliano Grodzki, and the founders have kept low profiles. What is clear is the arc. Mining was a business at the mercy of one price. When bitcoin fell, margins collapsed. When it rose, rivals poured in. The company's chief executive, Ben Gagnon, described the shift plainly. A year before the rebrand, he said, the company had set out on a transformational plan to pivot away from bitcoin. The thought was that the same grids and sites that powered mining machines could host computers for artificial intelligence, where customers sign long contracts and pay for reliable power. The pivot had a price. Revenue reached US$229m in 2025, up 72%, yet losses widened. In the first quarter of 2026, revenue fell 23% to US$37m and the company lost US$145m. An analyst pointed to US$588m of convertible notes as a worry. To raise cash for the build-out, the company planned to sell bitcoin it had once been proud to hold. On April 1, 2026, Bitfarms completed its move from Canada to the United States and became Keel Infrastructure, trading as KEEL on Nasdaq and the Toronto exchange, with its only principal executive office in New York City. It describes itself as a North American digital infrastructure and energy company that develops and owns data centres and energy assets for demanding computing, including AI. By May it reported liquidity of about US$533m, made up of US$336m in cash and US$197m in unencumbered bitcoin. There is a founder lesson in the keel itself. A keel is the part of a ship you never see, and it is what keeps the vessel upright. A miner builds on a commodity it cannot control. A company that owns the power and the buildings owns something customers will pay to rent whatever the next technology is. Whether the bet pays off is still open. The share price, about US$4.30 in May 2026, reflects that. But the direction is clear: the people who mined coins are now trying to sell the ground they stood on. ## Figment (Canada) URL: https://startup.talk/canada/figment/ Founders: Lorien Gabel, Andrew Cronk, Ben Chan Founded: 2018; Based in: Toronto, Ontario; Sector: Blockchain staking and Web3 infrastructure In August 2021 a Toronto crypto company announced a C$63m round, and a surprising number of famous names in digital assets were on the cap table. Figment was founded in 2018 by Lorien Gabel, Andrew Cronk and Ben Chan, and the public record says little about their lives before it. What is on the record is the idea. Newer blockchains work by proof of stake: holders lock up their coins to help secure the network and earn rewards. That sounds easy until you realise it requires running servers around the clock, with real penalties for mistakes. Most large holders would rather pay someone else to do it. Figment became that someone. It offered staking, middleware and application-layer tools to token holders and developers, and in October 2020 it raised a C$2.5m seed round to build out. Ten months later it closed a C$63m Series B, about US$50m, co-led by Senator Investment Group and Liberty City Ventures, with Galaxy Digital, Anchorage, Bonfire Ventures, GSR and 40 North taking part. Bloomberg reported the valuation at US$500m. The company had 75 employees and said it wanted to expand across the whole Web3 stack. The timing was either lucky or sharp. Staking was moving from a hobby for early adopters to an infrastructure service institutions wanted, and Figment had been selling that service for three years. It also placed itself in the unglamorous layer: not a coin, not an exchange, but the plumbing in between. That position has costs. Staking revenue rises and falls with token prices, so the business lives inside crypto's cycles. Rivals include exchanges that bundle staking and protocols that offer it with no middleman. In October 2025, Preqin recorded new venture funding from C1 Secondaries Fund, a sign that investors still wanted exposure. Figment now works from Toronto, with offices in New York and Warsaw. The company's story is deliberately low on personality; its founders let the product speak. For first-time founders the point is clear enough. In a gold rush, the people selling reliable tools tend to outlast those chasing the coin of the day, but only if they stay alive through the quiet years in between. ## Nord Quantique (Canada) URL: https://startup.talk/canada/nord-quantique/ Founders: Philippe St-Jean, Julien Camirand Lemyre, Alexandre Blais Founded: 2020; Based in: Sherbrooke, Quebec; Sector: Quantum computing (bosonic qubits on superconducting circuits) Quantum computing is usually pitched with big numbers. Nord Quantique's pitch is about doing more with fewer parts. The company began in 2020 in Sherbrooke, Quebec, as a spinoff of the Institut quantique at the Universite de Sherbrooke. Its founders are named in press material as Philippe St-Jean and Julien Camirand Lemyre, with Professor Alexandre Blais, the institute's scientific director, described in funding announcements as a co-founder and scientific force behind the work. The story is one of science and money more than biography. The scientific bet is distinctive. Most quantum designs fix errors by surrounding each useful qubit with many extra ones. Nord Quantique builds error correction into a single qubit, encoding information in microwave photons inside a cavity with Schrodinger-cat and GKP codes, then adding software. The promise is a smaller, cheaper machine that reaches fault tolerance sooner. Proving that required patient early backers. Quantonation and Real Ventures funded a pre-seed in 2020. In February 2022 the company announced a C$9.5m seed round co-led by Quantonation and BDC Capital's Deep Tech Venture Fund, aimed at moving from concept to a first generation of chips. Quantonation's managing partner Christophe Jurczak said he had full confidence in the vision and capabilities of the team. Later the federal government named the company to the Canadian Quantum Champions Program, with funding of up to C$23m. The stakes are high because the rivals are not small. Google, IBM and Amazon all chase fault-tolerant machines, and others such as Alice and Bob in France use similar bosonic ideas. A startup in a university town has to prove its approach is not just clever but cheaper to build than theirs. In May 2026 the company raised US$30m from investors including Fidelity, Panache Ventures, Presidio Ventures and Quantacet, alongside existing backers. Reports put its valuation near US$1.4bn, a striking figure for a company whose first seed round was under C$10m four years earlier. The founding lesson here is institutional as much as personal. The company grew out of a research institute that had spent years on the physics, and its first investors backed that depth before there was a product. Deep technology rarely rewards speed; it rewards staying close to the lab until the idea is ready to leave it. ## Magnet Forensics (Canada) URL: https://startup.talk/canada/magnet-forensics/ Founders: Jad Saliba, Adam Belsher Founded: 2009; Based in: Waterloo, Ontario; Sector: Digital forensics software for law enforcement and corporate investigators Jad Saliba never planned to leave the street. He was a police officer in the Waterloo region, a patrol cop, until cancer took that job away from him. After a year of radiation and chemotherapy he was well enough to come back, but not to patrol. The service moved him into digital forensics, the unit that digs evidence out of seized phones and computers. It looked like a consolation posting. It turned out to be the whole story. Saliba started spending evenings and weekends writing a program of his own. He has described it as a fix for a problem he was having at work, a way to pull internet chat, browser and social media traces out of a drive without hours of manual digging. He called it Internet Evidence Finder. Colleagues borrowed it. Then officers in other agencies asked for copies. What almost nobody outside forensics knew was that this was not a startup idea. It was a tool built by a man who had been told his old career was closed, and who was working inside the unit that handled some of the worst cases an officer sees. The pull of the work, not a business plan, kept him writing code. Saliba was a gifted builder, not a seller. The company that grew around IEF needed someone who knew how to run one. He found Adam Belsher, a senior executive at BlackBerry, and the two partnered: Belsher as chief executive, Saliba as the technical founder. Magnet Forensics was formally founded in 2009, in Waterloo, the same town that had produced BlackBerry. The cost of that choice was a different kind of pressure. Their customers were police services and governments, buyers who move slowly and compare every tool against Cellebrite and others. Magnet kept pushing into mobile devices and cloud data, and moved to the Toronto Stock Exchange in April 2021 at C$17 a share. Then came a harder lesson. Magnet had planned to buy Grayshift, the Atlanta firm it had worked with for years. Thoma Bravo got there first, buying a stake in July 2022 and outbidding Magnet that autumn. Rather than fight, Magnet agreed to be bought. In January 2023, Thoma Bravo agreed to pay C$44.25 a share, about C$1.8 billion, and to merge the two companies, with Belsher and Saliba staying in senior roles. By the time of the sale, nearly 3,000 agencies in 90 countries were reported to use the product, a company-era figure, but it shows how far a spare-time tool from a sidelined cop had travelled. The lesson is easy to miss. Saliba did not set out to build a business. He built something because he had a job to do and a limited body to do it with. The company came later, once somebody else needed the same help. ## Assent (Canada) URL: https://startup.talk/canada/assent/ Founders: Rob Imbeault, Matt Whitteker, Andrew Waitman Founded: 2005; Based in: Ottawa, Ontario; Sector: Supply chain compliance and ESG software (SaaS) Andrew Waitman learned about Assent in a boxing gym. He was in his forties, a former chief executive of the Ottawa tech firm Pythian, and he had taken up the sport late. There he met Matt Whitteker, a fellow boxing enthusiast who, with Rob Imbeault, had started a small compliance software company in 2005 and organised the Fight for the Cure charity boxing matches on the side. Waitman left Pythian in August 2014. Whitteker invited him to look at Assent, at first as a volunteer. Within a short time Waitman was the chief executive. He later told the Ottawa Business Journal that Whitteker was the reason Assent exists. That detail rarely makes the funding headlines, but it explains the place. Assent's Ottawa office has a full-size boxing ring in it. The company uses it as a stage for announcements and all-hands meetings, not for settling arguments. Waitman was not new to engineering or to building companies. He had an electrical engineering degree from Waterloo, class of 1987. What he found at Assent was a product with unusual timing. Manufacturers were being asked by regulators and their own customers to prove where raw materials came from, and whether suppliers met rules on human rights, health and safety. Doing that by spreadsheet and email was misery. Assent sold software that collected the answers from suppliers. What almost nobody expected was how fast that boring work would pay. By the middle of the 2010s the company had about 25 employees and clients that included more than 30 percent of S&P 500 product companies, with contracts often in the range of C$50,000 to C$100,000 a year. The business could not keep growing organically at the pace those customers needed, so it took a C$20 million equity investment and planned to go from 165 people to 265. The bigger money came later. In early 2022 Vista Equity Partners led a US$350 million deal that valued Assent above US$1 billion, one of the largest rounds in Canadian software history. Much of it bought out earlier shareholders. By June 2024 recurring revenue had passed US$100 million, and in 2025 Vista raised its stake at a reported valuation near US$1.3 billion. Assent said it had not been acquired. Michael Southworth became CEO that May, and Waitman moved to executive chairman. There is a founder lesson here that does not appear in any pitch deck. Assent's second act began with an unforced introduction in a gym, a volunteer first look, and a founder willing to hand over the wheel. The best outside CEO can arrive through a friendship, if you are willing to ask. ## Ada (Canada) URL: https://startup.talk/canada/ada/ Founders: Mike Murchison, David Hariri Founded: 2016; Based in: Toronto, Ontario; Sector: AI customer service automation In early 2016 Mike Murchison and David Hariri had to tell their team it was over. Volley, the social app they had been building in Toronto, had found a following among young professionals but never reached the vision the two founders had sold their investors. They let everyone go and wound the company down. Murchison had come to Volley by an academic route. He studied cognitive science, psychology and human-computer interaction at the University of Toronto, graduating in 2012. As a student he sat in on courses taught by Geoffrey Hinton and spent time in Steve Mann's wearable computing lab. Pure AI was not yet a business. It was something he watched from the back of the room. The shutdown was the low point, but it was not the end of the relationship with Volley's backers. Version One Ventures, which had led the pre-seed round, did not pull away. What almost nobody expects from investors after a failure, they nudged the founders to look hard at a pain point they had lived while building Volley: talking to customers. The two founders had a particular advantage. Murchison had worked for several months as a frontline tech support agent before building Ada, so he knew what the job felt like. They chose customer support, and by summer 2016 Volley had turned into Ada. The first product was a chatbot that answered routine questions and could be edited by non-technical support staff, with set-up said to take under an hour. Medium, Kik and Wattpad were early customers. A seed round followed in October 2016, with a surprise term sheet from Bessemer, usually a later-stage investor, after one of its associates who had followed Volley since his Google days made the connection. The company grew from there. Accel led a US$44 million Series B in March 2020. In May 2021 Spark Capital led a US$130 million Series C, valuing Ada at a reported US$1.2 billion and pushing total funding to about US$200 million. Customers later included Meta, Shopify and Verizon. Murchison told University of Toronto interviewers that AI should be treated like a new employee who needs coaching, and his stated mission is to make customer service extraordinary for everyone. The lesson is a quiet one. A closed company is not a closed career. Murchison's failure gave him two assets he could not have bought: a loyal investor and an irritating, specific problem he already understood. ## Shareworks (Canada) URL: https://startup.talk/canada/shareworks/ Founders: John Kenny, Mark Van Hees Founded: 1999; Based in: Calgary, Alberta; Sector: Equity compensation and cap-table software (SaaS) When Morgan Stanley agreed in February 2019 to pay C$19.15 a share for Solium, a 40 percent premium, many outside Calgary asked the same thing: who was Solium? The company administered employee share plans for some of the biggest corporations in Canada, and for decades it had been a quiet success. A Calgary Herald columnist once called it the city's best-kept tech secret. It began in 1999 with two financial advisers, John Kenny and Mark Van Hees. Both worked in Calgary, a city where oil and gas companies were paying staff in options and shares long before most Canadians had heard the term equity compensation. Kenny's earlier career included Mackie Research Capital and Canaccord Genuity, the kind of Calgary brokerage floors where such plans are managed. Their idea was narrow. Companies needed a clean way to administer share plans, and spreadsheets were no answer. Solium built software for it, and its first customers were names on the Calgary skyline, TransAlta, Encana and Shell Canada. The company moved quickly to the public market, listing on the TSX Venture Exchange in 2001, only two years after founding. It became profitable in 2004. That is a long road for a small company, and the document trail shows no big venture round to rely on. It had to stay profitable by serving clients well. The next decade was about breadth. Solium bought Transcentive from Computershare in 2010, then CapMx in 2012, Capshare in 2017 and Advanced-HR in 2018. It opened offices in London in 2011 and Sydney in 2014. Its customer list moved from oil and gas boardrooms to technology firms and startups, which hold much of their wealth in shares nobody can easily sell. Marcos Lopez, who led the company as chief executive, stayed on as head of the business after Morgan Stanley completed the purchase on May 1, 2019. The price was about C$1.1 billion, or roughly US$900 million. The company was renamed Shareworks and folded into Morgan Stanley at Work, alongside retirement and financial wellness tools. For first-time founders the useful point is the patience on show. Kenny and Van Hees did not chase a flashy category. They solved a dull problem for a handful of demanding clients, reached profit in five years, and then kept acquiring until the whole world needed what they had built. ## Carbon Engineering (Canada) URL: https://startup.talk/canada/carbon-engineering/ Founders: David Keith, Geoffrey Holmes Founded: 2009; Based in: Squamish, British Columbia; Sector: Direct air capture of CO2 and synthetic fuels David Keith was born in Wisconsin because his father, a British-born field biologist, happened to be in graduate school there. The family moved on. Keith grew up in Britain and then Ottawa, where his mother taught history at Carleton University. It was a household of birdwatching and fieldwork. He would go birdwatching with his father as a child. In his teens he spent summers in a National Research Council laser lab, and he spent part of a gap year in Nunavut as a research assistant to a wildlife biologist on Devon Island, a trip he says fixed his love of the Arctic. He has also canoed the Yellowknife River and skied across rough sea ice from Iglulik to Arctic Bay. The odd thing is how those two worlds met. Keith studied physics at the University of Toronto, finishing in 1986, then took a doctorate at MIT in 1991, where he helped build the first working atom interferometer. The wilderness had taught him what a changing climate might cost. Physics taught him what could be measured. He turned to climate and energy, and some of his proposals were loud. A 2004 paper based on his modelling suggested huge wind farms might have unexpected climate effects. His interest in spraying sulphuric acid into the upper atmosphere drew sharp criticism, and a planned 2021 balloon experiment in Sweden was cancelled after opposition from Sámi reindeer herders. Carbon Engineering was the quieter path. Founded in 2009 out of research at the University of Calgary and Carnegie Mellon, it set out to pull CO2 from the air using a liquid that binds to the gas and a second cycle that releases it in concentrated form. Bill Gates, who had funded Keith's research since 2007, and Alberta oil financier N. Murray Edwards were early investors. What almost nobody in climate circles expected was how much of the money would come from oil. A pilot plant in Squamish captured about one tonne of CO2 a day from 2015. In 2017 the team made fuel from the captured gas. A 2018 study put costs between US$94 and US$233 per tonne. In 2019 Chevron, Occidental and BHP backed a US$68 million round. Occidental bought the company in August 2023 for US$1.1 billion, paid over three years, and said it would build 100 plants. Keith held about four percent before the sale. The lesson for a founder is a plain one. The same man who walked ice floes and sat in the lab could not choose between pure science and practical engineering, so he built a company that needed both, and he accepted partners who did not always look like climate allies. ## eSentire (Canada) URL: https://startup.talk/canada/esentire/ Founders: Eldon Sprickerhoff, Edmund Dengler Founded: 2001; Based in: Waterloo, Ontario; Sector: Cybersecurity (managed detection and response) Eldon Sprickerhoff put his house on the table before he had a single customer. It was late 2001. He had just come back to Canada from New York, where he had worked at a prime brokerage through the dot-com crash and the aftermath of 9/11. He had a degree in computer science from the University of Waterloo, a childhood friend ready to join him, and a plan to sell cybersecurity consulting. He took a $150,000 line of credit against his home. He told a podcast that he did not ask friends or family for money, and he stayed away from venture capital. That decision shaped everything that followed. The company, eSentire, was started with Edmund Dengler, a fellow Waterloo computer scientist with a master's degree. Its first customer was in New York, not Ontario: ING Asset Management, reached through a contact he had met while doing an earlier contract for the bank's prime brokerage. Sprickerhoff did a weekend network cutover for that contact, who later hired him for more work. The early years were lean. He has described them as a death march, with long hours, almost no pay for himself, and cheap JetBlue flights from Buffalo to New York to save money. He has said he had almost no business knowledge and the team made up the rules as they went. What almost nobody sees from outside is how slow it was. It took about seven years to reach US$1m in revenue. Investors who looked at the company did not understand a services business, so growth stayed organic, around 30% a year. Then came a change of structure rather than a change of idea. The consulting work, vulnerability assessments, turned out to be a doorway into something customers would pay for every month: round-the-clock watching of their networks. Recurring revenue climbed from about half of the business to more than 90%. About eight years in, he hired J. Paul Haynes as CEO, and the company raised roughly US$1.5m from Canadian investors. Revenue went from US$1m to US$10m in about three years, which Sprickerhoff credited to a clearer story for buyers who were not technical. Against the big vendor SecureWorks, he said, eSentire won about 95% of head-to-head deals by offering a customised burger instead of a standard Big Mac. The price of that growth was control. Warburg Pincus bought in during 2017, a recapitalisation that let original investors cash out, and Sprickerhoff sold a majority stake and kept a small piece. Kerry Bailey became CEO in 2018. By February 2022, after a US$325m raise led by Georgian and CDPQ, eSentire was valued above US$1bn, with recurring revenue above US$100m and a plan to hire 400 more people. The first cheque was his own risk and his own house. For a founder with more nerve than business training, that may be the only credential that counted for the first seven years. ## Q9 Networks (Canada) URL: https://startup.talk/canada/q9-networks/ Founders: Osama Arafat, Paul Sharpe Founded: 2000; Based in: Toronto, Ontario; Sector: Data centre hosting and managed services In 2002, while the internet industry was burying its dead, Osama Arafat bought a corpse. Exodus Communications had gone bankrupt, and its giant Brampton data centre, about 160,000 square feet, was up for sale. Q9 Networks, the Toronto hosting company Arafat had started two years earlier with Paul Sharpe, agreed to take it. Arafat said the deal let Q9 meet demand much faster than if it had built a second site itself. The bet looks sharper once you remember the timing. Q9 launched in May 2000, within weeks of the market peak. Its first data centre opened in September of that year in downtown Toronto, near Bay Street. That address mattered. Banks and financial firms wanted their servers close, and Q9 sold them the idea that a data centre could be an outsourced utility, not a back-room closet. He and Sharpe kept the company small in story and large in steel. Over the next decade Q9 built out sites in Ontario, Alberta and British Columbia, ending up with 11 data centres. The 2010 announcement of a C$125m, 240,000 square foot facility beside the Brampton sites created what was then Canada's largest commercial data centre campus. It also created risk, because every building had to be filled with customers before the debt came due. The corporate story was not smooth either. In 2008, ABRY Partners agreed to buy Q9 for C$361m, and that buyout did not go through. Four years later, an investor group including Bell and Ontario Teachers' agreed to buy the whole company for C$1.1bn, with Arafat and Sharpe expected to keep running it as a stand-alone business from Toronto. The price tag, three times the failed offer, is the real reveal. Q9 never pitched itself as a technology breakthrough. It was patient capacity, built where customers worked and bought when sellers were desperate. For a first-time founder, the lesson is unglamorous. If you can survive the year everyone else is leaving, the empty buildings left behind are the cheapest growth you will ever buy. ## Nesto (Canada) URL: https://startup.talk/canada/nesto/ Founders: Malik Yacoubi, Damien Charbonneau, Karim Benabdallah, Chase Belair Founded: 2018; Based in: Montreal, Quebec; Sector: Fintech: digital mortgage lending and mortgage technology In 2018, Malik Yacoubi left a full-time executive job to start over. That is how EY described the moment when it named him its 2026 Entrepreneur of the Year for Eastern Canada. He teamed up with co-founders, reportedly Damien Charbonneau, Karim Benabdallah and Chase Belair, and launched a digital mortgage lender in Montreal in May of that year. The business plan was plain: Canadians were borrowing hundreds of thousands of dollars through a slow process, and software could make it faster. The reception was cold. Banks were sceptical. Brokers, the people who had always arranged mortgages, saw Nesto as a threat that could cut them out. A new lender needed both groups. His company has made a series of hard choices. The company raised a C$76m Series B, then C$80m in December 2022, taking total funding to C$167.5m, at a moment when many fintechs were retreating. The turn came in 2024 with the purchase of CMLS Group, described as Canada's third-largest mortgage finance company. Headcount jumped from about 350 to 1,100 overnight. Nesto was no longer just a consumer app; it administered mortgages worth tens of billions of dollars and began building white-label technology for banks and other financial institutions, the same institutions that once doubted it. The tool behind that shift is Maestro, an AI platform the company says can underwrite a deal in under two minutes, where a human process once took a day or more. Yacoubi has said that pairing deep vertical expertise with AI is where the magic can happen. The June 2026 Series E tells how far the company has moved. C$302m was raised, with C$107m in new capital and C$195m sold by earlier holders, at a valuation of C$1.47bn. New backers included La Caisse and Fidelity. The company reports more than C$37bn in mortgages originated that year and says it is profitable. The lesson for a founder is in who Nesto eventually sold to. The people who looked like opponents in 2018, banks and brokers, became the customers once the product was good enough and the company was large enough to be trusted. ## Svante (Canada) URL: https://startup.talk/canada/svante/ Founders: Brett Henkel Founded: 2007; Based in: Burnaby, British Columbia; Sector: Carbon capture technology and manufacturing The company that now builds machinery for Chevron began in a Vancouver garage. Brett Henkel had worked at QuestAir Technologies, a gas purification company, and then left. In 2007, he co-founded Inventys, working out of his garage. Karen Miller, another QuestAir alumna, later became Svante's vice-president of finance. They were not starting from nothing: they had spent years learning how to pull gases apart with adsorption. The idea they carried into the garage was simple to say and hard to build. Instead of washing flue gas with chemicals, put a material that grabs carbon dioxide on rotating beds, and let it capture and release the gas in a cycle. Nothing about the climate business was kind in 2007. Carbon prices were uncertain, governments were fickle, and investors preferred software. The company survived long enough to find patient backers. Chevron Technology Ventures made its first investment in 2014. The turn came with scale. In 2019 the company renamed itself Svante, after the Swedish scientist who first used models to link atmospheric carbon dioxide to temperature. It raised US$100m in a 2021 Series D, with Suncor and Chrysalix among the investors, and then US$318m in December 2022, led by Chevron with Temasek, 3M Ventures and GE Vernova joining. That money was meant for steel, not research. A pilot ran at Chevron's Kern River facility in California, and Svante broke ground on a factory in Burnaby in 2022 to mass-produce the filters. In 2024, Canada Growth Fund committed up to C$100m. The price of that ambition is exposure. Capturing carbon is only worthwhile if governments or customers pay for it, and the company has no revenue figures public. Chief executive Claude Letourneau said in 2022 he was optimistic about reaching unicorn status, a private valuation of US$1bn, though no valuation has been disclosed. The founder's lesson is the garage. The idea that is now a US$490m-plus enterprise began with one engineer leaving a company he knew and betting that the same chemistry could be pointed at a bigger problem. ## Semios (Canada) URL: https://startup.talk/canada/semios/ Founders: Michael Gilbert Founded: 2010; Based in: Vancouver, British Columbia; Sector: Agtech: precision orchard sensing and pest management Every ten minutes, a sensor in an almond orchard sends Michael Gilbert's company a reading. By 2020, Semios had more than a million such sensors in the field and was collecting about 350 million data points a day. It managed more than 150,000 acres, including over 500 farms in the United States. Gilbert, a doctor of philosophy by title and the company's founder and chief executive, started it in Vancouver in 2010. The idea was to treat a tree crop like an industrial process. Sensors track temperature, moisture and pests; software tells the grower when to irrigate, when frost is coming, and when to deploy pheromone-based pest control instead of blanket spraying. Growers pay a subscription, which means Semios earns only if the farmer keeps renewing. That is a hard business to start. Farmers are cautious and hardware breaks outdoors. Funding came late. In 2016, Reid Carter, who had headed Brookfield's timberland investment arm, led a C$8m seed with other individuals, and the Canadian government added C$950,000. Then growth took off. Gilbert's company has said revenue roughly doubled each year on average from 2015. That opened the door to bigger cheques. In February 2020, Morningside Group led a US$75m round, bringing total external capital to US$115m. A US$25m loan from CIBC Innovation Banking followed that March. In 2021 came US$100m more, and the company bought the farm-software firm Agworld. The cost of scale is that Semios is now in the business of keeping sensors alive across thousands of kilometres of trees, and the company keeps its revenue and profit private. The company has also been named to the Global Cleantech 100 list since 2018, a sign that outside judges noticed the model before the revenue figures were public. The lesson is in the subscription. Semios did not sell a gadget; it sold a growing season's worth of reassurance, and every grower who renewed was a vote that the data paid for itself. ## Visier (Canada) URL: https://startup.talk/canada/visier/ Founders: Ryan Wong, John Schwarz, Jan Schwarz Founded: 2010; Based in: Vancouver, British Columbia; Sector: People analytics and workforce planning software (SaaS) Before a visit from executives of a big company, someone had to clear away the takeout containers. The meeting place was a townhouse in Vancouver. Upstairs, a bedroom doubled as the boardroom. Downstairs, servers were stacked on the kitchen counter. There was no air conditioning, and the team could not open the windows because of construction across the street. The product they were about to demonstrate was, by their own later account, a click-through prototype made mostly of mock-ups. This was Visier in 2010, and the people inside called themselves data geeks. They were building it while still holding full-time jobs, and they were later nicknamed the Townhouse Crew. John Schwarz, who had earlier led the analytics company Business Objects, was involved, and so was Ryan Wong, who would become chief executive. At its most crowded, thirteen people worked in about 1,500 square feet. Their problem was not only the heat. One of the founders put it plainly: they had to create the market they wanted to sell into. Companies kept employee data in payroll and HR systems, but few leaders could ask a simple question of it, such as why people were leaving, and get an answer without waiting for a report. Visier wanted to hand HR those answers ready-made. What almost nobody outside the townhouse saw was how thin the proof was at the start. Wong later recalled reaching ten customers and the first payment, which looks tiny by today's standards. The first big deal, with Commonwealth Bank of Australia, changed what the team believed was possible, and the bank remains a major customer. The next test was technical. Rather than rely on someone else's database, the developers built their own in-memory analytics engine from scratch. They worked through the Christmas holiday and released it a few weeks later in early 2012, after the first customers had gone live in late 2011. A founder said there were moments when he wondered whether all the effort would pay off. Money followed the evidence. Foundation Capital led the first round, a US$6m Series A in 2011, and more rounds came in 2013, 2014 and 2017. In June 2021, Goldman Sachs Asset Management led a US$125m Series E that valued the company above US$1bn, taking total funding to about US$219.5m. The founders' reaction to the label is the most human part of the story. Wong called it exciting but stressful. Schwarz said he would celebrate quietly until the private valuation became a public one. The lesson is in the townhouse. Visier did not wait for a market, a boardroom or a finished product. It showed up with mock-ups, a kitchen full of servers and a clear question, and let one customer after another turn the picture into software. ## GeoComply (Canada) URL: https://startup.talk/canada/geocomply/ Founders: Anna Sainsbury, David Briggs Founded: 2011; Based in: Vancouver, British Columbia; Sector: Geolocation compliance and fraud prevention software (online gambling, streaming, financial services) For ten years, GeoComply said no to investors. That is the detail that stands out in the company's own announcement in March 2021. Founder Anna Sainsbury said GeoComply had received countless investment enquiries over the past few years. It had never taken an institutional cheque. Then Blackstone Growth and Atairos arrived with a minority investment, terms undisclosed, and Sainsbury confirmed that the move made her Vancouver company a unicorn. The business she built with David Briggs, founded in 2011, solves a problem most people never notice. When a phone is used to place a bet online, the operator must prove the bettor is physically inside a state or country where betting is legal. A wrong answer risks a licence. GeoComply built software that checks location from the device and flags people trying to fake it. The company moved to Vancouver in 2016, drawn by the talent in the city. That timing mattered. In 2018, the United States opened the door to state-by-state sports betting, and every new state created a new rule about where a wager could legally start. Operators who had never thought about geography now needed it every second. What the record shows is a founder who built slowly and kept control. She described the company's mission as empowering the future of digital trust. Following the first institutional round, Norwest Venture Partners and Arctos made minority investments in January 2023, about eighteen months after Blackstone. In May 2023, GeoComply bought Vancouver-based OneComply, its first major acquisition, extending its reach from location into wider compliance. Maclean's put Sainsbury on its 2024 Power List, noting the company's role in enforcing out-of-state betting laws. There is a harder side to the story. A product that depends on law is exposed to law. Each state that opens or closes its market changes demand, and fraudsters keep finding ways to make a phone lie about where it is. The lesson for a first-time founder sits in the pause. GeoComply turned away money for years, found its market when regulation moved, and took capital only when it could pick partners who would not ask it to give up the company. Patience can be a strategy, if the product is already needed. ## Enverus (Canada) URL: https://startup.talk/canada/enverus/ Founders: Allen Gilmer Founded: 1999 (as Drillinginfo; list shows 1998); Based in: Austin, Texas (major Calgary presence); Sector: Energy data analytics and SaaS for oil, gas and power In December 2018, an Austin company wrote a cheque of about C$46m to a software firm in Calgary. The buyer was then called Drillinginfo. The seller was Cortex Business Solutions, which automated the invoices that oil and gas companies swap with their suppliers. The deal put Enverus, as the company would soon be known, on the Canadian map and explains why this American-headquartered business is on a list of Canadian startups. Its software is used in Calgary offices every day. The story starts in Texas, around 1999, when Drillinginfo began collecting public drilling records and selling them as searchable data. Allen Gilmer is credited with founding it, though we have not been able to verify the details of his early life. A geologist's instinct to look for patterns in the ground, applied to paperwork, seems to have been the idea. The records were public. The value was in making them usable. An independent oil company that wanted to know who was drilling next to its lease, or which acreage was changing hands, no longer needed a landman and a stack of filings. A subscription replaced the stack. The hard part came after the data. Oil and gas companies spend freely in booms and cut hard in busts, and a business that sells to them rides the same cycle. Drillinginfo's answer was to move from information into software that people used daily. It bought MineralSoft in 2017, Oildex in September 2018 from Accel-KKR, and then Cortex. In 2020, it renamed itself Enverus. The new name reflected a wider ambition, with the company's reach going beyond oil and into power and renewables. Private-equity firms, among them Genstar Capital and Insight Partners and later Hellman & Friedman, have owned stakes along the way. Revenue is not published, and founders' accounts of the early lean years are not easy to find. What the record shows is a pattern of buying useful tools, joining them under one login and selling the bundle to a demanding industry. A first-time founder can read a quiet lesson in that sequence. The first product made information searchable. The durable business came from sitting inside the customer's daily work, where leaving is expensive and staying is easy. ## FreshBooks (Canada) URL: https://startup.talk/canada/freshbooks/ Founders: Mike McDerment, Levi Cooperman, Joe Sawada Founded: 2003 (product launched 2004); Based in: Toronto, Ontario; Sector: Cloud accounting and invoicing software for small businesses The mistake took one click. In 2003, Mike McDerment was running a small design firm in Toronto and billing clients with Word. A client came back for more work, so he did what any busy person would do: he opened last time's invoice, changed the details and saved it. He saved it over the original. A record of what he had billed, and what he was owed, was gone. That was the itch. McDerment decided there had to be a better way to bill, and since no one had built it, he started to build it himself. He worked out of his parents' basement in Toronto, and by his own later account the effort went on for about three and a half years. FreshBooks launched in 2004, with Levi Cooperman and Joe Sawada alongside him, first as an online invoicing tool aimed at technical freelancers and small shops. It was a slow climb. For a decade there was no outside money. The company paid its way through subscriptions and kept its team small while larger accounting products dominated the category. McDerment ran it as chief executive from 2003 to 2021. What almost nobody outside the company saw was how much it had to rebuild. In January 2015, McDerment incorporated a second company, BillSpring, to build new products without being held back by the old code. In 2016 it was folded back in and became the new FreshBooks. The original version lived on as FreshBooks Classic until 2022, when the last users were moved across. Money arrived after the product had proved itself. Oak Investment Partners led a US$30m Series A in 2014, with Georgian Partners and Atlas Venture. A further US$43m came in 2017. In August 2021, the company announced an US$80.75m Series E and US$50m in debt, reportedly putting its value above US$1bn. That year, McDerment handed the chief executive role to Don Epperson and stayed on as executive chair. FreshBooks says more than ten million people have used the product. Revenue is not published. What they do show is a founder who turned his own embarrassing error into a company, then had the discipline to throw away a working product and start again. The best ideas often begin as a small personal injury, and the founder who takes it seriously can turn it into a business. ## Vention (Canada) URL: https://startup.talk/canada/vention/ Founders: Etienne Lacroix, Max Windisch Founded: 2016; Based in: Montreal, Quebec; Sector: Manufacturing automation platform (cloud design, modular hardware and AI software) As a child, Etienne Lacroix was obsessed with Legos. He has told Chief Executive magazine that it was early training in creative problem-solving, and in how to build big things from small, standard parts. It is hard to read the story of Vention without thinking of that box of bricks. Lacroix became a mechanical engineer, and he spent his first decade in the trade. At GE Lighting he worked on LED product development, including the company's first LED effort for commercial agriculture. Then came a Harvard MBA in 2012 and a job at McKinsey, advising manufacturers on operations and product development. He saw plenty of factories and plenty of delay. Getting a custom machine built could take months. In 2014 he had the idea that would not leave him. Web technology, he reasoned, could let an engineer design a machine in the cloud, order the standard parts and receive them the next day, ready to assemble. It was Lego for factories. But he did not jump. What nearly no one outside his circle knew was that he held the idea for two years while still at McKinsey. In 2016 he left, and with Max Windisch he founded Vention in Montreal. A beta of the 3D design tool MachineBuilder followed in 2017, backed by about US$3.5m from White Star Capital, Bolt and Real Ventures. The company soon moved beyond parts. It launched the MachineMotion controller in 2018 and the code-free MachineLogic software in 2019, so that a factory worker could program a machine without writing code. Bain Capital Ventures led a Series A, and Georgian and Fidelity led a US$95m Series C in 2022, just as headlines were calling the market rapidly cooling. In January 2026, Vention raised a US$110m Series D, with Nvidia's venture arm taking part. The company says it has deployed more than 28,000 machines at over 4,000 factories, and total funding is above US$260m. Lacroix's own account of the hard years is short. Asked about failure, he tells leaders to learn from it, and we do not invent more. What the record does show is the patience of a man who waited two years for the right moment. A founder who carries an idea long enough, while learning how the customer actually works, often builds a sturdier company than one who leaves first and asks questions later. ## Jobber (Canada) URL: https://startup.talk/canada/jobber/ Founders: Sam Pillar, Forrest Zeisler Founded: 2011; Based in: Edmonton, Alberta; Sector: Field-service management software for home service businesses Two strangers kept ending up at the same coffee shop tables in Edmonton. Sam Pillar and Forrest Zeisler were both freelance programmers, each working on a personal project, each hunched over a laptop in the cafes that gave them cheap caffeine and a place to sit. They noticed each other because they kept noticing each other. Eventually they talked. Neither was building a company that day. They were building a living. Zeisler happened to know someone at a local painting business that wanted software to keep track of clients, quotes and invoices. The painters did not need an app that would change the world. They needed their paperwork to stop eating their evenings. Pillar later recalled being excited simply to be solving a real problem for a real business. So the two freelancers built it, listening closely to what the painters said they needed. Then they did the obvious thing: they asked who else had the same headache. Plumbers, landscapers, cleaners and electricians turned out to have exactly the same one. The ambition at the start was almost comically small. As it has been described, the target was US$10,000 a month, enough to cover costs and pay the two of them a modest salary. That is the part the later headlines leave out. Jobber did not begin as a plan to build Edmonton's biggest technology company. It began as a plan not to need a boss. What nobody quite expected was how patient it would be. Jobber took its first real institutional money in 2015, an US$8m Series A led by OMERS Ventures. A quiet inside round followed in late 2018, which Pillar counts as the Series B. Even then the company had raised only about US$16m in total. In January 2021 Summit Partners led a US$60m round. By then more than 100,000 service professionals in 47 countries were using the product, and recurring revenue had grown 90% in a single year. The price of building in Edmonton shows up in small ways. When the company wanted to hire people it called software engineers, it ended up in a legal dispute with Alberta's engineering regulator over what that phrase legally means. The pandemic brought a short hiring freeze, then more hiring. In February 2023 General Atlantic led a US$100m Series D, the largest venture round in the city's history. Pillar said Jobber had not needed to raise, and described the terms as founder-friendly. His guide, he said, was Warren Buffett's advice to be greedy when others are fearful, and Jobber had spent the boom watching rather than splurging. The pattern is easy to miss. Pillar and Zeisler never needed to guess at a market. They had one customer with one paper problem, and they built until it went away. Everything after that was repetition, done more carefully than their competitors thought necessary. ## Axelar (Canada) URL: https://startup.talk/canada/axelar/ Founders: Sergey Gorbunov, Georgios Vlachos Founded: 2020; Based in: Toronto, Ontario (team distributed); Sector: Blockchain interoperability network (crypto) Sergey Gorbunov still had a university office when his company hit a billion dollars. That is the odd detail in Axelar's story. In February 2022 the University of Waterloo announced that a computer science professor's startup had raised US$35 million and become a unicorn. Gorbunov, a cryptographer with a PhD from MIT, was the company's CEO. He was also still a professor, someone whose research dealt with secure systems and protocols for networks you cannot trust. That research was the real starting point. Blockchains are built to distrust everyone, which makes them good at holding value and terrible at talking to each other. A token on one chain could not simply walk over to another. Gorbunov later described the missing connection as the industry's biggest obstacle. He did not meet the problem alone. Georgios Vlachos, a fellow MIT graduate with a master's degree, had been on the founding team at Algorand, where the two had worked together. In 2020 they started Axelar to build a network whose only job was to move messages and assets between chains. The early money came from people who understood the pitch. A US$3.75 million seed round in 2020 included the venture firm DCVC, Binance's investment arm and the AngelList founder Naval Ravikant. A US$25 million Series A followed in mid-2021, in the middle of the crypto boom. What almost nobody outside the industry saw was the risk hiding inside the idea. A bridge between blockchains concentrates value in one place, and bridges became favourite targets of attackers through 2021 and 2022. Axelar's answer was to run its own network of validators and to treat security as the product rather than a feature. The valuation arrived just before the market turned. The February 2022 round, led by Dragonfly and Polychain with Rockaway and North Island Ventures also in, priced the company at US$1 billion. Then crypto went cold. Its AXL token was changing hands near US$0.04 in late 2025 according to market trackers, a long way from the headlines about a unicorn. Whether that gap matters depends on what Axelar is. As a token it has been through a hard cycle. As infrastructure, it remains one of several networks competing to connect blockchains, alongside LayerZero, Wormhole and Chainlink. The lesson is quieter than the valuation. Gorbunov did not leave the lab to chase a trend. He took a research question, how do you trust a message from a system you do not control, and treated the company as the experiment that would test it. ## Benevity (Canada) URL: https://startup.talk/canada/benevity/ Founders: Bryan de Lottinville Founded: 2008; Based in: Calgary, Alberta; Sector: Corporate purpose software (employee giving, volunteering, grants) Imagine a billion-dollar company whose first home smelled of shawarma. In 2008 four software developers sat in a small, peach-coloured office above a Calgary shawarma shop. They worked for Bryan de Lottinville, who had already done well. He had been chief operating officer of iStockphoto, the stock photo marketplace that Getty Images bought in 2006. The money was made. The question was what to do with the rest of a career. His own answer, as he has told interviewers, was a vow: to spend the second half aligning what he did with what he said he cared about. That is an unusual place to start a software company. Most founders chase a gap in a market. De Lottinville started with a discomfort. The gap turned out to be real. Companies were giving money away, matching employee donations and running volunteer days with spreadsheets and cheques. Giving was treated as a handout, not a system. De Lottinville wanted it to run like any other piece of corporate software, with employees choosing causes and the company handling the paperwork behind the scenes. Building that was slower than it sounds. By his account in a 2018 interview, the first couple of years went into the back end: the platform itself and the legal receiving entities needed in different countries so that a donation could actually reach a charity. There was no shortcut. It was infrastructure, built before anyone paid for it. Then came the quiet compounding. In 2010 Benevity became a certified B Corporation, one of only 15 in Canada. In 2016 Deloitte ranked it the third fastest-growing company in the country, on more than 5,800 per cent revenue growth. A year later it moved into its biggest office yet, in Calgary's Bridgeland neighbourhood. By 2019 revenue was about US$100 million, and a US$40 million round came from JMI Equity and General Atlantic. In December 2020 the investor Hg bought a majority stake in a deal valuing the company at about US$1.1 billion. De Lottinville stepped from chief executive to executive chairperson and Kelly Schmitt took over as CEO in 2021. For a first-time founder the most useful detail is the order of events. The vow came first, then the product, then years of unglamorous plumbing. The shawarma shop was only the address. ## Beacon Software (Canada) URL: https://startup.talk/canada/beacon-software/ Founders: Nilam Ganenthiran, Divya Gupta Founded: 2024; Based in: Toronto, Ontario; Sector: Acquirer of vertical software businesses, adding AI Nilam Ganenthiran had run Instacart as its president. In January 2024 he started over with a company that bought other people's overlooked software businesses. That was the move that surprised the industry. He had been a partner at D1 Capital, a large investment firm, and then a senior executive at a grocery delivery giant. A Canadian, he could have chosen almost any next step. He chose to buy small software and services companies in neglected corners of the economy: education, finance, logistics, recreation. The pitch was compact. Many of those firms were profitable, run by founders who had never needed venture capital and had never been given serious engineering help. Beacon would buy them, add AI and shared tech, design and fintech teams, and leave the founders to run their own shops. He joined forces with Divya Gupta, an American engineer and former Sequoia partner who had also worked at Databricks, and the early investors came fast. Amar Varma of Mantle and Fidji Simo, who runs applications at OpenAI, were among them. Then the pace became startling. By November 2025 Beacon had raised a US$250 million Series B led by General Catalyst, Lightspeed and D1 Capital, bringing total funding to about US$335 million and the valuation to US$1 billion. It was buying roughly one company every two weeks, more than 30 per cent of them Canadian, including Let's Camp, PowerUp and Viefund. Ganenthiran drew a hard line against the obvious comparison. Beacon, he told BetaKit, is the anti-private equity firm. It buys to hold forever, avoids cost-cutting, and prefers sellers to stay on. He also said the bootstrapped founders had been easier partners than venture-backed ones, who tend to arrive with different expectations. By mid-2026 the numbers had moved again. A US$225 million Series C came from General Catalyst and HarbourVest, sooner than he expected. More than 30 businesses had been acquired, and portfolio earnings were up more than 50 per cent, partly because the acquired companies had adopted AI. Beacon hired Mark Schaaf, formerly CTO at Superhuman and Instacart, as chief operating and product officer, and Goutham Buchi, formerly AngelList's CTO, as CTO. What remains to be seen is whether hold-forever survives integration at that speed. Buying companies is a quick skill. Keeping thirty founders motivated inside one structure is a slower one. The lesson for founders may be a reversal. Beacon makes money not by inventing something new, but by taking seriously what others had overlooked. Sometimes the opening is the business nobody was bothering to improve. ## HIVE Digital Technologies (Canada) URL: https://startup.talk/canada/hive-digital-technologies/ Founders: Founded: 2017; Based in: Vancouver, British Columbia; Sector: Bitcoin mining and high-performance computing data centres In fiscal 2026 HIVE mined 2,885 bitcoin and lost US$148 million. Both facts are true, and together they describe a company in the middle of changing its mind. HIVE began as a bet that crypto mining could be a public-market business rather than a garage hobby. In 2017 it listed on the TSX Venture Exchange, billed as the first cryptocurrency mining company to do so. Vancouver gave it a home, but the machines went where the electricity was cheap. In 2018 and 2019 it expanded to Sweden and Iceland, leaning on renewable power. That choice defined the brand. HIVE sold itself as green: hydro and geothermal rather than coal. It also made a quiet decision to keep part of what it mined, a practice the crypto world calls HODLing, and has held bitcoin since March 2019. For years the company mined Ethereum on graphics cards. Then, in 2022, Ethereum switched to proof of stake and the work disappeared overnight. HIVE refocused on bitcoin, which needs specialised machines, and in 2023 renamed itself HIVE Digital Technologies to reflect a broader scope. The graphics-processor business did not die. It turned into something investors found more exciting, high-performance computing for AI, under a banner called BUZZ. The numbers show the pivot half-finished. In the year to March 31, 2026, revenue reached US$297.8 million, up 158 per cent. Mining accounted for US$278.3 million and BUZZ hosting US$19.5 million. Hashrate stood at 25.1 exahash per second. But the GAAP net loss was US$148.4 million, mostly depreciation and unrealised investment losses, and adjusted EBITDA turned negative in the final quarter, when gross operating margin dropped to 24 per cent. By November 2026 the market valued the company at about US$474 million, and it trades on NASDAQ as well as in Canada. The sceptical reading is that a miner calling itself an AI company is following the money. The generous one is that miners already own the hard things AI needs: cheap power, cooling, buildings and the people who can run them. Contracted annual recurring revenue from computing was about US$35 million at year end. For a founder, the story is about assets outliving their original use. Ethereum mining ended in a day. The hardware, the power contracts and the know-how did not. What the company does with them next will decide whether 2022 was an ending or a transfer. ## Lightspeed (Canada) URL: https://startup.talk/canada/lightspeed/ Founders: Dax Dasilva Founded: 2005; Based in: Montreal, Quebec; Sector: Point-of-sale and commerce software (retail, hospitality, e-commerce) Before Dax Dasilva was a founder, he was a thirteen-year-old apprentice sweeping up behind a software developer. He had been given a Macintosh by his father, and he used it to learn how to build program interfaces. At thirteen he talked his way into an apprenticeship. It was an odd start for a boy from Vancouver whose parents had arrived in Canada as refugees. They had fled Idi Amin's Uganda in 1972, with little more than the idea that a country could be safe. That is the fact that rarely makes the headlines about Lightspeed. The company is usually described through its numbers. The man behind it came from a family that began again from nothing, and he spent his teens in the odd space between code and conscience. At fourteen he came out as gay. As a young adult he took part in the 1993 Clayoquot protests against old-growth logging. At the University of British Columbia he began in computer science, then moved to art history and religion. So when he moved to Montreal and started Lightspeed in 2005, he was not a typical software founder. His idea was simple and a little moral: independent shops should be able to compete with the big chains. A small retailer should not need a giant company's budget to have a good till. What almost nobody expected was how long that small idea would take to grow. For seven years Lightspeed was a modest Montreal outfit. Then in 2012 Accel Partners led a US$30m round, followed by iNovia in 2014 and the Quebec pension fund CDPQ in 2015. Dasilva used the money to buy his way sideways. In 2014 he bought a Belgian restaurant-software startup, POSIOS, and the next year the Dutch e-commerce firm SEOShop. A retail till became a platform for restaurants and online stores too. In March 2019 Lightspeed listed on the Toronto Stock Exchange. Then came two large purchases, ShopKeep for US$440m in 2020 and Vend for US$350m in 2021. They made it big, and they made it a target. In September 2021 short seller Spruce Point accused it of inflating its numbers. The shares fell 11% on the day, and the company called the report misleading. By July 2022 the stock had fallen to about $22, and layoffs followed in 2023 and 2024. Dasilva stepped down as CEO in February 2022. Today Lightspeed reports revenue of roughly US$1.2bn for fiscal 2026 and about 3,000 employees. The founder's own story is the quiet payoff. The refugee family's son who learned code from a gift in his father's hands built tools for other small, independent people. The lesson is not that scale is easy, because it clearly wasn't. It is that the pull you feel at thirteen can still be steering you at forty. ## Wrapbook (Canada) URL: https://startup.talk/canada/wrapbook/ Founders: Ali Javid, Cameron Woodward, Hesham El-Nahhas, Naysawn Naji Founded: 2018; Based in: Toronto, Ontario and New York; Sector: Payroll, insurance and compliance software for film, TV and live-event production In November 2021 a payroll company for film crews announced that it was worth US$1bn. Wrapbook was three years old. It had raised a seed round of C$4.7m in August 2020 and a C$34m Series A in March 2021, only eight months before the headline round. Tiger Global led the US$100m Series B, joined by Andreessen Horowitz, Jeffrey Katzenberg's WndrCo and others. The founders behind that number are Ali Javid, Cameron Woodward, Hesham El-Nahhas and Naysawn Naji. The company launched in 2018, with a team that sits between Toronto and New York. All its product and engineering staff were in Ontario, and about 90 people worked there by late 2021, roughly half of them in Ontario. The market it picked was not glamorous. Making a film or a TV show means hiring dozens or hundreds of people for a few weeks, signing paperwork for every one of them, then paying them correctly across states and unions. The big names in this corner of Hollywood had been there for decades. Wrapbook wagered that the work could be moved online, and that crews would like a profile they carry from job to job instead of filling in the same forms again. The early numbers bore out the bet. By 2021 Wrapbook supported more than 1,000 projects a month across North America. The share of workers who reused their profile with more than two employers rose to 23% in 2021, double the 2020 rate. Studios, networks, production companies, commercials and live events were all on its customer list. What it cost is visible in the later numbers. In September 2024 Wrapbook raised another US$20m, led by Bessemer Venture Partners. This time it was valued at about US$750m, below the US$1bn of 2021. Many companies of that vintage were repriced after the cheap-money years, and Wrapbook was one of them. Yet the company kept growing and kept raising, which is not true of all its peers. In total it has raised roughly US$150m. There is a founder lesson in the shape of that curve. A giant valuation is a measurement of a moment, not of a business. The part that held was the dull, repeatable one: a worker's file that follows them to the next job, and a producer who does not chase paperwork at midnight. The company that wins in an old, crowded back office usually wins by making one chore vanish. The scoreboard moves. The chore either disappears or it does not. ## Tilray (Canada) URL: https://startup.talk/canada/tilray/ Founders: Brendan Kennedy Founded: 2013; Based in: Leamington, Ontario and New York (production base Nanaimo, British Columbia); Sector: Cannabis, beverage alcohol and wellness The hardest conversation Brendan Kennedy says he had about getting rich was not with a banker. It was with his in-laws. By the autumn of 2018 he had good reason to have it. Tilray had floated on Nasdaq that July at US$17 a share, the first cannabis company to trade on a major American exchange. By September the stock touched US$214, and Bloomberg estimated his net worth at no less than US$2.4bn. Headlines called him the marijuana billionaire who does not smoke. He came from a very different world. Kennedy is the sixth of seven children in a San Francisco family. He studied architecture at Berkeley, took a master's in engineering at the University of Washington, and later an MBA at Yale. He worked at SVB Analytics, an affiliate of Silicon Valley Bank, where he met Christian Groh. With Groh and Michael Blue he started Privateer Holdings, Seattle-based, after he experimented with cannabis and saw an industry waiting for a serious investor. Tilray was the Canadian leg of that bet. It was created in 2013 under Privateer and became one of Canada's first licensed producers. In late 2014 it took money from Founders Fund, which was described as the cannabis industry's first institutional investment. In 2016 it ran a Health Canada-approved clinical trial, a first for a cannabis company. What almost nobody saw coming was how fast the peak would pass. By August 2019 the stock was near US$29. Kennedy's pay fell from about US$32m in 2018 to US$3.5m in 2019, and by February 2021 his net worth was reported at US$238m. He stepped down as CEO in 2020, as Tilray merged with Canadian rival Aphria in December and Irwin Simon took over. The merged company kept the Tilray name and the Nasdaq ticker, and it tried to widen out. It closed a large greenhouse in 2020. It bought into craft beer, and in 2023 it paid US$85m for eight Anheuser-Busch brands including Redhook and Widmer. It struggled with its brewing deals; some had been declining for years. In 2024 revenue was about US$789m, and the company posted a net loss of roughly US$222m. The figures tell one part of the story. The other sits in what Kennedy said about money itself. Sudden wealth had changed the terms of his family, and even a cannabis billionaire had to explain it across a dinner table. For first-time founders the lesson is plain. A valuation can double in a summer and halve in a year, but the people who knew you before the number are the ones you will still have to face. ## Copperleaf (Canada) URL: https://startup.talk/canada/copperleaf/ Founders: Judi Hess Founded: 2000; Based in: Vancouver, British Columbia; Sector: Decision analytics software for asset-heavy industries (utilities, energy, infrastructure) Judi Hess had already sold one company to a giant before she walked into Copperleaf. She was a software engineer at MacDonald Dettwiler in the 1980s, then rose to president of Creo, the Vancouver-area printing technology company. In 2005 Eastman Kodak bought Creo for almost US$1bn. Most people would have stopped there. Instead, in 2009, she joined a smaller Vancouver company called Copperleaf as CEO. Copperleaf had begun in 2000 as a consultancy. By the late 2000s it was turning into a software maker, and the transition mattered more than anything else. Consultants sell hours. Software sells the same idea again and again. Its product helps utilities, pipelines and infrastructure owners decide which assets to repair or replace first, and when, with a limited budget. It is easy to underrate how dull and how large that problem is. A power utility owns thousands of transformers and lines. Money for upgrades is never enough. Copperleaf built a way to rank every possible project by value and risk, and to show the board why. The company later said its software helped manage more than US$2.9 trillion in assets around the world. What almost nobody outside the industry saw was how long the climb took. It spent more than twenty years from the first consultancy to a stock market listing. In October 2021 Copperleaf went public on the Toronto exchange at C$15 a share, raising C$140.1m. Orders reached about C$1.4bn, and the Globe and Mail put its market value at C$1.78bn, making it one of the few female-led Canadian tech firms with a ten-figure valuation. The next chapters came fast. Hess moved to vice-chair and chief strategist, handing the chief executive role to enterprise sales veteran Paul Sakrzewski. In June 2024 IFS agreed to buy the company, and the deal closed that year at C$12 a share, about C$1bn. That was below the IPO price. IFS said it would keep investing in Vancouver. The arc is unusual. Someone who had already helped build and sell one company chose to build another. Then she stepped back from the chair when a sales-led leader could take it further. For founders the useful lesson may be the quiet one: knowing which stage of a company you are best at is a skill, and stepping aside at the right moment is part of it. ## Docebo (Canada) URL: https://startup.talk/canada/docebo/ Founders: Claudio Erba Founded: 2005; Based in: Toronto, Ontario (originally Biassono, Italy); Sector: Cloud learning management software (AI-powered) Claudio Erba started Docebo while he was still a lecturer, standing in front of university students and teaching them how content management systems worked. He studied economics and marketing at the Catholic University of the Sacred Heart in Milan. For several years in the early 2000s he was a guest lecturer at the University of Florence, teaching about content systems. In 2005 he founded Docebo, from the Brianza region near Milan. In a 2019 Forbes Italia profile he credited that region for the company's work ethic. What almost nobody realises is how local it was. For years Docebo was mainly an Italian business, selling cloud-based learning software to Italian customers. There was no global plan, only a lecturer's idea that companies needed an easier way to train their staff online. In 2012 an Italian venture firm, Principia SGR, put in EUR 2.4m. Erba told EconomyUp in a 2013 interview, 'We bet on digital and we won.' He also spoke about the strain of becoming more industrial as the company grew. In 2013 Docebo opened its first American office in Athens, Georgia. The real turn was a Canadian cheque. In 2015 Klass Capital invested US$3m, and the company moved its headquarters to Toronto. That is not an obvious thing for an Italian founder to do. It put a small Italian software firm into a larger market, with a public-market ecosystem to match. The company went to the Toronto Stock Exchange in October 2019 and raised C$75m. Fourteen months later, in December 2020, it listed on Nasdaq and raised US$144m. It uses the same DCBO ticker on both. Docebo today serves customers in more than 90 countries. As of the third quarter of 2025 its annual recurring revenue was US$235.6m and it counted 3,978 customers. The price of building something this size is that a founder eventually has to step back. On 29 February 2024 Erba moved from chief executive to chief innovation officer. Alessio Artuffo became interim CEO on 1 March 2024 and was made permanent that September. There is an unspoken lesson in the sequence. The classroom is where the idea formed, and Erba never fully left it: he started as a teacher, and he ended in a role about ideas. A founder does not always need to run the company forever. Sometimes the job is to start the lesson and let someone else lead the class. ## Celonis (Germany) URL: https://startup.talk/germany/celonis/ Founders: Alexander Rinke, Bastian Nominacher, Martin Klenk Founded: 2011; Based in: Munich, Bavaria; Sector: Process mining and execution management software Fifteen hundred handwritten letters. That was the sales team at Celonis in its first months: three students, a pile of envelopes, and a theory that a receptionist will bin a printed letter but hesitate over a handwritten one. The three were Alexander Rinke, Bastian Nominacher and Martin Klenk, all at Munich universities and all working through a student consultancy called Academy Consult. In 2011 it sent them to Bayerischer Rundfunk, the Bavarian broadcaster, where some internal processes took up to five days. They watched how work actually moved. Then, in their student dorm, they wrote simple software that did one thing: measure the gaps between timestamps. The biggest gaps showed where work was stuck. They presented the result on a few sheets of A4. The broadcaster loved it. Processes that had taken five days fell to a few hours. What almost nobody sees behind that tidy story is how little of a company it was. Rinke, who a BBC piece says started his first business at 15 by tutoring school pupils, had a product nobody could easily describe. Investors, he and the others have recalled, called it too difficult to understand and too complex. So Celonis stayed bootstrapped. Even the name nearly went wrong: the founders weighed Zelos, after the Greek spirit of zeal, and Zelonis, then dropped the latter because it would sit at the end of every alphabetical list. They settled on Celonis just before the notary appointment. That is where the letters came in. Without money for marketing, they wrote to chief executives by hand. Siemens followed the broadcaster as a customer, partly through that campaign. The turn came through a very German doorway. In 2012 Celonis joined SAP's Startup Focus programme, and in 2015 it signed a reseller deal with SAP, putting its software in front of the people who ran the world's corporate back offices. Venture money finally came: US$27.5m in 2016, then a billion-dollar valuation in 2018, and US$290m the next year. In 2019 the three received the German Future Prize. The price of success has been dependence. The channel that carried Celonis to big companies belongs to a company that now sells competing tools, and the founders have spent years making their product stand on its own, with acquisitions such as Integromat in 2020. By 2021 an US$1bn round valued the company at US$11bn; a 2022 extension led by Qatar's sovereign fund pushed that to about US$13bn. Today it has around 3,000 employees. The lesson is not about process mining. When no one will fund an idea they cannot understand, show them one small number on one sheet of paper, and if nobody answers the phone, write by hand. ## N26 (Germany) URL: https://startup.talk/germany/n26/ Founders: Valentin Stalf, Maximilian Tayenthal Founded: 2013; Based in: Berlin, Berlin; Sector: Digital bank (fintech) The first thing Valentin Stalf and Maximilian Tayenthal tried to sell was a prepaid card for teenagers. That is not the story N26 tells today. In August 2013 the two Austrians registered a Berlin company under a name almost nobody now remembers, Papayer GmbH. Stalf, born in Vienna in 1985, had studied accounting and finance at the University of St. Gallen and, while still a student, worked in M&A at Deutsche Bank and in strategy consulting at Roland Berger. Then he became an entrepreneur in residence at Rocket Internet, where he helped build payment companies. He knew what a bank looked like from the inside and what a payments startup looked like from the outside. The teen card launched in February 2014. In June a EUR2m seed round arrived, and with it a decision to throw most of the idea away. The company became Number26, dropped the teenagers and built a proper current account with an app. It had no licence yet; it borrowed one, running on a partner's backend until 2016. Then came the turn. In 2016 the German regulator, BaFin, granted a banking licence and the name shrank to N26. Accounts opened across the eurozone. Tencent and Allianz X came in with US$160m in 2018, and in January 2019 a US$300m round at US$2.7bn made it Europe's most valuable mobile bank. In October 2021 a US$900m round valued it at US$9bn. What almost nobody in the cheering years saw was the cost of speed. Accounts were closed in 2016 over suspected money laundering. By 2021 BaFin had fined the bank EUR4.25m over weak anti-money-laundering controls, installed a supervisor and capped new customers at 50,000 a month. Italy barred new customers for a time. The bank withdrew from the UK in 2020 and the US in 2022, and in 2025 the regulator threatened fresh sanctions. In August 2025 Stalf stepped down as co-CEO after tension with investors, and Tayenthal left operations at the end of that year. Mike Dargan took over in April 2026. Still, the numbers are real: revenue of EUR120.4m in 2021, customers in 21 countries, and a company that now offers share trading as well as accounts. For first-time founders the lesson sits in the name. N26 only exists because its founders were willing to bury Papayer after eight months. A bank, it turns out, is easier to start than to run. ## DeepL (Germany) URL: https://startup.talk/germany/deepl/ Founders: Jaroslaw (Jarek) Kutylowski Founded: 2017; Based in: Cologne, North Rhine-Westphalia; Sector: AI language translation and writing At twelve, Jarek Kutylowski sat down in a German classroom and could not understand a word. He had been born in Poland and moved to Germany at that age, and he has said he went to school there without speaking German. A speaker biography adds a second detail: he had been programming since he was ten. Put those together and you get a boy for whom language was a wall and code was a way over it. He became a computer scientist and ended up at Linguee, an online dictionary that matched words to real translated sentences and paid for itself with advertising. By 2016 Kutylowski, the company's chief technology officer, was leading a team building something different inside it: a translation system based on neural networks. He saw, he later told interviewers, that these networks could make a breakthrough in machine translation, and he decided to build something great. What almost nobody outside Cologne expected was where it would be built. On 28 August 2017 the system launched as DeepL Translator, and the company behind Linguee became DeepL GmbH. At launch it ran on a 5.1-petaflop supercomputer in Iceland, powered by hydroelectric and geothermal energy. Seven languages at first: English, German, French, Spanish, Italian, Polish and Dutch. Reviewers noticed that the output sounded less like a machine and more like a person. They also noticed that Google offered far more languages. The next years were a slow narrowing of that gap. Windows and Mac software came in 2019, the year he became CEO. Chinese and Japanese arrived in 2020. In January 2023 a funding round of about EUR100m valued the company at EUR1bn, making it Cologne's most valuable startup. In May 2024 a US$300m round lifted the value to US$2bn. The hard part is that the ground keeps moving. A 2025 academic benchmark ranked general-purpose AI models above DeepL on translation, and the company has answered with its own translation language model in 2024, an AI agent in 2025 and voice-to-voice translation in 2026. It now serves more than 200,000 business customers. The lesson is hidden in his childhood. The person who knows how it feels to be unable to say what you mean is often the one who builds the tool for it. ## Personio (Germany) URL: https://startup.talk/germany/personio/ Founders: Hanno Renner, Ignaz Forstmeier, Roman Schumacher, Arseniy Vershinin Founded: 2015; Based in: Munich, Bavaria; Sector: HR software (SaaS) Before Personio raised its first real money, the company's bank account held about 200 euros. Hanno Renner told CNBC as much, years later, when the company was valued at more than US$6bn. The founders first had to pay outstanding invoices, a standard condition of a funding deal, and the money that remained was barely enough for a week of lunches. It was an odd place for a man who had spent his student holidays crossing oceans. Renner earned a yachtmaster's licence so he could work as a skipper, and during college breaks he was paid to sail boats around the world. He said it was intense and a harder job than most people think. It paid for his studies in industrial engineering and technology management at TU Munich and the CDTM, the entrepreneurship centre where he met his three co-founders, Roman Schumacher, Ignaz Forstmeier and Arseniy Vershinin. He also spent time at Columbia University in New York. In 2015, soon after graduating, the four started Personio. They had no office, so they built their first product wherever they could find space at college. The idea was dull in the best way: HR software for small and medium companies that were still running on spreadsheets and email. What almost nobody saw was how they paid for the early months. Customers paid for the first version, and the founders used that revenue to buy software licences, rent a small office and hire a handful of people. Word of mouth did the selling; reports describe growth from around 20 to over 100 companies in 2016. Only then, in July 2016, did a EUR2.1m seed round arrive, with Global Founders Capital among the investors. A US$12m Series A led by Northzone followed in 2017. In January 2021 a US$125m round made Personio a unicorn; by autumn that year it was valued at US$6.3bn, and in 2022 at US$8.5bn. By 2024 it served about 12,000 customers in 70 countries and employed roughly 2,000 people. The cost of that climb is that every later valuation had to be defended. The 200 euros are a useful reminder of what the company was made of before anyone was watching. A founder who has skippered a yacht across open water has already learned to keep sailing when the shore is out of sight. ## Flix (Germany) URL: https://startup.talk/germany/flix/ Founders: Daniel Krauss, André Schwämmlein, Jochen Engert Founded: 2011; Based in: Munich, Bavaria; Sector: Intercity bus and rail travel platform For almost a year, three men in Munich could not decide whether to buy a bus. Daniel Krauss later called that deliberation the most important early decision FlixBus made. Should the company rent, lease, buy or partner with the family bus firms that already ran Germany's roads? The founders spent a year touring the country to meet operators before they chose to keep the vehicles out of the business entirely. It was not their first attempt at being founders. Krauss and André Schwämmlein were school friends from Franconia. In 2005 they started a company together and sold it two years later. Both then studied in China; Krauss worked at an automotive supplier in the United States and later at Microsoft in Munich while finishing a master's degree in Erlangen. Jochen Engert, a football teammate of Schwämmlein's, joined the venture. They collected thousands of ideas, Krauss has said, from knitted socks to online portals. The trigger was a law. Germany's plans to loosen its long-distance passenger transport rules, discussed from 2009, meant a closed market was about to open. In 2011 they formed a partnership in Munich, later turned into a limited company. The first angel investor was Heinz Raufer, founder of hotel.de, whom they met through a business-plan contest; Harald Baumann of Baumann Bus brought money and industry knowledge. By the end of 2012 the team numbered 30. In February 2013 the first buses rolled, on three routes in Bavaria. What almost nobody outside the industry saw was how close it came to damage. MeinFernbus, a Berlin rival, had more bus partners and about nine months of head start. FlixBus had stronger technology and more aggressive marketing. They fought for about a year until both saw that the war threatened them. In January 2015 they merged and kept the more international name, with MeinFernbus's green colour. Postbus followed in 2016, and by 2015 the buses had carried 20 million passengers. The price has been public scrutiny: crashes, an American investigation into wheelchair access in 2025 and a German fine over green claims. Still, the business carried a total revenue of about EUR2bn in 2023, with adjusted EBITDA of EUR104m, and in 2024 EQT and Kuehne + Nagel took a 35% stake for EUR1bn. The lesson is in the year of indecision. Krauss now says the right investor and the right partners matter as much as cash. ## Helsing (Germany) URL: https://startup.talk/germany/helsing/ Founders: Torsten Reil, Gundbert Scherf, Niklas Köhler Founded: 2021; Based in: Munich, Bavaria; Sector: Defence technology (AI software, drones, autonomous aircraft) Torsten Reil spent his twenties teaching digital bodies how to fall over. He was a biologist at Oxford, working on how living things move, and from that research he built NaturalMotion, a company whose software made video-game characters stumble and recover like real people. Zynga bought it in 2014. For most founders that would have been the end of the story. Reil, who also studied evolutionary systems at Sussex, was only getting to the question that actually bothered him. That year Russia annexed Crimea. Reil has said it was part of what pushed him toward defence. In the same years, across the table in Berlin, a McKinsey partner named Gundbert Scherf was serving as a Special Representative at the German defence ministry, from 2014 to 2016. One man understood machine learning and the other understood how a ministry buys things. What almost nobody outside the trade noticed was the third piece. Niklas Köhler, a physicist, had started a deep-learning company called Hellsicht in 2017. When the three founded Helsing in Munich in March 2021, that small company was folded in. The pitch was plain: AI would decide who sees the battlefield first, and Europe was not building it. Scherf later put it this way in a funding announcement, saying AI would be crucial for democracies to defend their values. The cheque that made it real came in November 2021, about €100m from Prima Materia, the vehicle of Spotify founder Daniel Ek, who now chairs the company. It was a strange sight, a music-streaming billionaire funding weapons software, and it cost him. Musicians pulled their catalogues from Spotify in protest, in 2021 and again in 2025. The company grew faster than its critics could count. It reached a €1.7bn valuation in 2023, the first European defence unicorn, then €12bn in 2025. It supplied 4,000 HX-2 drones to Ukraine, bought the aircraft maker Grob, flew an AI pilot named Centaur in a Swedish Gripen jet, and unveiled an autonomous combat aircraft. In July 2026 it raised about US$1.8bn at an US$18bn valuation. The price has been trust. Rheinmetall, an early partner, walked away in 2024 and sided with a rival. In April 2025 Bloomberg reported that former employees, investors and military experts alleged overpriced drones and shaky software. Some industry voices call the company secretive. Helsing keeps certain production sites quiet, saying it fears Russian sabotage. Reil's path is an odd one: from teaching animated characters to fall, to building machines that decide in a split second. The lesson for a founder is not about weapons. It is that an exit can be a rehearsal. The skills he sold in 2014 were the ones that let him walk into defence a decade later with something to offer. ## Trade Republic (Germany) URL: https://startup.talk/germany/trade-republic/ Founders: Christian Hecker, Thomas Pischke, Marco Cancellieri Founded: 2015; Based in: Berlin, Berlin; Sector: Fintech (neobroker and bank) The name on the first door was not Trade Republic. It was Neon Trading. In 2015 three founders sat in the start-up incubator of comdirect, a German online bank, in Munich. Christian Hecker had studied philosophy and spent years in finance. Thomas Pischke was a physicist. Marco Cancellieri was a computer scientist. Their idea was easy to state and almost absurd for the time: a German version of Robinhood, a phone app where ordinary people could buy shares without paying a bank's fees. The early years were slow. Germans are famous for keeping money in savings accounts, and shares have never been a national habit. The company needed more than a clever app. It needed a patient owner. In 2017 Sino AG, a Düsseldorf financial firm, took a majority stake, and its chief, Ingo Hillen, put in about three million euros and became a managing director. It was small money for a financial company, and it kept the founders alive. The technical heart was Pischke's. Investors have said he built the trading architecture so that trades could be executed extremely cheaply, which is why Trade Republic could charge a flat €1 per order. Hecker, meanwhile, was described by backers as the unofficial product chief behind the design of the app. The venture funds arrived in 2019 and 2020 with over €60m from Creandum, Project A, Accel and Founders Fund, and Sino lost its majority. In May 2021 came US$900m at a US$5bn valuation. The price of speed came quickly. In January 2021, during the GameStop frenzy, the app stopped customers from buying certain shares, and more than 4,000 people complained to the regulator, BaFin. In June 2022 the company cut staff when markets turned. In 2024 and 2025, customers complained about slow customer service and delayed depot transfers, and a consumer group sued over claims about interest and deposit protection. The company kept going. It won a full banking licence from the European Central Bank in December 2023, ending its dependence on partner banks. For the year to 2024 it reported a profit of €34.8m on about €340m of revenue. By April 2026 it said it served about 10 million customers holding around €150bn, and a share sale in December 2025 valued it at €12.5bn. The lesson for founders is in the first name. Neon Trading was a product idea. Trade Republic became a bank. The change came from someone patient enough to hold the stake before the venture money believed. ## Black Forest Labs (Germany) URL: https://startup.talk/germany/black-forest-labs/ Founders: Robin Rombach, Andreas Blattmann, Patrick Esser Founded: 2024; Based in: Freiburg im Breisgau, Baden-Württemberg; Sector: Generative AI (image and video models) The paper that changed the internet's pictures was written by research assistants. In April 2022 Robin Rombach, Andreas Blattmann and Patrick Esser, working under the professor Björn Ommer at LMU Munich, published a method for generating images. It became the foundation of Stable Diffusion. By the end of that year the tool had been downloaded by millions, and the company that carried the name, Stability AI, was the one that got the credit. The three had gone to work there. Then they left. In 2024 they founded Black Forest Labs in Freiburg, in the Black Forest, with seven others. They chose a small German town over Silicon Valley, and the name made the point. The first cheque was US$31m, from Andreessen Horowitz and a handful of prominent individuals, among them Brendan Iribe, Michael Ovitz and Garry Tan. In August 2024 they released Flux.1. It scored above DALL-E 3 and Midjourney on a popular rating system in mid-2024, came in versions developers could use, and showed up inside xAI's Grok chatbot within weeks. Mistral's Le Chat added Flux in November. What almost nobody saw coming was the break. In December 2024 xAI moved to its own model, Aurora, and Musk's company ended the cooperation. The founders were praised for refusing to work with it again. They had other customers. Nvidia partnered in January 2025, Adobe put Flux.1 Kontext inside Photoshop in September 2025, and Flux.2 followed in November. In December 2025 a US$300m round led by Salesforce Ventures and AMP valued the company at US$3.25bn. Total funding passed US$450m. By mid-2026 it had about 100 employees, with 30 in San Francisco, and called itself first of all a research lab. In June 2026 Martin Scorsese joined as an adviser, a choice that drew pushback from the Art Directors Guild. The cost is a cloud that has never fully lifted. The founders do not disclose what the models were trained on. Ars Technica suspected scraped internet images; the founders told an interviewer they did not use the LAION set but would not list their sources. Critics also worry about how real the pictures look. And in a field that moves by the month, one benchmark in August 2026 put Flux 2 only fifteenth of 148. For a first-time founder the story is a quiet one. Credit and ownership are different things, and the people who write the method can still build the company after someone else has carried the name. ## Raisin (Germany) URL: https://startup.talk/germany/raisin/ Founders: Tamaz Georgadze, Frank Freund, Michael Stephan Founded: 2012; Based in: Berlin, Berlin; Sector: Fintech (savings and investment marketplace) A teenager in Georgia once wrote to the president of the United States, and the president wrote back. The reply, reported in a Tagesspiegel profile, came from George Bush Senior. The boy, Tamaz Georgadze, had written in the name of Eduard Shevardnadze, then Georgia's head of state. It is the kind of detail that reads like a joke until you notice what it shows: a young man who understood that doors open when you write to the person on the other side. He went on to work, in 1994 and 1995, as an aide to the President of Georgia in the foreign relations department. He studied international economics at Tbilisi State University, then law and agricultural economics at the University of Giessen in Germany, where he earned a doctorate. Then he joined McKinsey. He stayed ten years and became a principal. The idea that took him away from consulting was almost dull. German savers kept money at home, at low rates, while banks across Europe paid more. Georgadze, with Frank Freund and Michael Stephan, built a single website where a saver could open one account and place money with banks in other countries. The company, incorporated on 18 December 2012, began as Weltsparen and later became Raisin. In a podcast he described the early days in a Berlin basement. What almost nobody tells you about boring ideas is how long they take. A company that moves money must be trusted before it is used, and it must sign banks before it has savers. By one profile, Raisin needed roughly a decade to turn a profit. The prize at the end was scale. By June 2025 savers had put more than 75bn through the platform and a company profile counts more than 4bn in returns for savers. A Series E in 2023 reportedly valued the business at about US$1.1bn. In December 2024, a secondary sale of over €100m in shares put its value above €2bn. Raisin has never commented on its valuation. The price was patience. Fintech valuations fell sharply after 2021, and some outlets reported that Raisin lost its unicorn tag before recovering. The company converted to a European company form, Raisin SE, in 2025. Georgadze's life offers one repeatable lesson. The teenager who wrote letters to the powerful and the executive who set up a savings site are the same person. He asked for something that sounded unlikely, in writing, and kept asking until the other side answered. ## BioNTech (Germany) URL: https://startup.talk/germany/biontech/ Founders: Uğur Şahin, Özlem Türeci, Christoph Huber Founded: 2008; Based in: Mainz, Rhineland-Palatinate; Sector: Biotechnology (mRNA vaccines and cancer immunotherapy) A teacher looked at a Turkish boy in Cologne and recommended the lowest secondary school. Uğur Şahin had been born in İskenderun in 1965. At four he moved with his mother to Germany to join his father, who worked at the Ford factories in Cologne. He loved football and borrowed popular science books from the library. When it came time to choose a school, his primary teacher suggested a Hauptschule, the track that leads away from university. It was a German neighbour who stepped in. Şahin went to the gymnasium instead. What almost nobody remembers when they tell the vaccine story is that it did not begin with a virus. It began with two doctors in Homburg. Şahin, by then a hematologist and oncologist, met Özlem Türeci, a medical student at Saarland University Hospital, born in Siegen to a surgeon from Istanbul. As a child she had been inspired by the nuns who worked at her father's Catholic hospital. What they shared was a belief that the immune system could be taught to hunt tumours. They co-founded Ganymed in 2001 and married in 2002. Fox Business reported that they left the lab at lunchtime to marry at the registry office and were back in their white coats afterwards. In 2008 they started BioNTech in Mainz, with money from the Strüngmann brothers that outlets put between €150m and €180m. Katalin Karikó, a future Nobel laureate, joined in 2013. The company listed on Nasdaq in October 2019. Then, in January 2020, they put the company's effort behind a COVID vaccine. They called it Project Lightspeed. They paired with Pfizer, started trials in April, and announced 95% efficacy in November. By August 2021 the market value had briefly passed US$100bn. Bloomberg put Şahin's fortune at US$7.22bn in February 2022. He and Türeci became the first Germans of Turkish roots among the country's hundred richest. The family reportedly lives modestly. The price came afterwards. A vaccine offer disclosed in 2021 had asked for €54.08 per dose in June 2020, and the price was later cut to €19.50. Demand collapsed. For 2025 the company reported revenue of €2.87bn and a net loss of €1.14bn. In May 2026 it announced it would end COVID vaccine production in Germany and close several sites, shifting toward oncology. In March 2026 it was reported that Şahin and Türeci would leave by the end of the year to start a new biotech. The boy who nearly missed the gymnasium ended where he began: in a lab, working on cancer. The lesson is not that talent breaks through. It is that someone, a neighbour in this case, has to open the door at the right moment, and you must remember to hold it for others. ## Auto1 Group (Germany) URL: https://startup.talk/germany/auto1-group/ Founders: Christian Bertermann, Hakan Koç Founded: 2012; Based in: Berlin, Berlin; Sector: Used-car trading / online automotive marketplace Two cars belonging to a grandmother in Germany. That is where, by the company's own legend, Auto1 begins. Christian Bertermann tried to help her sell them and found himself buried under offers that smelled wrong: lowballers, strangers, haggling with no rules. He has described the used-car trade as unprofessional, and the story goes that this irritation, more than any grand plan, pushed him to build a business. He was not a car man by training. Bertermann studied business administration at the University of Mannheim and then went to the fast world of Berlin start-ups, where he ran product at CityDeal and then international product at Groupon. His future partner, Hakan Koç, came at it from another angle: a law degree from Bucerius Law School in Hamburg, followed by a stint inside Rocket Internet and a product job at the furniture retailer home24. Together they founded Auto1 in Berlin in 2012. The idea sounds almost boring now. A private seller enters a car, the company makes a fast offer, the car is inspected, and then it is sold on to a dealer through an online marketplace. But that boring idea attacked the exact thing Bertermann had felt at his grandmother's side: the seller's fear of being cheated, and the dealer's fear of buying blind. What nobody expected from two product people was how physical the business would be. Cars have to be collected, stored, checked and moved across borders. Auto1 became a logistics company wearing a software jacket, with hundreds of branches where sellers can drive in. It also became a company that needed enormous amounts of money. In 2016 it launched Autohero to sell cars directly to consumers, and in 2018 it set up a financing venture with Allianz and Deutsche Bank. The roads were not always smooth: in 2021 German press questioned how those two banks had dealt with investors linked to Northern Cypriot banks. That was an allegation from reporters, not a finding. On 4 February 2021 the company listed in Frankfurt, planning to raise about EUR 1.8bn, much of it earmarked for Autohero. Forbes soon estimated each founder at roughly USD 1.3bn on paper. By 2025 the group reported revenue of EUR 8.17bn and says it moves about 3,000 vehicles a day. SoftBank still held about 14.8% at the end of 2025, with Bertermann at 12.7% and Koç at 9.1%. The quiet lesson sits at the start. A billion-euro company began as someone trying to do a small favour for a relative and noticing how badly the market treated her. If you want a big idea, start where an ordinary person you love got treated unfairly. ## Taxfix (Germany) URL: https://startup.talk/germany/taxfix/ Founders: Mathis Büchi, Lino Teuteberg Founded: 2016; Based in: Berlin, Berlin; Sector: Fintech / consumer tax filing app Most founders raise money to survive their first year. Mathis Büchi and Lino Teuteberg had already done something rarer: they had built a profitable company without outside capital. That company was Smallpdf, a humble tool for merging and compressing PDF files, and it taught two Swiss friends how little a good product needs to be flashy if it removes a daily irritation. Then they moved to Germany and noticed another irritation, a far bigger one. Millions of people were owed money by the tax authorities and never claimed it, simply because the paperwork felt like a threat. Büchi later talked about seeing people lose money by avoiding their returns. The company's own account is that the two had years of friendship and shared projects behind them before deciding to take on a topic everyone dreads. They founded Taxfix in Berlin in 2016. The early product label, as one German outlet called it in 2016, was a friendly tax bot. The bet was simple: do not ask people for tax vocabulary, ask them plain questions and do the translating behind the scenes. For a country where tax software had long been built for accountants, that was a different tone. What made investors lean in was the size of the unfiled pile. Index Ventures led a USD 65M Series C in 2020 and Neil Rimer joined the board. In September 2021 Büchi stepped back from the chief executive role to become chairman, and Martin Ott took over while Teuteberg kept shaping the product. The next year looked like the dream. In April 2022 Taxfix raised USD 220M at a valuation above USD 1bn, joining Berlin's unicorn list. But the timing was brutal. As funding dried up, the company cut 120 jobs, about one fifth of staff, in May 2023, and headcount fell from over 500 to just over 300 across several rounds. Still, the app kept growing in a seasonal business that peaks every spring. A third-party estimate puts 2024 revenue near USD 56.6M, though that figure is not company-confirmed. The lesson is that the unglamorous edge counts. The founders who made PDFs simple earned the right to make taxes simple, and they did it with a reputation for profitable thinking. When money flooded in, the discipline they had learned from bootstrapping was the thing the company had to relearn. ## Parloa (Germany) URL: https://startup.talk/germany/parloa/ Founders: Malte Kosub, Stefan Ostwald Founded: 2018; Based in: Berlin, Berlin; Sector: Enterprise AI / conversational customer service agents Before Parloa was a USD 3bn company, it was a consulting shop that built voice apps for Alexa. In 2016 Malte Kosub and Stefan Ostwald started Future of Voice, which Kosub says was one of the first companies in Germany to build Amazon Alexa applications. Nobody was sure yet that talking to machines would be more than a party trick. Kosub came from Hamburg, and studied business administration at the University of Hamburg until 2016. Ostwald shared his obsession with voice technology. They had no giant research lab behind them, only timing: speech-to-text was improving fast, and brands wanted to be on the new devices. Consulting paid the bills. It also showed them something. Working with large companies, the pair noticed a gap in serious technology for customer contact centres, the place where a company's patience is tested a thousand times a day. One profile says they were building software on the side while consulting kept the lights on. In 2018 they founded Parloa. Their choice was to be voice-first and AI-ready from the beginning, while most competitors had begun with chat. Kosub has said he was convinced by early speech breakthroughs that talking is a more natural interface than typing. It was a bet that looked early, then looked lucky. The hardest thing, he has said, was not the algorithms but people. Finding hires who combine skills and cultural fit can be challenging, a quiet admission from a founder whose product is supposed to replace queues of agents. The money followed the product. A USD 66M Series B led by Altimeter arrived in April 2024. In May 2025 a USD 120M Series C valued the company at USD 1bn. Then, in January 2026, General Catalyst led a USD 350M round at USD 3bn, tripling the valuation in about eight months. The company now has offices in Berlin, Munich, New York and London, with a reported headcount in the hundreds. The lesson is easy to miss. Parloa's edge was built while it was a service business, learning customer problems at close range and writing software only once the gap was clear. Consulting is not a detour, it can be the research phase. ## Isar Aerospace (Germany) URL: https://startup.talk/germany/isar-aerospace/ Founders: Daniel Metzler, Josef Peter Fleischmann, Markus Brandl Founded: 2018; Based in: Ottobrunn (Munich), Bavaria; Sector: Space launch / small satellite launch vehicles On 30 March 2025, Spectrum lifted off from the Andoya Spaceport in Norway. Thirty seconds later it was over. A vent valve had opened when it should not have, the rocket lost attitude control as it began a roll, and the flight was terminated. After seven years and hundreds of millions of euros, the first attempt lasted half a minute. To understand why Isar's people kept going, you go back to a student club. Daniel Metzler, Josef Fleischmann and Markus Brandl were active in WARR, the rocketry group at the Technical University of Munich. WARR students had already built their own engines and a suborbital rocket. Rockets were not a business idea for these three. They were something you built on weekends, with classmates, and then did again better. In 2018 they founded Isar Aerospace near Munich, naming it for the river that runs through the city. The first money was modest and came from believers: Bulent Altan, a former SpaceX leader, along with the Viessmann family's venture arm and UVC Partners. Altan's early nod mattered, because it told other investors that somebody who had seen a space start-up from the inside thought this one was real. Then the cheques got large. In April 2020 Earlybird and Airbus Ventures led a USD 17M round. That December Lakestar led EUR 75M. Porsche SE joined in 2021. In March 2023 a USD 165M Series C came, followed by USD 70M from the NATO Innovation Fund, and the company passed EUR 400M raised. It leased Andoya for 20 years, won a EUR 10M EU prize in 2022 and became the first private company CNES picked to launch from French Guiana. All of it rode on a rocket that had to work. After the 2025 failure the team found the cause and rebuilt. The second launch, called Onward and Upward, was delayed by a boat in the safety zone, a leaking pressure vessel and high winds. On 5 September 2026 it reached orbit and released five CubeSats and an experiment. There was a cost beyond money. Days after the triumph, the story turned: Fleischmann stepped down as CTO to start his own firm, Bavaria Instruments, keeping his Isar stake. Co-founders do not always want the same next chapter even when they share the same rocket. The lesson for founders is about the first thirty seconds. A hardware company can fail in public, on a live stream, and still be the same company a year later, if the reason it began was never a business plan but a club of friends who liked building things. ## Quantum Systems (Germany) URL: https://startup.talk/germany/quantum-systems/ Founders: Florian Seibel Founded: 2015; Based in: Gilching (Munich area), Bavaria; Sector: Defense / dual-use drones (eVTOL fixed-wing UAS) Florian Seibel's first customers were not soldiers. They were people who wanted to look at fields and farms from above. The former Bundeswehr helicopter pilot, who flew military aircraft and worked as a researcher at the University of the Bundeswehr in Munich, had set out in 2015 to build small drones for civilian use. The seed was a patent. Seibel and a colleague had patented a drone prototype while he was working on his doctorate. Then he did something many academics fear: he left the PhD behind and founded Quantum Systems in January 2015. His service background gave him credibility with the aircraft but also an itch about how unmanned systems get built and bought. The early years were modest. The first Trinity prototype flew in 2016, the Tron eVTOL fixed-wing system followed in 2017, and production of Trinity began in 2018. In a 2018 interview with DroneBelow, Seibel offered a founder's mantra: what can go wrong will go wrong. For years it was a plain engineering company, with civilian customers and a modest base of backers that eventually included HV Capital, Project A and Thiel Capital. The turn came from outside. After Russia's 2022 invasion of Ukraine, European armies suddenly wanted small, reliable, fast-to-field drones. The company moved from agriculture toward defense. In October 2022 the first Vector went to the Dutch Ministry of Defence. In 2024 Australia selected Vector under its Land 129 programme. The cost was a change of identity. A company that began with farmers and surveyors was now a defense supplier, which meant political scrutiny, long procurement cycles and a different kind of customer. It raised EUR 63.6M in 2023 and EUR 160M in May 2025, becoming Europe's first dual-use unicorn. In December 2025 Germany approved a framework agreement for 747 drones, and in July 2026 the company raised USD 1.2bn at an USD 8bn valuation. For first-time founders, the lesson is about optionality. Seibel did not start with a defense strategy. He built a technology that could fly vertically and cruise efficiently, kept the company alive through years of small markets, and was ready when the world changed its mind about what drones were for. ## Staffbase (Germany) URL: https://startup.talk/germany/staffbase/ Founders: Martin Böhringer, Lutz Gerlach, Frank Wolf Founded: 2014; Based in: Chemnitz, Saxony (Sachsen); Sector: Employee communications software (B2B SaaS) From his office window in Chemnitz, Martin Böhringer can see a forklift rental and a car repair shop. He told Business Punk in 2024 that this is exactly why he likes it. Berlin, he said, is the startup mainstream, and moving there would have been too easy. He was not new to the game when he made that choice. What few people mention is that Staffbase was not his first company. Böhringer and his friend Lutz Gerlach had already built another startup, Hojoki, before they started again. They met at TU Chemnitz, where Böhringer studied business informatics and Gerlach business administration. Both went on to finish doctorates there. The idea, Böhringer has said, practically took shape on campus. They added a third founder, Frank Wolf, who had worked at T-Systems in Dresden. In 2014 the three launched what was, in the simplest description, an employee app. The first six months were spent with a handful of chosen customers, some with five employees, some with half a million. That range tells you the problem they were circling. Most workplace software assumes a desk and a company email. Nurses, drivers, shop staff and factory workers have neither. A phone in the pocket was the only channel left, and big companies badly wanted one. The cost of staying in Saxony was explaining it. Investors asked, again and again, why a software company lived in Chemnitz. Böhringer has said the question faded only once the large funding rounds arrived. Frank Wolf eventually moved to New York, about three years in, and opened a second office, so the company could sit near its biggest customers while its engineers stayed home. The rounds did arrive. General Atlantic led a $145M Series D in March 2021, then a $115M Series E a year later that valued Staffbase at $1.1bn. In May 2022 the founders were named Saxony's entrepreneurs of the year. Total money raised stands near $300M. Now Böhringer talks about an IPO, but in an unusual way. In 2025 he described it not as an exit but as an extended financing round, a way to pay back early investors while the company keeps growing. That is the quiet thread through the story. The two men who had one startup behind them chose the same town, the same friends and the same campus for the second attempt. For a first-time founder, the lesson is not that location does not matter. It is that you can pick the place you understand best and make everyone else come to you. ## Tier Mobility (Germany) URL: https://startup.talk/germany/tier-mobility/ Founders: Lawrence Leuschner, Matthias Laug, Julian Blessin Founded: 2018; Based in: Berlin, Berlin; Sector: Micromobility (e-scooter and e-bike sharing) Before there were scooters, there were hammocks. Lawrence Leuschner's father imported goods from Latin America, and some of them came back defective. Most were thrown away. The teenage Leuschner struck a deal to take the damaged ones for nothing and sold the seconds at flea markets near his hometown in the Taunus. He has described the next step as a relief. Online auction sites such as Alando and eBay meant he no longer had to stand at a market early on Sunday mornings. He could sell from home. A couple of years later, just as he turned adult, he and friends started Trade-a-Game, a marketplace for used video games. He left his studies for it. Angels, among them Oliver Samwer, backed the team, and in 2006 they moved to Berlin. Later, after watching An Inconvenient Truth, he pushed the business into used electronics. It became Rebuy, now a market leader in Europe. In a podcast he put it simply: a person always needs a purpose. Then, after more than ten years as CEO, he stopped. In 2017 he took a sabbatical, went surfing, and stepped back from daily work at Rebuy while remaining a shareholder. The break grew into an 18-month trip through Asia and California, including eight months driving a VW Bulli from Patagonia to Mexico. In San Diego in April 2018 he saw people riding e-scooters and enjoying themselves. He thought it could become big. He believed Europe's dense cities were even better suited to it. With Matthias Laug, an early Lieferando figure, and Julian Blessin, who came from Bosch's scooter-sharing venture Coup, he started Tier. Only three months separated the business plan from the first test market, funded by a rushed seed round. The founders made one unusual choice. Instead of the American model of freelancers charging scooters at home, Tier ran its own workshops. Leuschner admitted the first scooters did not last long. By June 2019 the company reported two million rides. The boom had a price. In August 2022 about 180 people lost their jobs and expansion stopped as the company turned toward profitability. In 2023 it sold the Spin business to Bird for $19M. In March 2024 it merged with Dott, backed by a €60M equity injection, and by October the Tier name had gone from the app. The combined group reported €173.3M revenue for 2025. Leuschner's story is a long way from the hammock stall, yet it follows the same rule. He kept finding value in what others discarded, whether a returned hammock or a second-hand game console, and acted on it fast. ## Solaris (Germany) URL: https://startup.talk/germany/solaris/ Founders: Marko Wenthin, Andreas Bittner Founded: 2015; Based in: Berlin, Berlin; Sector: Banking-as-a-service (fintech infrastructure) In 2025, one of the men who founded Solaris went to a Berlin court to contest the rescue that was keeping the company alive. Andreas Bittner had left the firm five years earlier, when his five-year contract ended on 30 June 2020. Now he was challenging the capital increase through which Japan's SBI Group took control. It is an odd ending for a story that began with a simple bet. In 2015 Bittner and Marko Wenthin set up a banking platform inside Finleap, a Berlin company builder. They did not want a pretty app. They wanted the licence behind it. In March 2016 the bank launched with a full German licence, and an idea that was almost boring in its logic: let other companies rent the plumbing. The pitch worked. Mastercard signed a partnership within months. In February 2017 the company raised €26.3M from investors including UniCredit, Arvato and SBI, and by March it reported over twenty customers and the ability to serve six European countries. Roland Folz arrived as CEO in 2017. A €56.6M round followed in 2018, then €60M led by HV Capital in 2020. But a bank is not a startup. It answers to a regulator. BaFin ordered a special audit in January 2022 after finding organisational weaknesses in an earlier review, and Folz publicly welcomed it. Late that year the regulator demanded corrective steps on risk management and money-laundering prevention, and fined the company €6.5M for filing suspicious-activity reports late. The firm took a new name, Solaris SE, in November 2022, the year it reported about $130.9M in revenue. The cost showed in 2024. Reports described a heavy loss, the closing of the e-money unit, and the loss of roughly a third of the workforce. Then SBI, which had been an early investor in the 2017 round, led a €140M Series G in 2025 and became the majority owner. The dispute with Bittner shows the awkward side of that kind of rescue. Early shareholders get diluted. The people who built the thing sit outside the room. For a founder of infrastructure, the lesson is plain. A licence is a moat, but it also owns you. Solaris proved banking could be built like software, and then spent years finding out that regulators do not accept software's habit of moving first and fixing later. ## Forto (Germany) URL: https://startup.talk/germany/forto/ Founders: Michael Wax, Erik Muttersbach, Ferry Heilemann, Fabian Heilemann Founded: 2016; Based in: Berlin, Berlin; Sector: Digital freight forwarding / logistics Shipping a container from Asia to Europe has long meant phone calls, PDFs and faxed customs forms. In 2016 a handful of Berlin founders decided that this was a software problem. They called their company FreightHub. The founding team mixed operators and investors. Michael Wax became chief executive. Erik Muttersbach and Ferry Heilemann were co-founders, and Fabian Heilemann, also a partner at the venture firm Earlybird, took the chairman's seat. Wax studied at Technische Universität München, and Ferry Heilemann is listed as a WHU graduate. Beyond that, little is public about how they met or what they gave up. The first product was modest. It sold spot-priced full container loads to small and mid-sized businesses, the customers the big forwarders treated as an afterthought. A shipper could get a price online and book it. By late 2017 Northzone had led a $20M Series A. Progress came in steps. A $30M Series B arrived in May 2019, with the shipping giant Maersk's venture arm among the backers. That is a quiet detail worth noticing. The industry the startup wanted to change put its own money in early, which gave Forto access to knowledge and relationships that a pure software company would struggle to buy. By then the name had shifted from FreightHub to Forto. Then came the pandemic freight boom. In June 2021 SoftBank Vision Fund 2 led a $240M round that valued the company at $1.2bn. About 2,500 customers were using its digital tools. Fabian Heilemann credited industry partnerships, constant innovation and a culture centred on efficiency. A second round followed in March 2022, $250M led by Disruptive at a reported $2.1bn valuation, almost double in eight months. What happened after is the harder chapter. The freight market that lifted every forwarder in 2021 fell hard, and valuations built in that moment depend on the next cycle. Total funding is variously reported between roughly $360M and $600M, and revenue has largely stayed private. For a founder, the useful point is where Forto aimed. It did not try to replace the shipping line. It took the paperwork around the box and made that fast and clear, and let the incumbents become partners instead of enemies. ## Enpal (Germany) URL: https://startup.talk/germany/enpal/ Founders: Mario Kohle, Viktor Wingert, Jochen Ziervogel Founded: 2017; Based in: Berlin, Berlin; Sector: Residential solar, batteries and heat pumps (energy-as-a-service) Mario Kohle was born in 1984 in the GDR and grew up in Strausberg, a town east of Berlin. His grandfather, Horst Kohle, had played for the East German national football team. Few people who see Kohle in a boardroom think of that. The family name carried one kind of fame. Kohle chose another. He studied business administration at WHU, the Otto Beisheim School of Management, finishing in 2008. As a student he became fascinated by e-commerce. In 2008, with Robin Behlau, he started Käuferportal, a platform that connected homeowners with local tradespeople for building and renovation jobs. It was later renamed Aroundhome. A student became a chief executive before most of his classmates had settled into their first jobs. In 2016 he sold it to ProSiebenSat.1 and General Atlantic, reportedly for €100M. Most founders would have taken a long holiday. Kohle was thinking about something else. He has described a Greta Thunberg moment in 2015, when climate change stopped being an abstract topic for him. What almost nobody expected was the next move. A software founder walking into solar roofs, a business of cables, steel and installers. With Viktor Wingert and Jochen Ziervogel he founded Enpal in 2017, selling rooftop systems to homeowners on a lease instead of a large purchase. The industry doubted that a startup could do the hard physical part. Early backers such as Lukasz Gadowski and Alexander Samwer bet that it could. Quality was the weak point. A solar company lives or dies on installation. So in 2021 Enpal opened its own training centre for installers and electricians, a costly choice, and it worked. Revenue went from €56M in 2020 to €115M in 2021 and almost €415M in 2022, the first profitable year. By the end of 2023 the company had put about 60,000 systems on German roofs, and staff numbers had climbed from 324 in 2020 to some 2,500 two years later. SoftBank and Leonardo DiCaprio's fund joined the investor list, and the company passed one billion euros of revenue in 2025. Kohle has said he does not chase valuations, and the record bears him out. He built a trade school before he built a brand. For first-time founders, there is a tidy lesson: the unglamorous layer under the idea, the people who actually do the work, is often where the moat sits. ## 1Komma5° (Germany) URL: https://startup.talk/germany/1komma5/ Founders: Philipp Schröder Founded: 2021; Based in: Hamburg, Hamburg; Sector: Energy / residential solar, batteries and AI energy management Before there was a unicorn, there was a farm. Philipp Schröder, the man who now runs one of Europe's most richly valued energy startups, reportedly grew up on an organic farm in northern Germany. It is a detail that appears in only one author profile, and he rarely dwells on it. But it helps explain a life spent around people who grow their own energy, in the literal sense. The career that followed was a string of bets on the sun. In 2008 he founded a small energy company, Nycon Energy. In 2011 he joined the solar developer JUWI. Then came sonnen, the battery maker, and after that a stint as head of Tesla in Germany, selling home batteries and cars to people who had never thought of their cellar as a power station. In 2015 he went back to sonnen. In 2018 he started CAPinside, a software company in an entirely different field. What nobody on the outside saw was the pattern. Each job had given him one piece of the same puzzle: panels from one employer, batteries from another, a consumer brand from a third. None of them put the pieces together for the household that actually pays. In July 2021 he started 1Komma5° in Hamburg to do exactly that. The pitch was a single package: solar panels, a battery, a heat pump, a wallbox, and a software layer that decides when to store power and when to sell it back to the grid. The installer was not an afterthought. 1Komma5° bought up local installation businesses and put them under one brand, because he had seen at Tesla and sonnen how often a good product died on a bad roof. Investors moved quickly. A €200m Series A followed in 2022, and Klarna co-founder Niklas Adalberth invested through his Norrsken Foundation. A pre-IPO round reportedly followed in 2025, valuing the company at about $1.2bn. Then the market turned cold. In October 2026 the company raised fresh money at a $1.35bn valuation, but the planned stock market listing stayed on ice. It is the quiet price of a capital-hungry model: you can grow fast by buying installers, but you cannot print patience. So the farm boy's company sits in an awkward adolescence. It has the valuation of a champion and the balance sheet questions of a builder. There is a lesson for first-time founders in how long Schröder waited. He did not start 1Komma5° at 25 with an idea. He started it after more than a decade of selling, installing and arguing with customers about the same product. The idea looked obvious only because he had been standing inside the problem for so long. Experience was his unfair advantage, and he spent thirteen years collecting it. ## Choco (Germany) URL: https://startup.talk/germany/choco/ Founders: Daniel Khachab, Julian Hammer, Rogério da Silva Yokomizo Founded: 2018; Based in: Berlin, Berlin; Sector: Food supply chain / B2B ordering software and payments Everyone assumes the name means chocolate. It does not. Daniel Khachab has had to explain this for years: Choco is named after the Chocó, a wet, wildly biodiverse strip of Colombia's Pacific coast, a place he had come to know while living in Colombia. He told the OMR podcast he liked it for what it stood for, a huge range of things and less waste. That is a surprisingly soft origin for a company that now moves orders between restaurants and wholesalers. Khachab came out of Rocket Internet, the Berlin startup factory, where speed is a religion. In 2018 he started Choco with Julian Hammer and a third co-founder, named in most accounts as Rogério da Silva Yokomizo. The problem they picked was unglamorous. Chefs ordered ingredients by phone, scribbled notes and WhatsApp voice messages, usually late at night. Wholesalers then typed it all up by hand. Choco turned that mess into something that feels like a messenger chat. Khachab compared apps to kitchen tools: you reach for them and they just work. What almost nobody expected was how little marketing it took. For the first two years there was no marketing department and not even a website. Wholesalers told their restaurant customers, who told other restaurants. The founders did, however, pay a significant sum for the domain name early on, which Khachab mentioned with some amusement. Then came 2020. Restaurants shut from one day to the next, and the customer base that Choco existed for was suddenly gone. Khachab has described the shock plainly. The company pivoted its energy to helping wholesalers keep selling food, and the business model matured: Choco earns on payment services, letting restaurants pay invoices later while wholesalers get their money immediately. Investors saw the recovery. A Series B of about $100m came in 2021, and in April 2022 a further $111m round valued Choco at about $1.2bn. By January 2022 it counted more than 15,000 restaurants and 10,000 wholesalers, and the United States had become its biggest market. The boom ended fast. In February 2023 the company laid off staff as it streamlined, a reminder that hiring from 350 towards 800 in one year is a bet on a market that does not wait. The lesson for founders is in the name. Khachab built a global software company around a place that mattered to him personally, and around a customer who was too busy to learn a new tool. Make the product feel like the thing they already do, and the brand story can stay a little strange. ## Ada Health (Germany) URL: https://startup.talk/germany/ada-health/ Founders: Claire Novorol, Daniel Nathrath, Martin Hirsch Founded: 2011; Based in: Berlin, Berlin; Sector: Digital health / AI symptom assessment A baby was sick, and nobody could say why. Claire Novorol, then a young clinician in clinical genetics, went looking. At Addenbrooke's Hospital in Cambridge she searched the medical literature and scientific databases until she found a rare genetic condition that matched. The baby was diagnosed. That episode, as the story is told, is where the idea began. It stuck with her because of what it exposed. The answer had existed in print all along. It had taken hours of one doctor's evening to find it, and most children with rare diseases do not have a doctor like that. Novorol did not leave medicine for software straight away. She started Doctorpreneurs, a network for doctors interested in building digital health products. Through it she travelled to Berlin and met two people who would become her co-founders: Daniel Nathrath, who became CEO, and Martin Hirsch, who became CTO. In 2011 they founded Ada. The first product was a decision-support tool for doctors working on rare diseases. It was accurate, but it was small. Doctors are a thin market, and the tool needed the very people who already had the least time. What changed things was a pivot. In 2016 the company turned the engine around and pointed it at patients, creating a symptom checker called Assess. Anyone could answer questions and get a list of possible causes. It was tested against rival apps in 2017 and did well, and a 2019 retrospective study found that its top suggestion matched the confirmed diagnosis in 83 of 93 rare disease cases, about 89 percent. The price of going consumer was scrutiny. A tool that tells strangers what might be wrong with them must be certified, audited and defended. Ada runs a quality system to the medical device standard ISO 13485 and its software is CE-certified as a Class IIa medical device in Europe. That is slow, expensive work, and it is the opposite of the move-fast story most apps tell. By 2017 the company had raised about $47m, and by 2023 it employed roughly 300 people. A later, larger Series B of over €105m has been reported. The lesson is a doctor's one. Novorol did not start from a market gap. She started from a single patient, and from the discomfort of knowing that the answer existed somewhere. Founders who begin with one person's problem rarely run out of reasons to keep going. ## Mambu (Germany) URL: https://startup.talk/germany/mambu/ Founders: Eugene Danilkis, Frederik Pfisterer, Sofia Nunes Founded: 2011; Based in: Amsterdam, Netherlands (founded in Berlin, Berlin); Sector: Fintech / cloud banking infrastructure (SaaS) The first customers were not banks. They were microfinance lenders in Mozambique, visited by a group of graduate students who did not yet know they were starting a company. Eugene Danilkis, Frederik Pfisterer, Sofia Nunes, Josh Coe and Joydeep Sengupta were five students in Carnegie Mellon University's Master of Human-Computer Interaction programme. Their capstone was a one-year project. The assignment was about microlending. They did field research with microfinance institutions in Mozambique, watching how small lenders tracked loans in notebooks and clumsy software. They saw a design problem. The lenders served poor customers, but the tools they used were built for large banks, and were slow and expensive to change. A product with a better interface and a more flexible core could change that. When the year ended, most people go to jobs. Three of them kept going. Danilkis managed the project, Pfisterer led the technology and Nunes brought the psychology, and their thesis work became the foundation of a platform. In 2011 they set up Mambu in Berlin. The first customers were microfinance providers in Latin America and Africa. Early backers included Commerzbank and the Berlin fund Point Nine. The turn came when the same flexible core turned out to suit banks and fintechs in rich countries, who wanted to launch loans and accounts without rebuilding their systems. A 2019 round of about €30m led by Bessemer marked the shift. In January 2021 a €110m round valued the company at $1.7bn, and in December 2021 EQT led a roughly €235m round at about €4.9bn. By 2022 it reported $165m of contracted annual recurring revenue. The cost showed up in June 2023, when Danilkis stepped down as CEO. He stayed as a shareholder and board member, and the CTO, Fernando Zandona, took over, becoming permanent CEO in August. Founder transitions are rarely clean, and companies valued at that level are rarely allowed to stay small. Mambu today is headquartered in Amsterdam, a long way from a Mozambican lender's notebook. The lesson is in the sequence. The founders did not start with a market size. They started with an observation made in person, in the places where the pain was sharpest, and they were willing to stay with the problem after the course ended. ## Scalable Capital (Germany) URL: https://startup.talk/germany/scalable-capital/ Founders: Erik Podzuweit, Florian Prucker, Adam French, Stefan Mittnik Founded: 2014; Based in: Munich, Bavaria; Sector: Fintech / digital wealth management, brokerage and banking In 2014 almost nobody in Germany knew what a fintech was. Erik Podzuweit has said the first job was just explaining what Scalable Capital did. It managed clients' money using software, and no independent German startup had done that before. Podzuweit had studied business administration and economics in Kiel and at Warwick, and then worked at Goldman Sachs, where his future co-founder Florian Prucker also worked. Prucker had studied business and electronics at the Technical University of Munich. Wealth management, as they saw it from the inside, was a product kept for the rich. They wanted to change that exclusivity. In December 2014 they started the company in Munich with Adam French and Stefan Mittnik, a professor of financial econometrics at LMU. Munich was a deliberate choice: engineers from the city's tech firms, a university down the road and, in Podzuweit's words, strong infrastructure for founders. The first product, a robo-advisor, launched in 2016. The hard part was trust. Handing savings to an app run by people you have never met is a strange ask, and Germans are famously wary of shares. BlackRock led a €30m round in 2017 and ING Germany signed on as a partner, which lent the young company a borrowed credibility. What changed the shape of the business was a decision to add a broker in 2020, where clients could pick their own ETFs. Growth followed. A Series E of about €150m led by Tencent in 2021 made Scalable a unicorn, and in the same year it bought the ETF portal justETF. It has not been smooth. In 2020 a former service provider suffered a data breach that affected more than 33,000 people in Germany and the UK, leading to court cases over damages. In 2025 consumer advocates warned it over how it described interest on cash, and the Financial Times reported access delays during April's market turmoil. Still, in 2025 it received a full banking licence from the European Central Bank. By August 2026 it reported more than a million clients and over €60bn in assets. The lesson for founders is about patience with a boring promise. Podzuweit and Prucker did not invent investing. They spent years explaining a simple idea until the market caught up, and they were still there when it did. ## Pitch (Germany) URL: https://startup.talk/germany/pitch/ Founders: Christian Reber, Adam Renklint Founded: 2018 (company launched 2018); Based in: Berlin, Berlin (Bundesland); Sector: Presentation / collaboration software (SaaS) In September 2019, a man who had sold a company to Microsoft for a reported $100 million to $200 million wrote on Twitter that he was still sad. Microsoft, he said, had wanted to shut his old app down. The man was Christian Reber, and the app was Wunderlist. Reber has told a podcast that he 'accidentally started a company' when he moved to Berlin at 18. That accident grew into Wunderlist, a to-do list app that became a fixture on many phones. In 2015 Microsoft bought it. It looked like the clean ending every founder dreams of. For Reber it was something more tangled. He has since said in interviews that he questioned whether the sale was a mistake, and he watched the product he loved get folded into a larger suite and then retired. Most founders would have taken the money and rested. Reber went looking for the next thing that Microsoft owned in people's heads. The answer was the slide deck. Every company, school and startup made them, nearly everyone hated making them, and the tool of choice was a desktop program designed decades earlier. His pitch to investors was blunt: build the 'YouTube for presentations', with sharing, comments and version history as native features rather than afterthoughts. What gave the idea weight was the people behind it. Wunderlist had taught him how a Berlin team could ship software that felt designed rather than assembled, and investors remembered. BlueYard Capital backed the first round in 2018. A $30 million round followed in October 2019, with Thrive Capital and Index Ventures involved, and Slack among the names. By then the company had been in existence barely a year and a half. The real test came in May 2021. Pitch raised $85 million from Lakestar and Tiger Global at a valuation reported at $600 million, even though only a couple of thousand people were paying $10 a month for the Pro tier. About 125,000 workspaces were using the product, with Intercom, Superhuman and Notion among the named customers. Reber leaned on his nationality, telling TechCrunch that 'as a German business, we have a special relationship with data privacy.' Then came the part nobody on a funding-announcement day talks about. Pitch was selling into a market where the incumbents are free or bundled, and a wave of AI tools after 2023 changed what people expected from slides. Third-party estimates put revenue at roughly $9.4 million in 2024, a modest figure against the ambition of 2021. Leadership changed too, with Adam Renklint listed as CEO in recent profiles. The lesson inside Reber's story is a quiet one. He did not escape the grief of Wunderlist by winning once. He went back to the same hard problem, productivity software that people love, and tried to own it this time. Whether Pitch gets there is still open. But a founder who treats a sale as a wound, not a trophy, tends to build differently the second time. ## Aleph Alpha (Germany) URL: https://startup.talk/germany/aleph-alpha/ Founders: Jonas Andrulis, Samuel Weinbach Founded: 2019 (27 February 2019); Based in: Heidelberg, Baden-Württemberg; Sector: Generative AI / large language models Jonas Andrulis has said that being called 'Germany's answer to OpenAI' has cost him the odd sleepless night. It is an honest line from a man whose company was handed a national burden before it had much revenue to carry it. Andrulis trained as an industrial engineer at the Karlsruhe Institute of Technology, a short drive from Heidelberg. Then he went to Silicon Valley and spent years at Apple, where one profile places him managing AI research in a special projects group from 2016. A German profile suggests Apple is where his world-changing ambition took shape. He returned in 2019 and registered Aleph Alpha in Heidelberg with Samuel Weinbach on 27 February. 'Let's change the world!' is a phrase he reaches for in keynotes. He also says, 'I believe we have a chance that's worth fighting for.' The pitch was specific. European governments and companies were nervous about sending their data to American models, and nobody on the continent was building a serious alternative. Aleph Alpha aimed to make transparent models, Luminous among them, aimed at businesses and administrations rather than consumers. A €5.3 million seed came in 2020 and €23 million followed in 2021. In 2021 the team added models that read both text and images. The moment that made the company a national story came in November 2023, when a round was announced at $500 million. It was read as a sign that Europe could field a rival. What almost nobody noticed at first was how the number was built. Trade press later reported that only about €110 million was genuine equity, with around €300 million going to research funding for a subsidiary and €60 million in order commitments. The books were harder still. Aleph Alpha missed an internal 2023 sales target of $6 million. Its statutory accounts for the year showed turnover under €1 million and a loss of €18.9 million. Meanwhile US labs were spending at a scale no European startup could match. The strategy bent. The company moved away from racing on the largest models and toward helping clients run AI in-house. In January 2026 Andrulis ceased to be a managing director, and the company moved to the supervisory board. In April 2026 Cohere announced it would acquire Aleph Alpha, with a definitive merger agreement in September, and the Heidelberg site is to stay as a research centre. In October 2026 the company still released Kolibri, an open-weight model under the Apache 2.0 licence. For first-time founders, the lesson is uncomfortable. A big headline number can become a promise you have to live up to in public. Andrulis got the attention, the sleepless nights and a chance worth fighting for, and he learned that carrying a country's hopes is not the same thing as having a business that pays for them. ## Lilium (Germany) URL: https://startup.talk/germany/lilium/ Founders: Daniel Wiegand, Sebastian Born, Patrick Nathen, Matthias Meiner Founded: 2015; Based in: Weßling/Gauting near Munich, Bavaria; Sector: Electric vertical take-off aircraft (eVTOL), aerospace Daniel Wiegand has been flying gliders since he was 14. That is the line that shows up in one profile of the Lilium co-founder, and it explains a lot. Before he was an aircraft entrepreneur he was a teenager who knew exactly what it felt like to hang in the air with no engine at all. He went on to study aerospace engineering at the Technical University of Munich. One profile says the idea for a flying taxi arrived on a semester abroad in Glasgow, after he found a YouTube video of the American V-22 Osprey. It is the kind of origin story founders tell, and it may have grown in the telling. What is documented is that in 2015 he, Sebastian Born, Patrick Nathen and Matthias Meiner, all engineers or doctoral students at TUM, started Lilium. The name is a nod to Otto Lilienthal, the German glider pioneer. The early support was unglamorous. The European Space Agency's Bavarian incubation centre gave them a start. Then the money got serious. In September 2017 a round of more than $90 million came in, with Atomico, Tencent, LGT and Frank Thelen's Freigeist all involved. A few months earlier, in April 2017, a two-seat unmanned Lilium Eagle had flown at a small airfield near Munich. In May 2019 the five-seat Phoenix followed. The pitch was irresistible. A jet with dozens of electric fans in the wings, quiet enough for cities, fast enough for regional trips, built in Bavaria. In March 2020 more than $240 million arrived, bringing total investment above $375 million. In September 2021 the company merged with the SPAC Qell and began trading on Nasdaq at a value of about $2.4 billion, though it had no meaningful revenue and a 2020 loss of €188 million. The price of that ambition showed up slowly. Wiegand was replaced as CEO in 2022, when Klaus Roewe took over. Certification of a new kind of aircraft is long and expensive, and the cash kept leaving. In October 2024 the two operating subsidiaries filed for insolvency, and Nasdaq delisted the company on 7 November. A rescue seemed to arrive in December, when an investor group led by Mobile Uplift agreed to take over the business and keep around 775 jobs. The promised money, more than €200 million, never came. In late February 2025 Lilium filed for insolvency again, ending operations for over 700 people. The founder lesson is a hard one. A prototype that flies proves physics. It does not prove a business. Wiegand built the first and could not afford the second, and the glider pilot's gift, patience in thin air, runs out when the funding does. ## Wefox (Germany) URL: https://startup.talk/germany/wefox/ Founders: Julian Teicke, Dario Fazlic, Fabian Wesemann Founded: 2015 (as FinanceFox); Based in: Berlin, Berlin (Bundesland); group seat Zug, Switzerland; Sector: Insurtech / digital insurance In 2009, a 22-year-old Berliner was selling chocolate. Julian Teicke, born in Berlin in October 1986, had helped start the UK branch of Chocri, a Berlin company where customers designed their own bars. Nothing about it suggested insurance. It suggested a young man who liked building a market from nothing. He had studied at the University of St. Gallen, and after Chocri he joined Groupon, where he headed European sales. In 2010 he moved to Switzerland to join DeinDeal, an e-commerce deals platform, and became its COO. Ringier bought it in 2015. He won the Swiss e-commerce award in 2013 and 2014. By then he had learned the pattern: find a clumsy industry, put a screen in front of it, and grow faster than the people who have been doing it for decades. The clumsy industry he picked was insurance. In 2015 he founded FinanceFox with Dario Fazlic and Fabian Wesemann, an app that let customers see all their policies in one place. Within two years it reported more than 100,000 customers. Once it won permission to sell insurance products it rebranded as wefox in 2017, and in 2018 it launched its own carrier under the One brand. What almost nobody noticed in the glow of that growth was how much it cost to be both a broker and an insurer. Wefox passed a €1 billion valuation in 2019. In June 2021 it raised $650 million at $3 billion, led by Target Global. A $400 million round in 2022 pushed it to $4.5 billion. It bought insurers and brokers across Europe, and critics said the acquired firms never fully adopted its technology. Then the arithmetic turned. The group reportedly lost €125 million in 2022 and €141 million in 2023. Teicke stepped down as CEO in March 2024 after nine years. A bid from Mubadala, valuing the group at around €550 million, was turned down. A 2024 round valued wefox at about €500 million, roughly 90% below the peak. It withdrew from the German market, sold Assona to Ecclesia and, in December 2024, its Liechtenstein insurer. In July 2025 it raised €151 million after restructuring, and by March 2026 it was preparing a €15 million convertible emergency loan. Teicke later took a non-executive role and sits on the board of the Jewish Museum Berlin. His story carries a warning for a generation of founders who learned from chocolate and coupons that growth is the product. In insurance, growth is a liability until the losses are priced. ## GetYourGuide (Germany) URL: https://startup.talk/germany/getyourguide/ Founders: Johannes Reck, Tao Tao, Tobias Rein, Martin Sieber, Pascal Mathis Founded: 2009 (Zurich); HQ moved to Berlin 2012; Based in: Berlin, Berlin (Bundesland); originally Zurich; Sector: Online travel experiences marketplace For the first two years of GetYourGuide, there were almost no customers. Johannes Reck has said the site took three to five bookings, and that three of those came from his mother. He has told the story with a laugh. At the time it can hardly have been funny. The idea began in 2009 in Beijing. Reck, then a student at ETH Zurich, arrived a day early for a trip. His classmate Tao Tao showed him around, and the two saw how hard it was for a traveller to find a good local guide. They began building a peer-to-peer platform to connect tourists with amateur guides. They also considered a couch-surfing model and dropped it, Reck has said, because they doubted travellers would sleep in strangers' homes. The company was incorporated in Zurich that year with money from friends and family. Venture capital did not help. Reck has said investors turned them down more than a hundred times in the early days. For four years the team stayed alive without a single VC cheque. They had met at university, and they kept working on the product alongside their studies. The turn was not glamorous. The amateur-guide idea was not working, so they changed what they sold. They built reservation systems for professional tour providers and put the providers' inventory on one booking site. That gave them product-market fit. In 2012 the headquarters moved to Berlin, and a Series A of $14 million came in 2013. Further rounds followed: $25 million in 2014, $50 million in 2015, $75 million in 2017. The real leap came in 2019, when SoftBank's Vision Fund led a $484 million round that made GetYourGuide a unicorn, among the largest rounds a European startup had raised. By then it had passed 25 million tickets sold. In 2018 it began selling tours under its own brand, and in 2021 it began operating in the United States. There were hard years. Italy's competition authority fined the company in April 2025 over Colosseum tickets, accusing it with others of profiting from resales and making face-value tickets harder to get. In May 2023 it raised $194 million at a $2 billion valuation. In October 2025 it said it had become profitable for the first time, with roughly €1 billion in sales over the previous year. The quiet lesson is in the first two years. Reck had a mother who bought tickets and a classmate who showed him a city. Most great marketplaces begin with a founder who can survive a very small number of customers and still change the product. ## Babbel (Germany) URL: https://startup.talk/germany/babbel/ Founders: Markus Witte, Lorenz Heine, Thomas Holl, Toine Diepstraten Founded: 2007; Based in: Berlin, Berlin; Sector: Consumer edtech / language learning subscriptions The company that now sells language lessons to millions was first registered under a name almost nobody remembers: Lesson Nine GmbH. Four men who had worked together at a Berlin software firm, Markus Witte, Lorenz Heine, Thomas Holl and Toine Diepstraten, founded it in August 2007, and for a while the name said all there was to say. A lesson, and a number. They did not start with a plan to build a giant. They started with a website. In January 2008 it opened as a free community beta, a place where people could learn a language online in plain, practical steps. Free is a comfortable word for a founder. It brings users quickly and demands nothing back. The decision that mattered came next, and it was a less comfortable one. The founders chose paid content over advertising and over a long free tier. It meant fewer sign-ups and harder conversations. It also meant that every user who stayed was a customer, not a number to sell to advertisers. What almost nobody remembers is how small the early backing was. A case study from the University of British Columbia names Kizoo Technology Ventures and the Berlin state-linked IBB Beteiligungsgesellschaft as the first investors in 2008. These were modest, local cheques, not the glamour money of later years. Later, a round of about $22m led by Scottish Equity Partners, with Reed Elsevier Ventures and Nokia Growth Partners alongside, gave the company room to grow abroad. In 2013 it bought PlaySay, a San Francisco startup, to get a foothold in the United States. Growth brought its own tensions. In 2020 staff in the Berlin office elected a works council. In 2022 the company gave free access codes to Ukrainian refugees, a gesture that suited a firm built on helping people speak to each other. By 2022 revenue stood at about €247m with roughly 1,000 employees. Then came the part founders rarely plan for. Witte, who had run the company as CEO, stepped down in 2019 and Arne Schepker took over. In October 2024 Witte came back as CEO. By mid-2025 Tim Allen held the job and Witte moved to chair the supervisory board. A founder had left, returned, and then stepped up to the sidelines within six years. There is a quiet lesson in the name Lesson Nine. The founders began with something plain and almost forgettable, and let the product, not the branding, earn the right to a bigger name. The decision to charge from day one looks obvious in hindsight. In 2008 it was the whole company. ## Trivago (Germany) URL: https://startup.talk/germany/trivago/ Founders: Rolf Schrömgens, Stephan Stubner, Peter Vinnemeier, Malte Siewert Founded: 2005; Based in: Düsseldorf, North Rhine-Westphalia; Sector: Travel metasearch / hotel price comparison In March 2020 trivago's sales dropped about 95% from a year earlier. Hotels were shut, borders closed, and a company built on clicks from people planning holidays watched the clicks stop. For a founder who had built the thing from nothing, it was the harshest test since the first day. Rolf Schrömgens was born in 1976 and took a management diploma at the HHL Leipzig Graduate School of Management. What is easy to miss is that he was already a founder by the time trivago existed. From 1999 to 2001 he was founder and vice president at ciao.com, a consumer review site. He had learned early that people trust other people's opinions more than ads, and that a website can be a marketplace for trust. In January 2005 he and three partners, Stephan Stubner, Peter Vinnemeier and Malte Siewert, started trivago in Düsseldorf. Stubner resigned as managing director soon after launch, a quiet casualty of the early days. The product compared hotel prices across booking sites and charged them for the clicks it sent. Money came in steps. In 2007 it raised $1.14m from HOWZAT Media. In 2010 Insight Venture Partners bought about a quarter of the company for roughly $52.9m. And in December 2012 Expedia announced a deal to buy a majority stake for about $632m, completed in 2013. By 2015 revenue passed $500m, and in December 2016 trivago listed on the Nasdaq. The cost of that scale was dependence. The company lived on traffic from Google and on the goodwill of booking sites that paid for clicks. In 2019 Schrömgens stepped back from the chief executive role after about fifteen years, and in 2023 Johannes Thomas took over. In April 2022 an Australian court fined the company AUD 44.7m for misleading consumers, a reminder that a comparison site is judged on whether it compares honestly. The recovery has been slow but real. For 2025 revenue grew 19%, net income reached €11.2m and the company held €130.9m in cash. Expedia still holds the majority. The lesson is in the order of things. Schrömgens had spent years in an earlier startup learning how users think before he built the company that made him known. The overnight success was a decade of preparation, and the crash of 2020 was survived by the habits that preparation built. ## Flink (Germany) URL: https://startup.talk/germany/flink/ Founders: Christoph Cordes, Oliver Merkel, Julian Dames, Saad Saeed, Nikolas Bullwinkel Founded: 2020; Based in: Berlin, Berlin; Sector: Quick commerce / grocery delivery By early 2025 two of Flink's founders had walked away from the company they started. Christoph Cordes and Oliver Merkel left, and Julian Dames was alone at the helm of a business that had once been worth $2.1bn on paper. It was an odd ending for a company that had outlived most of its rivals. Cordes was no first-time founder. In 2009 he co-founded fashion4home, an online furniture and home shop, and from 2016 to 2020 he sat on the management board of home24 SE. He had spent more than a decade learning how hard it is to move bulky goods to customers who want them now. Grocery delivery, with its razor-thin margins, was the same problem at higher speed. At the end of 2020 he, Merkel, Dames, Saad Saeed and Nikolas Bullwinkel started Flink in Berlin. The idea was copied from a pattern already visible in Turkey and the United States: small dark stores, bike riders, groceries in minutes. In February 2021 Cherry Ventures, Northzone and Target Global led the first round, and within months the money was flowing. By the end of 2021 the company was valued at $2.1bn and had about 12,000 employees. What set Flink apart was a quiet decision to partner rather than fight. Rewe, the German supermarket giant, invested, and a group around DoorDash followed. In 2021 Flink overtook Gorillas to become the largest third-party quick-delivery service in Europe. Total funding reached around $750m. But speed has a price. In 2022 it bought the French rival Cajoo for about €100m, and Carrefour became a shareholder. That same year it left Austria. In April 2024 its French arm filed for bankruptcy. The empire had been built too wide and had to be cut back. What is left is a smaller, more focused company. One market profile puts 2024 revenue near €600m and says Flink reached EBITDA profitability, though other reports give lower numbers and the figures conflict. A 2025 round of about $100m, led by Prosus Ventures and Btomorrow Ventures, was reported as the company searched for fresh capital. For first-time founders the lesson is in who survives. Cordes arrived with a decade of e-commerce scars, and the company that outlasted its rivals was the one with a supermarket partner and a lean rebuild. In fast-burn markets, the winner is often the one that stops growing first. ## Gorillas (Germany) URL: https://startup.talk/germany/gorillas/ Founders: Kağan Sümer, Jörg Kattner Founded: 2020; Based in: Berlin, Berlin; Sector: Quick commerce / grocery delivery In October 2021 Gorillas fired 350 employees in Berlin. They had joined a wildcat strike over unpaid wages and heavy loads, and the company, then barely a year and a half old, answered by sending them letters. It was the moment the cheerful startup with the slogan 'faster than you' met the cost of its own speed. Kağan Sümer, the Turkish entrepreneur who co-founded the company with Jörg Kattner in May 2020, had built it on a borrowed idea. The model was based on the American goPuff: small neighbourhood warehouses, bike couriers, a promise of groceries in minutes. Berlin in lockdown was the ideal place to test it. The money arrived with a speed that surprised even investors. A $44m Series A in December 2020 was followed by $290m in March 2021 from backers including Tencent, DST Global and Coatue. In October 2021 came a round of nearly $1bn, with Delivery Hero adding about $200m. Gorillas became a unicorn in about nine months, the fastest in Germany. But the riders were carrying the weight. They reported heavy loads, back pain and pay problems. After the strike and the dismissals the company bought 1,200 e-bikes and introduced a bonus scheme, a late repair to a wound it had made itself. What followed was a retreat. In 2022 Gorillas bought Frichti in France and launched its own-label products, but it was also leaving Belgium, Italy and Spain, closing most of its UK operations outside London and cancelling Switzerland and Australia. Sümer said he wanted to raise $700m to expand and become profitable. Instead, in December 2022 the Istanbul rival Getir bought Gorillas for $1.2bn. In February 2023 a UK regulator banned an advert as irresponsible, and by June 2023 Gorillas was leaving France. The last operations ended in April and May 2024. The founders' own personal paths are thinly documented, and Sümer was reported to be leaving his role after the sale. What the record does show is a company that tried to do three hard things at once: grow faster than any German company before it, treat riders as the core asset, and make money on groceries that sell at near-cost. The lesson for first-time founders is the order of operations. Growth can be bought. Trust from the people who actually deliver the product cannot be, and when the strike came, the speed that had made the brand became the story against it. ## Blinkist (Germany) URL: https://startup.talk/germany/blinkist/ Founders: Holger Seim, Tobias Balling, Niklas Jansen, Sebastian Klein Founded: 2012; Based in: Berlin, Berlin; Sector: Consumer edtech / non-fiction book summaries For about two years after Blinkist 1.0 launched in 2012, the product that later reached millions of readers was barely holding on. Blinkist's own account is blunt about it: money was short, downloads were low, the product had problems, and nobody knew how to reach the right people. The four founders, who had been friends since university, kept going anyway. The idea came from a habit, not a market study. Sebastian Klein has recalled taking notes while reading non-fiction books at university. Together with Holger Seim, Tobias Balling and Niklas Jansen he turned that habit into a company on August 21, 2012. A podcast fragment even hints at an earlier, vaguer name, but it is too garbled to rely on, and the founders' own telling is that they began with a broad idea about learning before they narrowed it. That narrowing was the hard part. The first version did not click. Users downloaded, looked, and left. What the founders needed was a format that fit a life with no free time. They landed on the 'Blink': a book boiled down to fifteen minutes that could be read or heard on the way to work. It worked slowly and then quickly. After the public launch in 2013 the product grew into a catalogue of more than 6,500 titles, available in English, German and Spanish. In 2018 Insight Venture Partners led an $18.8m round, and the company counted about $35m in assets that year. Earlier backers included Greycroft, Headline, IBB and MGO. The price of success was a narrowing of ambition. A consumer subscription for non-fiction summaries is a pleasant business, but hard to turn into a giant one. In May 2023 the Australian edtech company Go1 acquired Blinkist, a deal reportedly driven by demand from employers who wanted learning content for their staff. The investors joined Go1's shareholder list. The company's story is told in public by its founders, and little of it concerns family or childhood. What the record shows is stubbornness. Two years of poor numbers, and still a team that stayed together. The lesson for first-time founders is the one the founders gave themselves: the format that works is rarely the first one you build. The business only appeared after they stopped explaining books and started respecting their readers' time. ## Tado (Germany) URL: https://startup.talk/germany/tado/ Founders: Christian Deilmann, Johannes Schwarz, Valentin Sawadski Founded: 2011; Based in: Munich, Bavaria (Bayern); Sector: Smart home climate control (smart thermostats, energy services) Most people name a company after a word that sounds good. The founders of a Munich startup named theirs after a feeling: coming home to a warm flat. The name tado is built from two Japanese greetings. Tadaima is what you say when you walk through your front door: I'm back. Okaeri is what the house answers: welcome back. In 2011, when Christian Deilmann, Johannes Schwarz and Valentin Sawadski started the company in Munich, the first thing their product could do was exactly that. It noticed your phone coming home and turned the heating up. That trick, geofencing, sounds small now. Then it was a bet that people would rather let a phone decide when they were in than program a thermostat, which almost nobody ever did. The first product, tado Heating, reached customers in Germany, Austria and Switzerland in November 2012, and the UK followed in 2013. The bet was easy to describe and brutal to build. Europe does not have one kind of heating. It has gas boilers, oil burners, district heat, old wiring and new, in every country. By 2014 the box could control more than 5,000 different heating systems, and that is where the early years went: not into slogans, but into integrations. In July 2014 an investor group led by Target Partners and Shortcut Ventures put in 10 million euros. In 2015 the company added a controller for air conditioners and showed it at an event in New York. Nobody outside the company sees the dull work behind a smart home gadget, but it is the moat. A rival can copy an app in a month. It cannot copy years of knowing how a thousand boilers misbehave. Money followed. Siemens, E.ON and Amazon were among the backers, and by early 2023 TechCrunch counted around 200 million dollars raised. More than 3 million thermostats had been sold by 2022. The company also bought Awattar, which helps households use power when it is cheap. Then came the part founders rarely enjoy telling. In January 2022 tado agreed to go public through a SPAC at a valuation of about 450 million euros, later trimmed to around 400 million as markets shook. In September 2022 the partner walked away. Deilmann said the cancellation reflected public capital market conditions. Tado raised 43 million euros privately in January 2023 instead, led by Trill Impact Ventures, and said it aimed for profit that year and had not laid anyone off. The quiet lesson sits in the name. The founders picked one human moment, coming home, and built everything around making it better. Funding cycles cooled, a listing slipped away, and the company still had something simple to point to: a warm house when you walk in. ## Marley Spoon (Germany) URL: https://startup.talk/germany/marley-spoon/ Founders: Fabian Siegel, Laura Maria Fernandez Founded: 2014; Based in: Berlin, Berlin (Bundesland); holding company registered in Luxembourg; Sector: Meal-kit delivery (food e-commerce) In April 2026 a company most customers had never heard of went bankrupt, and a Berlin meal-kit business felt it at once. FreshRealm, the US partner that packed and shipped Marley Spoon's boxes, filed for Chapter 11 on 27 April. Marley Spoon had to find someone else to do the work, and it did: Misfits Market took over as fulfilment and supply partner for the American business. That is what a meal-kit company really is. Behind the recipe cards and friendly marketing sits a hard logistics business, run on thin margins, fresh ingredients and other people's warehouses. Marley Spoon was founded in Berlin and grew by sending boxes of pre-portioned ingredients to homes in Europe, the United States and Australia. Public accounts of the founders' early days are scarce, so this profile sticks to what the record supports. What the record shows is a cycle many consumer startups know. Demand surged when people were stuck at home. Then it cooled. Net revenue for 2025 came in at about 238 million euros, down more than 22 percent on the year before. In the first half of 2026 it fell further, to 105.6 million euros from 138.9 million a year earlier, a drop of about 24 percent. The company paid to stay visible. Marketing rose to about 21 million euros in that half, around a fifth of revenue, against roughly 15 percent a year earlier. Operating EBITDA, which had been a small profit of 5.6 million euros in the first half of 2025, swung to a loss of 6.8 million. Management guided to revenue of 175 to 185 million euros for the full year 2026 and a loss at the operating level. There is also a brand story. In 2025 analysts noted that the budget brand Dinnerly was being discontinued in Europe while it carried on elsewhere. Cutting a brand that works in one market and keeping it in another is the sort of choice that shows how thin the margins are. What can a first-time founder take from it? Growth that is lifted by an unusual moment is not the same as lasting demand. A business that depends on a single supplier can be hurt by that supplier's failure in a single week, and the founders who survive are the ones who have a second name on speed dial before they need it. ## Sennder (Germany) URL: https://startup.talk/germany/sennder/ Founders: David Nothacker, Julius Köhler, Nicolaus Schefenacker Founded: 2015; Based in: Berlin, Berlin (Bundesland); Sector: Digital freight forwarding / road logistics The first sennder office was a garage. It was late 2016, there were five employees, and the business they were building was not the one they had started with. David Nothacker had begun it a year earlier while doing his MBA at INSEAD, and the project was about delivering parcels. That was the obvious thing to try: parcels were what people thought of when they thought of online commerce. Within months the model had to change. Parcel delivery is a crowded, low-margin game owned by giants. So he pivoted. He was joined by two people he knew from his earlier job at consulting firm Roland Berger, Nicolaus Schefenacker and Julius Köhler. In mid-2016 the company changed direction to full truckloads, the big trucks that haul pallets between factories, warehouses and ports. It was less glamorous and much larger. The idea was simple to say. Europe's road freight runs through thousands of small trucking firms, many of whom still find work by phone. If software could match a shipper's load to the right truck quickly, everybody would waste fewer empty kilometres. They registered the company in Berlin, took small angel cheques in 2016, and by the end of 2017 had a partnership with Scania and about 30 people. Then the money came, in steps. HV Capital and Project A led the Series A in early 2018, and Accel and Lakestar followed in 2019. By the end of 2019 there were 200 staff. In 2020 the company merged with the French firm Everoad and bought Uber Freight's European business, and headcount went from 300 to 700 in a single year. In early 2021 it raised 160 million dollars and passed a billion-dollar valuation. The next chapter was about size. Sennder bought Cars&Cargo and another firm, and by mid-2022 had 1,000 employees. In 2025 it took over the European surface transportation operations of C.H. Robinson. The company now says it is profitable and expects to approach 1 billion euros in annual revenue. The lesson is in the garage. Sennder did not win because the first idea was right. It won because the founders gave up the first idea quickly, found a boring market with real pain, and spent the next ten years buying their way up it. Giving up early is sometimes the most valuable decision a founder makes. ## Volocopter (Germany) URL: https://startup.talk/germany/volocopter/ Founders: Stephan Wolf, Alexander Zosel Founded: 2011; Based in: Bruchsal, Baden-Württemberg; Sector: Electric air taxis (eVTOL aviation) In 2013 a small company from the town of Bruchsal asked ordinary people for half a million euros. It got the money in two hours and 35 minutes, a European crowdfunding record at the time. The company was e-volo, which later became Volocopter. Its founders, Stephan Wolf and Alexander Zosel, had started it in 2011 with a strange idea: a multicopter big enough to carry a person, powered by many small electric rotors instead of a single large one. On 21 October 2011 the first prototype, the VC1, flew. Many people in aviation thought it was a toy. The crowd at Seedmatch disagreed, and its confidence bought the company time. That money mattered because aircraft are expensive and slow to certify. The company needed years, test pilots, regulators and a manufacturer. In 2017 a larger investment of about 25 million euros arrived. In 2019 Geely led a 55 million dollar round. That year Volocopter flew a demonstration over Singapore's Marina Bay, with a temporary vertiport built for the show. In 2022 it raised 170 million dollars to launch air taxi services, and in 2023 it flew in Rome and New York. Flying was not the hard part. Paper was. A passenger aircraft needs certification, and an electric air taxi has no road map. Plans to carry athletes at the Paris 2024 Olympics were dropped because the engine was not certified. The Singapore launch was put on hold indefinitely in November 2023 because the company lacked local funding partners. The cost of waiting for approval is that cash keeps burning. On 30 December 2024, Volocopter filed for insolvency because it could not keep its funding going. The firm kept operating during the proceedings. In March 2025 a rescue came from China: Diamond Aircraft, owned by Wanfeng, took over the company. It stays in Bruchsal. Passenger flights were pushed to 2026, and the latest report says the VoloXPro should be certified in Germany as a light sport aircraft by the end of 2026. Founders can read two things in this. The crowd was right that people wanted to see the thing fly. The cheque that matters, though, is the one that carries you across the years between prototype and permit, and nobody can crowdfund that. ## Neura Robotics (Germany) URL: https://startup.talk/germany/neura-robotics/ Founders: David Reger Founded: 2019 (founded 26 Mar 2019 as Han's Robot Germany GmbH; renamed Neura Nov 2020); Based in: Metzingen, Baden-Württemberg; Sector: Robotics (cognitive cobots, humanoids, AI software) David Reger got his first big robotics job by pretending. In 2011 a Swiss machinery maker called Mabi told him he could work on robots or on 3D printers. He picked robots. He had never worked with one, and he has admitted he pretended he had. He taught himself at night, from YouTube videos. The man who would one day run a company valued in the billions grew up near Rottweil, the eldest of eleven children. His father worked in a factory making auto parts, his mother stayed at home, and on Saturdays the boy went to the factory to clean machines. He skipped university and did a three-and-a-half-year apprenticeship as a mold maker. At 21 he moved to San Francisco and spent about a year and a half as a case manager for homeless people, then came home with a new appreciation for Germany's safety net. Within a year of joining Mabi, his small team showed a large industrial robot arm at the Automatica fair in Munich. He stayed seven years. By 2018 he wanted to build machines that could learn new tasks and pass that knowledge to each other through a shared platform. The board told him, he says, that it was five or six years too early. He thought waiting would make him late. So he left. He teamed up with Han's Robot, a Shenzhen maker of collaborative robots, and in March 2019 set up a German company, Han's Robot Germany. He hired about a dozen engineers and scientists, some from Mabi. The Chinese partner owned 70 percent. The company was renamed Neura in 2020. In June 2023 Reger bought out Han's stake for 11 million dollars. A month later Neura raised 55 million. In January 2025 came a 120 million euro Series B, and in June 2026 an announced Series C of up to 1.4 billion dollars led by Tether, with Nvidia, Amazon, Qualcomm and Bosch also investing. People close to the deal put the valuation near 7 billion dollars. Forbes estimated his stake, close to 20 percent, at roughly 1.5 billion. The company has not published revenue. Its last public filing, from 2022, showed a loss of nearly 7 million dollars, though it says orders exceed a billion. The quiet lesson is in the pretending. Reger did not wait to feel qualified. He said yes, learned the thing at night, and kept saying yes until the thing was his. ## Mister Spex (Germany) URL: https://startup.talk/germany/mister-spex/ Founders: Dirk Graber, Björn Sykora, Philipp Frenkel, Thilo Hardt Founded: 2007 (site online April 2008); Based in: Berlin, Berlin; Sector: Online and omnichannel optical retail (glasses, contact lenses) In 2007, Dirk Graber had a problem that no amount of clever code could fix. He wanted to sell glasses on the internet, and the people who made the glasses did not want to sell them to him. Many frame makers, the Berlin business press later recorded, were wary of an online seller who might undercut their opticians. Graber had been talking the idea through with Lukasz Gadowski, the founder of Spreadshirt, who helped him sharpen the model. He then pulled in Björn Sykora, a friend from his student days, and two IT builders, Philipp Frenkel and Thilo Hardt. So the first job was not a website. It was persuasion. Get one maker to say yes, then the next. Without frames there was nothing to photograph, nothing to ship, nothing to sell. The company was registered at the end of 2007 in Berlin and the shop went live in April 2008. A few early backers, among them DN Capital, XAnge and High-Tech Gründerfonds, put in a reported seven million euros by 2010. France came in 2010. In 2011 Graber brought in Mirko Caspar as co-CEO, and in 2013 the group bought the Swedish contact-lens seller Lensstore. What the founders had sensed early was that eyewear is a repeat, low-drama purchase that people would happily make from a sofa. What they underestimated was how long the physical world would keep pulling at them. Mister Spex opened shops, then more shops. It was no longer just a web business. In July 2021 it listed in Frankfurt, selling 15 million shares at 25 euros and raising 375 million euros. It was a peak. Forecasts were cut afterwards, and the share price slid. Caspar left in 2023. In 2024, Graber, by then the last founder still at the company, stepped down. A programme called SpexFocus shut all eight international locations and laid off around 150 people. Tobias Krauss took over as CEO in April 2025. The company reported revenue of 174.9 million euros for 2025, runs 67 stores in Germany, and in 2026 began handing logistics and production to Arvato and Rodenstock while closing its Berlin-Spandau site. The lesson is in the first month, not the IPO. Graber's real start-up capital was a few suppliers who took a chance on an outsider. Everything that followed, the cash, the bell in Frankfurt, the retreat, rested on that first yes. ## Sono Motors (Germany) URL: https://startup.talk/germany/sono-motors/ Founders: Laurin Hahn, Jona Christians, Navina Pernsteiner Founded: 2016 (GmbH, Karben, Hesse); prototype work reportedly began earlier; Based in: Munich, Bavaria; Sector: Solar-integrated electric vehicles; later solar technology for buses and trucks In December 2019, Laurin Hahn and Jona Christians had a decision to make. They could take venture money and the strings attached, or go back to the people who had already paid deposits on a car that did not yet exist. They said at the time that conventional financing would have meant giving up their principles and the rights to their patents. So they went back to the crowd. It is a choice that only makes sense once you know how they began. Early reports describe a basement and a home garage in Munich, and a pair of teenagers, as TechCrunch put it, building a solar car prototype. Hahn is described as an electrical engineer by training; Christians as someone who studied computer science and experimental physics. Navina Pernsteiner completed the trio when Sono Motors was formally founded in 2016. Crowdfunding was not a gimmick for them. It was the only door open. The first Indiegogo campaign raised about 550,000 euros from 1,686 backers, and a Seedrs round added 1.8 million. Business angels Matthias Willenbacher and Marita Hansen backed them early. In July 2017 two working prototypes, built with the small-series maker Roding, were shown off. The real test was the 2019 campaign. The aim was 50 million euros by 30 December. The date passed; they extended to 20 January 2020 and ended with about 53.3 million euros in pledges, though not all of that was binding. Reservations for the Sion reached around 13,000. The problem with promises is that they must be built. In November 2021 the parent listed on NASDAQ and raised about 135 million dollars. Production moved from Sweden to Finland. Then the money would not stretch. In 2022 the founders appealed to the community with a #savesion campaign for 3,500 more pre-orders at a minimum of 26,900 euros each. In February 2023 they ended the Sion. Insolvency under self-administration followed in May. A restructuring plan and an investor deal in November 2023 saved the company; the application was withdrawn in January 2024 and the business shifted to a solar kit for buses and trucks, now under the SonoSolar name. Reports in 2026 say bankruptcy has since returned. The founders' original instinct was sound: the customers were the investors. The weakness was not belief but arithmetic. Reservations are not revenue, and a car is a factory before it is a car. ## Jodel (Germany) URL: https://startup.talk/germany/jodel/ Founders: Alessio Borgmeyer, Tim Schmitz, Niklas Henckell, Alexander Linewitsch Founded: 2014; Based in: Berlin, Berlin; Sector: Anonymous hyperlocal social network (consumer app) In October 2014, a student at RWTH Aachen put an app on the market that let strangers speak to each other without names. Alessio Borgmeyer was still a student, and the app was small enough to be defined by a rule: you could only see what people within a few kilometres were posting. That constraint, not a clever algorithm, was the idea. Posts were capped at 150 characters, later raised to 240. Readers voted up or down, and anything that reached minus five disappeared. The result felt less like a global feed and more like a hallway wall in a student dorm. Borgmeyer built a team around him. Tim Schmitz took finance, Niklas Henckell the community, and Alexander Linewitsch the product. The company set up in Berlin. In November 2016 came a feature called Heimat, allowing users to post to their hometown even when they had left it. In 2017 the Traveller Feed let visitors talk to locals in foreign cities. Money followed. In June 2017 the company announced six million dollars from investors in Silicon Valley and Europe, among them Redalpine Venture Partners and Global Founders Capital, with a stated plan to enter the US. Advertising began in March 2018, local promotions in 2019, a premium subscription in 2021. What the app never fully solved was the cost of anonymity. Moderation is done by volunteer users, whose selection criteria are not public, and critics say majority voting nudges them toward the crowd instead of the rules. In November 2017 police identified a Trier student who had posted a shooting threat. A University of Passau study showed that the location of about 96 percent of 38,000 sampled posts could be pinpointed within 10 metres, which undercut the promise of anonymity. Critics also said it did not meet Germany's NetzDG reporting duties. Today The Jodel Venture GmbH employs about 60 people in Berlin and the app was at version 7.132 in early 2025. Survey data from May 2025 puts it at about four percent of social media users in Germany. The quiet lesson: a hard limit, here distance, can be a product. The harder lesson is that a promise like anonymity carries a debt you keep paying for years. ## Tubulis (Germany) URL: https://startup.talk/germany/tubulis/ Founders: Dominik Schumacher Founded: ; Based in: Munich (Martinsried), Bavaria; Sector: Biotech, antibody-drug conjugates (cancer) When Gilead said in April 2026 that it would pay up to 5 billion dollars for a Munich start-up, the figure that stood out was the upfront part: 2.7 billion euros in cash, with 1.6 billion more tied to milestones. For a company whose products are still in clinical trials, that was a vote of confidence. Tubulis develops antibody-drug conjugates, or ADCs, which pair an antibody that seeks out a cancer cell with a toxic payload that kills it. The aim is to hit the tumour and spare healthy tissue. Its lead candidate, TUB-040, is being tested in ovarian and lung cancer, and a second molecule, TUB-030, has shown early promising data. The company is led by co-founder and chief executive Dominik Schumacher. The money tells the story. In October 2025 Tubulis closed a Series C round of 308 million euros, around 358 million dollars, which took total financing to roughly 495 million euros. Schumacher said the funds would let it test therapies in earlier lines of treatment and across more tumour types. An earlier upsized Series B2 raised 128 million euros. Six months after the Series C, the buyer arrived. Gilead's chairman and CEO Daniel O'Day was quoted in the announcement. The plan was that Tubulis would stay as an independent research unit inside Gilead and that Munich would remain an ADC innovation hub. Closing was expected in the second quarter of 2026. The pace says something to founders. In biotech, the market does not wait for approval: it pays for evidence that a platform can produce candidates that work in humans. A start-up that reaches that point can choose between independence and a big partner. Tubulis, with plenty of cash, chose the second. ## CureVac (Germany) URL: https://startup.talk/germany/curevac/ Founders: Ingmar Hoerr, Steve Pascolo, Florian von der Mülbe, Günther Jung, Hans-Georg Rammensee Founded: 2000; Based in: Tübingen, Baden-Württemberg; Sector: Biotech, mRNA vaccines and therapeutics Before Ingmar Hoerr ever held a pipette professionally, he drove an ambulance. From 1988 to 1990, he served his civilian service as a paramedic with the German Red Cross in Nürtingen, near Stuttgart, after finishing school at an agricultural high school. It was a long way from the labs of Tübingen, but it points to a man who learned early that medicine is about timing. He was born in Neckarsulm in 1968 and studied biology at the University of Tübingen from 1990, including a year at Madurai Kamaraj University in India. His doctoral work, completed in 1999 in Günther Jung's institute in cooperation with Hans-Georg Rammensee's group, asked a plain question. Does mRNA vanish the moment it enters a body? He vaccinated mice with a stabilised RNA construct and showed that it did not decay as quickly as people assumed. On 9 September 1999 he applied for a first patent. A year later, with colleagues Steve Pascolo and Florian von der Mülbe, and with Jung and Rammensee, he founded CureVac in Tübingen. Prizes came first: a Baden-Württemberg innovation prize in 2007, then a 2 million euro EU vaccine prize in 2014. The Gates Foundation funded vaccine projects in 2015, and by 2017 equity investment had reached about 305 million euros and a valuation of 1.4 billion. Then came 2020. In March, Hoerr was back as CEO after the company replaced its chief executive following a reported meeting with Donald Trump, and a German newspaper claimed the US wanted CureVac to relocate. The company denied any offer. That same month, Hoerr suffered a severe health issue, and the details have not been made public. In June the German state bank KfW paid 300 million euros for about 23 percent, and in August CureVac listed on Nasdaq, raising 213 million dollars. The fall came in June 2021. The Covid candidate showed 47 percent efficacy, short of the 50 percent target. CureVac dropped it that October and turned to next-generation shots with GSK. In June 2025 BioNTech agreed to buy it in an all-stock deal valued at 1.25 billion dollars, and completed the takeover on 18 December 2025. The lesson is uncomfortable. Hoerr was early, but early is not the same as first across the line. A technology can be right and the company still lose the race. ## Magazino (Germany) URL: https://startup.talk/germany/magazino/ Founders: Frederik Brantner, Lukas Zanger, Nikolas Engelhard Founded: 2014; Based in: Munich, Bavaria; Sector: Warehouse robotics / intralogistics Not every founder gets to the finish line together. By 2020, Magazino's team of three founders was down to two: Nikolas Engelhard, one of the original trio who built the company out of a Technical University of Munich research environment, had left and later turned up at another Munich firm, Motius. That quiet departure sits inside a louder story. Magazino began in 2014, in Munich, with a hard bet: that a robot could do what warehouse automation had mostly avoided, picking single boxes of mixed sizes from normal shelves without a rebuilt warehouse around it. Conveyor belts and giant cube systems existed. A robot that walks up to an ordinary rack and grabs a shoe box did not. Frederik Brantner became chief executive and Lukas Zanger chief operating officer. Moritz Tenorth later joined as chief technology officer. They built TORU, a tall, narrow mobile robot that stores and retrieves cartons. It was slow, expensive work. Hardware does not iterate like software; every mistake meant metal. The early backers were an unusual mix of the people who would use the machine and the people who would fund it: the High-Tech Gruenderfonds, Siemens, the logistics group FIEGE, which put money in during 2017, and the fashion retailer Zalando, whose warehouses became the showcase. A fleet of dozens of TORU robots working beside human pickers is what turned a research idea into a product. What the company needed next was a patient owner. In 2020 the forklift maker Jungheinrich took a stake; in 2022 it raised that to 21.7 percent. Then, in August 2023, it bought everything. The price was never disclosed, though one market data site reported roughly €20 million for the last 74.5 percent, a number that should be treated cautiously. The founders did not disappear. Brantner and Zanger stayed on, and the company now sits inside one of Europe's biggest intralogistics groups, with about 130 people in its robotics team. The deal gave Jungheinrich, in its own words, one of the largest mobile robotics development teams on the continent. There is a lesson in the order of events. Magazino did not sell because it failed; it sold because hardware startups run out of runway before the market arrives. A strategic investor who first buys a minority, learns the technology and then takes the rest is a path a deep-tech founder can plan for, rather than fall into. It also shows what staying costs. One co-founder left in the middle of the climb, and the two who remained traded independence for the chance to see the robot reach more warehouses. Whether that was the right trade is a question only they can answer. The robots, at least, are still picking. ## NavVis (Germany) URL: https://startup.talk/germany/navvis/ Founders: Felix Reinshagen, Georg Schroth, Robert Huitl, Sebastian Hilsenbeck Founded: 2013; Based in: Munich, Bavaria; Sector: Reality capture / indoor mapping / digital twins In 2007, a doctoral student at Stanford's GPS Lab put a strange idea on paper: could a camera, not a satellite, tell you where you are indoors? Georg Schroth was working on his doctorate at the Technical University of Munich, and that visit to Stanford is where the NavVis story starts, six years before the company existed. GPS fails under a roof. That is the itch Schroth chased. Back in Munich, at the university's media technology chair, he and fellow PhD researchers Robert Huitl and Sebastian Hilsenbeck turned the idea into a research project: match what a phone camera sees to a stored image map of the building. Research papers do not pay salaries. What changed things in 2013 was an outsider. Felix Reinshagen, a McKinsey consultant, joined the three scientists, and the project became a company with support from the university and its entrepreneurship centre. The first products were not phone apps. They were machines. The M3, a trolley wheeled through corridors that captured images and laser data, launched in 2014, alongside a web viewer. A Series A that year, led by the Bavarian state fund BayBG with MIG, paid for it. In 2015 came €7.5 million from Target Partners and the American angel Don Dodge. The turn was quiet. Mapping buildings for navigation was a small market. Mapping them for owners, engineers and builders, who needed an accurate digital copy of a factory or a construction site, was a large one. NavVis opened in New York in 2016 and Shanghai in 2018, the year it raised $35.5 million and won the university's presidential entrepreneurship award. The hardware kept shrinking, from trolley to a wearable scanner in 2020 and handheld devices later. The software viewer was renamed IVION in 2021. Today the company counts over 300 employees from more than 60 nationalities and lists more than 2 billion square metres of buildings and sites on its platform. The lesson is in the sequence. A research idea about GPS indoors was right about the problem and wrong, at first, about who would pay. The founders kept the technology and changed the customer. It took six years from a doctoral sketch to a company, and a consultant who knew how to sell. ## Gropyus (Germany) URL: https://startup.talk/germany/gropyus/ Founders: Markus Fuhrmann, Philipp Erler, Bernd Oswald Founded: 2019; Based in: Vienna, Austria (Berlin and Richen, Baden-Wuerttemberg operations); Sector: Proptech / industrialised timber housing Barely three months after he stepped down as chief executive, the company Markus Fuhrmann started was in trouble. In August 2026 he moved to the supervisory board of Gropyus; on 3 November 2026 several of its subsidiaries filed for insolvency, citing an acute liquidity shortage. Fuhrmann was no stranger to building fast. Before Gropyus he co-founded Lieferheld, the German food delivery service that became part of Delivery Hero. Philipp Erler, who stayed on as chief executive, came from the technology side of Zalando. Gropyus, founded in 2019, had a plan that sounded like software ambition applied to concrete problems: build apartments in a factory, in timber, with robots, and design them on a digital platform so each one costs less and is built faster. The pitch fitted a country short of housing. The money came from one source above all. Vonovia, Germany's largest landlord, put in €100 million in 2023, another €100 million in 2024, and a third €100 million in March 2026. Its subsidiary Buwog came to hold about a quarter. Total equity raised passed €400 million. A plant in Richen ran about 50 robots, and the group employed roughly 550 people across Vienna, Berlin, Richen and Steinhaus. But the market moved the wrong way. The company points to high construction and financing costs and long approval times; German approvals for new apartments fell about 35 percent between 2020 and 2025. Parent company revenue slipped from €11.2 million in 2024 to €7.9 million in 2025, though the operating loss narrowed from €23.6 million to €12.9 million. The end, for now, came from a gap in timing, not a lack of belief: a contractually agreed investor payment due did not arrive, and bridge funding from existing shareholders could not cover every entity. The parent, Gropyus AG, was said not to be affected. The lesson is cold. When one investor is also your biggest customer and almost your only source of capital, your runway is a single promise. Building houses like software needs the cash discipline of construction, because walls do not scale on faith. ## Zenjob (Germany) URL: https://startup.talk/germany/zenjob/ Founders: Fritz Trott, Cihan Aksakal, Frederik Fahning Founded: 2015; Based in: Berlin, Berlin; Sector: Flexible staffing marketplace / HR tech Two men met at the Berlin venture builder Atlantic Labs, and the idea they hatched was to put short-term jobs on a phone. Frederik Fahning and Cihan Aksakal, who had earlier founded the local deals site kaufDA, were soon joined by Fritz Trott, a former Rocket Internet manager. That mix, a product mind, a founder who had sold a company, and an operator, became Zenjob. The pitch was simple. Companies need extra hands for a weekend, a trade fair, a delivery peak. Students and other flexible workers want paid work that fits a timetable. Staffing agencies were slow and paper-heavy. Zenjob built an app to match the two sides in hours. The money followed in steps. A roughly €3 million round in 2016 came from Redalpine and Acton Capital Partners, with 500 Startups and Atlantic Labs. A Series A of about €8 million followed in 2017, and €15 million from AXA Venture Partners in 2018. One secondary account says an earlier Series A in 2016 fell through at the last minute; the claim appears only once, so treat it carefully. By the time later rounds arrived, including a $50 million growth round led by Aragon and a €45 million raise for expansion, the company had taken in around €107 million. Press reported revenue of about €60 million in the second pandemic year, a period when temporary work swung hard. That is the real fight in this business. A staffing marketplace takes a small cut of each hour, so growth means volume and volume means constant hiring on both sides. The pandemic shut events and restaurants overnight, which are exactly Zenjob's customers. The founders chose a plain problem and kept pushing at it for years instead of chasing a fashionable one. The lesson for a first-time founder is to pick a boring market with real daily pain, then stay long enough to outlast the competitors. ## Cognigy (Germany) URL: https://startup.talk/germany/cognigy/ Founders: Philipp Heltewig, Sascha Poggemann, Benjamin Mayr Founded: 2016; Based in: Duesseldorf, North Rhine-Westphalia; Sector: Conversational AI / customer service automation A bill for $955 million landed on a Duesseldorf software company in July 2025, nine years after three Germans started it. Philipp Heltewig, Sascha Poggemann and Benjamin Mayr founded Cognigy in 2016 on the thesis that conversational AI would remake enterprise customer service. It began as a niche in a city better known for fashion and advertising than for AI. The idea was not a chatbot for consumers but a platform that large companies could use to build voice and chat agents for their contact centres. Progress was slow, then sudden. A Series A of about $6 million from DN Capital arrived in 2019. Insight Partners led a $44 million Series B in 2021. By 2024, when Eurazeo Growth led a $100 million Series C, annual recurring revenue reportedly stood near $37 million, up from about $17.5 million the year before. Over 1,000 brands, among them Mercedes-Benz, Nestle, Lufthansa and Bosch, used the product. The turn came with the boom in large language models. Cognigy had spent years on the governance, integrations and enterprise plumbing that a flashy model alone cannot offer. Buyers who wanted AI agents without losing control found it ready. In April 2025 it moved its US headquarters to North Texas. On 28 July NiCE, the Israeli contact-centre software group, announced it would acquire Cognigy for about $955 million, including roughly $50 million held back, and closed the deal later in the year. Heltewig described the sale as a milestone offering growth for customers and employees. The price drew debate. One analyst argued the multiple was no AI premium at all and speculated founders took tens of millions each; that was his guess, not a reported fact. What is documented is the scale: about 300 employees and roughly $165 million raised. The founders' path suggests a useful rule: build the boring layer others will need, years before the market arrives. When demand turned, Cognigy was already there. ## Mistral AI (France) URL: https://startup.talk/france/mistral-ai/ Founders: Arthur Mensch, Guillaume Lample, Timothée Lacroix Founded: 2023 (incorporated 28 April 2023); Based in: Paris, Île-de-France; Sector: Artificial intelligence / foundation models Before Arthur Mensch ever trained a language model, he spent three years staring at pictures of the human brain. Between 2015 and 2018 he was a doctoral student at Inria and NeuroSpin, the neuroimaging centre at CEA Saclay, working on optimisation methods to read patterns in fMRI scans. It was a long way from chatbots. He had come to it by the most French of routes: raised in Ville-d'Avray on the western edge of Paris, a businessman's son whose mother taught physics, then the preparatory class at lycée Hoche and a place at École polytechnique. The brain work was only the first of many detours. A postdoc at ENS Paris on optimal transport. A stretch at New York University's Courant Institute with Joan Bruna. Then, in 2020, a job at Google DeepMind's small Paris office, where he stayed about three years and worked on large language models. By any measure he had arrived: a safe, well-paid seat at the one of the best labs on earth. What nobody outside that circle knew was how crowded his own head had become with a single question. If the best language models were being built behind closed doors in California, who in Europe would ever be allowed to look inside one? He left DeepMind in May 2023. With him were two old classmates from École polytechnique, Guillaume Lample and Timothée Lacroix, both of whom had been building large models at Meta. They incorporated Mistral AI on 28 April 2023, before Mensch had even resigned. In June they raised 105 million euros, an astonishing seed for a company with no product, from backers including Lightspeed, Xavier Niel and former Google chief Eric Schmidt. Then they shipped. In September 2023 came Mistral 7B, released openly, which the company said beat Llama 2 13B on every benchmark it tested. Three months later a mixture-of-experts model, Mixtral 8x7B, was claimed to beat GPT-3.5 on most tests. In December 2023 the company was valued at over two billion euros. By June 2024 it was 5.8 billion. The price of that speed is capital. Training frontier models eats money at a pace that no European startup had ever faced, and in March 2026 Mistral borrowed 830 million dollars to build datacenters near Paris and in Sweden. A reported stake from Samsung in September 2026 put its valuation at around 21 billion euros. Critics, meanwhile, pointed to a 2024 study showing its models could reproduce copyrighted text, a reminder that moving fast leaves marks. Mensch has become the public face of European digital sovereignty, once arguing that France should not become a US 'vassal state' in AI. It is a long way from the quiet scan rooms at Saclay. The lesson for a young founder is hiding in that detour. Nothing Mensch studied was wasted: the maths, the brain, the big-lab years all turned out to be the same education. Wander if you must, but wander toward the hard problem. ## BlaBlaCar (France) URL: https://startup.talk/france/blablacar/ Founders: Frédéric Mazzella, Francis Nappez, Nicolas Brusson Founded: 2006 (Mazzella bought Covoiturage.fr; company Comuto founded 2006); Based in: Paris, Île-de-France; Sector: Ridesharing / long-distance carpooling marketplace In 2013, the year his company got a name the whole of Europe would learn, Frédéric Mazzella finished paying off a loan. It was 70,000 euros, borrowed to put himself through INSEAD. That detail says more about him than any valuation. Mazzella was born in Nantes in 1976 and grew up in Sérigné, in the Vendée, where his father taught mathematics and his mother French. A scholarship took him to the lycée Henri-IV in Paris, then to the École normale supérieure to study physics, then to Stanford from 1999 to 2002. He also studied music at the Conservatoire and plays four instruments. On paper, a safe academic career lay ahead. The itch came in 2003. Returning home from Paris for a holiday trip, he noticed that nearly every car heading his way carried an empty seat. The idea stuck with him. In 2006 he bought a small site called Covoiturage.fr and, with Nicolas Brusson and Francis Nappez, began to turn it into something bigger. What almost nobody saw was how long it paid nothing. Mazzella did not take a salary until 2009. The business had been funded first by 600,000 euros from founders, friends and family. Then came a million-euro-scale cheque from ISAI in 2010 and 7.5 million from Accel and others in 2012. Meanwhile the INSEAD debt hung in the background. The turn was a hard commercial choice. Carpooling had been free. In 2011 and 2012 the team introduced online booking, which let them charge a fee, and the platform began to work as a business. In 2013 the site was renamed BlaBlaCar. Investors followed: 100 million dollars from Index Ventures in 2014, and in 2015 200 million more at a 1.6 billion dollar valuation. Growth came at a price. By 2017 the company had opened offices in India, Turkey and Mexico and then closed them, admitting it had overspent and over-hired. In 2018 it bought Ouibus from SNCF and turned it into BlaBlaBus. In 2022 it suspended investment in its Russian subsidiary. Revenue reached about 250 million euros in 2023. Mazzella handed day-to-day management to Brusson in 2016, staying as chairman. Xavier Niel once called him the biggest star of the French internet, a title that sits oddly with a man who waited three years for his first pay cheque. The lesson is in the order of events. He borrowed against himself for an education, then worked for years unpaid on an idea nobody thought would scale. The debt and the empty seats were the same thing: a bet that something owned by one person could be shared. ## Doctolib (France) URL: https://startup.talk/france/doctolib/ Founders: Stanislas Niox-Chateau, Ivan Schneider, Jessy Bernal, Steve Abou Rjeily Founded: 2013; Based in: Levallois-Perret / Paris, Île-de-France; Sector: Healthtech / appointment booking and practice software At seventeen, Stanislas Niox-Chateau lost the future he had trained for since childhood. A serious back injury ended his hopes of turning professional at tennis. He was no casual player. Born in Paris in 1987 and raised in Boulogne-Billancourt, the son of a nursery teacher and a computer scientist, he followed a sport-study programme from age ten. He won the international youth Open at twelve and was a six-time Paris junior champion, part of the same generation as Gaël Monfils and Jérémy Chardy. What many people who know the Doctolib name do not know is that he also fought a strong stutter as a young man. He eventually overcame it. Years later, through a family endowment called Okola, he would help fund support for children, stuttering among them. After the injury he turned to books: a top science baccalauréat, the Saint-Louis preparatory class, then HEC Paris from 2006, where he became vice-president of the junior consultancy. In 2010 he joined investment fund Otium Capital and sat on the board of La Fourchette, the restaurant booking platform later bought by TripAdvisor. He had seen up close how a simple booking tool could change an industry. In August 2013 he and co-founders Ivan Schneider, Jessy Bernal and Steve Abou Rjeily launched Doctolib, starting with about fifty practitioners. A clinic became its first user in June 2014. Money was small at first, four million euros in 2014 from investors including the PriceMinister founders. Hospitals, famously cautious, were the prize: the Paris public hospital network AP-HP integrated the service in 2016 and adopted it in 2017. A year later Doctolib bought its direct rival MonDocteur and its 55,000 professional users. The cheques grew. 26 million euros in 2017 with Bpifrance, 150 million in 2019 at a one billion euro valuation, and 500 million in 2022 at 5.8 billion, the highest of any French startup at that moment. Success brought scrutiny. Critics challenged its Amazon hosting and the encryption claims it had made. In 2021 it received a Big Brother Award. In November 2025 France's competition authority fined it 4.6 million euros for abusing a dominant position, a decision it is appealing. On the numbers, revenue was about 384 million euros in 2025 and the company says it turned profitable that fiscal year. The tennis court taught him a habit he never dropped, to treat every booking slot as precious. Losing the sport at seventeen gave him a second chance to start from zero. Many founders discover that the thing that breaks them is the thing that points them to the work. ## Back Market (France) URL: https://startup.talk/france/back-market/ Founders: Thibaud Hug de Larauze, Quentin Le Brouster, Vianney Vaute Founded: 2014 (12 August 2014); Based in: Paris (18th arrondissement), Île-de-France; Sector: Refurbished electronics marketplace Before he sold anyone a phone, Thibaud Hug de Larauze helped build schools in Haiti. It was a gap year, taken after he had finished a scientific baccalauréat in Toronto in 2006 and enrolled at the IESEG business school. He was born in 1988 into an old family from the Quercy region, whose coat of arms bears an open oyster. His father, Bruno, would go on to run the food group IDEA. Haiti was a break from that world of boardrooms. He returned to finish an entrepreneurship MBA in 2010 and a master's in international negotiation in 2011. Then came a job at Neteven, a company helping merchants sell online. His work there brought him into contact with refurbishing factories, places that repaired and tested used electronics but had almost no way to reach consumers. What almost nobody saw, he later seems to have realised, was that these workshops were doing good work in the dark. In 2014, with Quentin Le Brouster and Vianney Vaute, he started Back Market. Their pitch was simple: refurbishing beats recycling, and a used phone can be as good as new. The first money was modest, about 300,000 euros from private investors in 2015. Trust was the product. In 2019 a consumer magazine said the platform accounted for most of the complaints in its investigation, and critics pointed at the one-year minimum warranty and lack of a new-battery option. Back Market responded with vetting and random tests, later added a battery option, and the defective rate reportedly dropped from 10% to 5%. Criticism that refurbishment was happening in China led it to show the country where each device was repaired. Capital followed. 41 million euros came in 2018 from Daphni, the Arnault group and Eurazeo, 110 million with Goldman Sachs Growth in 2020, and 276 million in 2021. In January 2022 a 450 million euro round valued it at 5.1 billion. By the end of that year it had cut 13% of its staff, 93 of 715 people. The company reports first profitability in Europe in 2024 and worldwide in 2025. Today it works with about 1,700 refurbishers and in September 2025 announced physical stores, starting in the United States. In 2024 Challenges estimated his fortune above 375 million euros. The lesson is a quiet one. The idea came from standing next to people doing unseen work. Before building a platform, spend time where the thing actually happens. ## Qonto (France) URL: https://startup.talk/france/qonto/ Founders: Alexandre Prot, Steve Anavi Founded: 2016 (4 April 2016); Based in: Paris, Île-de-France; Sector: Fintech / business banking for SMEs and freelancers Alexandre Prot grew up in a family where banking was the family business. His father, Baudouin Prot, was chairman of BNP Paribas, one of the largest banks in Europe. His son went on to build a company designed to be everything a traditional bank was not. That contrast is the most personal detail the public record offers about Qonto's co-founder. Little else is documented about his upbringing, so the story has to be told through what he and his partner built. Steve Anavi, the other founder, came from a different corner. He studied at EPFL in Switzerland, took an MBA at INSEAD, worked at Deloitte, co-founded a start-up called Smok.io, and then ran operations at Groupon. He had seen the machinery of a fast-scaling consumer business and the gaps in how small businesses were served. In April 2016 the pair founded the company, then under the name Olinda. The idea was to give freelancers and small companies a proper business account without the queue at a branch. By January 2017 they had raised a 1.6 million euro seed round from Alven Capital, Valar Ventures and angels. In July of the same year they added 10 million euros from their two main investors and opened the service to the public. The crucial step came in 2018 when the French regulator granted a payment institution licence. It meant Qonto could hold customer money legally. Money kept coming: 104 million euros in January 2020 with Tencent, and in 2022 a 486 million euro round at 4.4 billion. It entered Spain, Germany and Italy in 2019, bought Germany's Penta in 2022 and the accounting tool Regate in 2024. Growth came with a cost. In 2023 the French Modern Pentathlon Federation lost 672,139 euros to a scammer and sued Qonto over its security duties. In June 2025 Qonto was ordered to repay the full amount. A month later, in July 2025, it applied for a full banking licence. Numbers now say the strategy worked. Revenue reached 495 million euros in 2025, with net income of about 93 million and some 600,000 business customers. In January 2026 the company dropped its old legal name for Qonto. For a first-time founder the lesson sits in the family tree. Prot grew up next to the thing he later challenged. Knowing an incumbent from the inside can be the best head start, as long as you choose to build what it would not. ## Alan (France) URL: https://startup.talk/france/alan/ Founders: Jean-Charles Samuelian-Werve, Charles Gorintin Founded: 2016; Based in: Paris, Île-de-France; Sector: Health insurance / insurtech In 2016 two men set out to do something the French insurance industry had not seen in three decades: ask the regulator for an insurer's licence of their own. Alan describes itself as the first company to be granted one since 1986. That is a strange place to start a startup. Insurance is a business of trust and capital, and a newcomer has neither. The founders were Jean-Charles Samuelian-Werve, who became CEO, and Charles Gorintin, who took technology. They picked health, the part of insurance that most French employees touch every year and complain about most. Claims forms, delays, phone queues. The pitch was simple: build the insurer like a software company, so a member can do everything in an app. The early years were not glamorous. Alan's own year-in-review for 2018 describes the first minimum viable product, and the company says it found its first product-market fit in 2017. By the end of 2018 it insured close to 25,000 members across about 2,000 companies. That is small by insurer standards, but for a company that did not yet have a track record it was proof that small French firms would hand their staff's health cover to a stranger. What almost nobody outside the industry notices is how much the licence mattered. Because Alan was an insurer, not a broker, it controlled pricing, claims and customer service itself. It could fix a bad process the same week it found it. The money followed. Index Ventures became the anchor backer, and in September 2024 Alan was valued at EUR 4.5bn. But growth in insurance is expensive: the company reported losses of EUR 61m in 2023 and EUR 56m in 2024, the price of paying claims on members it had won with low prices and good service. Then the numbers turned. In March 2026 Alan raised about EUR 100m at a valuation near EUR 5bn, with Index again leading and Greenoaks, Kaaf, SH and Belfius joining; angels included Shopify's Tobi Lutke and footballer Antoine Griezmann. It reported EUR 785m in annualised recurring revenue for 2025, up 53%, and said it was operationally profitable in France. It also won a contract to cover up to 135,000 civil servants and their families, the kind of customer a 2016 newcomer would not have dreamed of. The lesson for founders is the unglamorous one. Alan did not invent health insurance. It went through the hardest door in the industry, the licence, and then treated every form, delay and phone queue as a bug to fix. The moat was the boring permission nobody else had bothered to earn. ## Ledger (France) URL: https://startup.talk/france/ledger/ Founders: Eric Larchevêque, Nicolas Bacca, Thomas France, Joël Pobeda, David Balland Founded: 2014; Based in: Paris, Île-de-France; Sector: Crypto security hardware / fintech Before he sold a single crypto wallet, Eric Larchevêque made his living at a poker table. He had sold his first company, a group of dating and adult-entertainment sites called Montorgueil, in 2007 for about EUR 22m. Then he played professionally for two years, until 2009, and went back to building companies. He was born in 1973 in Boulogne-Billancourt, but grew up in Vierzon, a small industrial town in central France, where his father André made porcelain. He trained as an engineer at ESIEE Paris, graduating in 1996, and co-wrote a book that same year on building advanced PCs. A man who writes a book on assembling computers at 23 is a man who likes taking machines apart. The companies came one after another. In 2011 he launched Prixing, a price-comparison service, and sold it to HighCo in 2014. That year he also opened La Maison du Bitcoin in Paris, a physical place where ordinary people could come and ask what bitcoin was. It was an odd bet. Most people who visited were curious, not rich. The turn came from the people around him. In 2014 a group of French crypto builders came together under the Ledger name, among them Nicolas Bacca, a smart-card specialist who had created BTChip, regarded as the first bitcoin hardware wallet. Larchevêque brought the business sense. Bacca brought the chip. The idea was blunt: if your coins sit on an exchange, you do not own them; if they sit on a small device that never shows your keys to a computer, you do. The company grew into a unicorn. A $75m round arrived in 2018 and a $380m round in 2021 at a reported $1.5bn valuation. Larchevêque led until 2019, then handed the CEO job to Pascal Gauthier and became chairman. The price of being famous in crypto showed in 2025. His associate David Balland was kidnapped, and Larchevêque publicly called for firearms permits for crypto professionals. The company had been through the 2020 leak of customer contact details, and a long crypto winter before that. As of 2026, Ledger is reported to be preparing a New York listing at a value above $4bn, and Challenges estimated Larchevêque's professional wealth at about EUR 340m in 2024. He remains a strategic shareholder. His other bets have not all worked. ALGOSUP, the computing school he co-founded in his hometown in 2022, closed in July 2025 for lack of financial balance. The poker years explain something about him: he treats a company as a hand to be played, and he has been willing to fold one and sit down at the next. ## Contentsquare (France) URL: https://startup.talk/france/contentsquare/ Founders: Jonathan Cherki Founded: 2012; Based in: Paris, Île-de-France; Sector: Digital experience analytics / SaaS Contentsquare started with a cheque smaller than the price of a Paris flat. In 2012 a club of software executives called Seed4Soft put in EUR 400,000 and bought a bet on a student. Jonathan Cherki was studying at ESSEC Business School in Paris when he began the project inside the school's incubator. He told the French start-up magazine Maddyness in 2024 that he started it as a student. Profiles describe the early work happening in an attic and at home. One weakly sourced profile places his birth in Marseille in 1986, with a grandfather who ran an import business in dried legumes, but that detail should be treated with caution. He had noticed a gap. Brands believed they were offering customers one experience, and customers were having another. A shop owner can see a shopper wander off. A website owner could not, until someone showed where people actually moved the mouse and where they gave up. The quiet years were long. Seed money in 2012. Then nothing large until June 2017, when a $20M Series A arrived. Five years is a long time to live on a seed round. It forced the product to pay attention to paying customers, not only to investors. When the money did come, it came fast. $42M in January 2018, $60M in January 2019, $190M in May 2020. Then in May 2021 SoftBank's Vision Fund 2 led a $500M round, then the largest in French tech. In July 2022 a $600M Series F led by Sixth Street, $400M of it equity and $200M debt, doubled the valuation to $5.6bn. Between rounds it bought its way into a bigger market: Clicktale and Pricing Assistant in 2019, Dareboost and Adapte Mon Web in 2020, Hotjar and Upstride in 2021, Heap in 2023, and Loris AI in 2025. It joined the French government's Next40 list in 2025. There is a price to buying so much. Combining products and teams across several countries is slow work, and the software market cooled after 2022. Third-party estimates put annual revenue near $370M in late 2024, and about 1,800 people work there. The takeaway for a first-time founder is the shape of that first cheque. A small seed forced Cherki to build something customers would pay for before the large rounds arrived. When the large rounds came, they were buying proof, not a promise. ## Deezer (France) URL: https://startup.talk/france/deezer/ Founders: Jonathan Benassaya, Daniel Marhely Founded: 2007; Based in: Paris, Île-de-France; Sector: Music streaming In January 2010 Jonathan Benassaya lost the company he had built. Deezer's board replaced him as CEO with Axel Dauchez. The numbers behind it were brutal: of roughly 12 million users, fewer than 15,000 were paying. Benassaya was born in Montpellier in 1980 and trained at the engineering school ENSAM and at ESSEC. He began in an investment bank, then in 2005 founded an online advertising agency aimed at the video-game industry, later renamed Magic Party. He took the business to Shanghai and Beijing to grow it, then decided to sell. Back in France he met Daniel Marhely. In 2006 they built Blogmusik, a site that let people stream music found on the web. It was popular and, in 2007, shut down as illegal. What the founders did next is the part most people skip. They did not walk away. They negotiated with the rights bodies, including SACEM, and relaunched in August 2007 as Deezer under licensing agreements. A pirate site had turned itself into a licensed business. Legal music is expensive. In the first half of 2008, revenue of EUR 875,000 did not cover licensing fees. In October 2008 the company raised $8.4M from AGF Private Equity and CM-CIC Capital Privé, taking total investment to $15.8M. The big money arrived in October 2012: $130M from Access Industries for international expansion. Access has held about 41% since 2016. The company that listed on Euronext Paris on 5 July 2022 was a different animal. Its shares fell 29% on the first day. It then spent three years proving it could make money, and in 2025 reported EUR 534M in revenue and its first full-year net profit, EUR 8.5M. It also flagged that up to 70% of streams on AI-generated tracks were fraudulent and blocked royalties for them. Benassaya had left by then. He left the presidency in November 2010 to work on other projects, including Stream Nation. For a founder, there is a harsh lesson in his story. The person who gets a company out of trouble with the law is not always the right person to get it through the long slog of making subscribers pay. Knowing that, and leaving anyway, is rarer than building the thing. ## Criteo (France) URL: https://startup.talk/france/criteo/ Founders: Jean-Baptiste Rudelle, Franck Le Ouay, Romain Niccoli Founded: 2005; Based in: Paris, Île-de-France; Sector: Advertising technology Jean-Baptiste Rudelle's first company began with a phone bill. He wanted cheaper international calls, so he built a call-back service called Kallback. It was a small, practical fix to a small, personal irritation. Everything he built afterwards came from the same habit. He was born in Paris in 1969. His father, Pierre-Marie, was a painter and his mother, Odile, a political science researcher. He studied telecommunications at Supélec, graduating in 1991, then took a master's at Imperial College London. His first jobs were at an EDF subsidiary and at Philips-Lucent, then consulting at Roland Berger and Arthur D. Little through 1999. In 1999 he co-founded Kiwee, which made SMS games, ringtones and logos for mobile phones. It sounds trivial now, but American Greetings bought it in 2004. That sale gave him a first exit, and a reputation. In 2005 he joined Franck Le Ouay and Romain Niccoli to found Criteo. What nobody expected was how slow it would be. The company spent its first four years on research and development. The first product launched in April 2008. The idea paid off. Criteo built display ads from a person's browsing behaviour and charged on a cost-per-click basis, so a shopper who looked at a pair of shoes saw those shoes again. Early funding was about $17M, and Bessemer put in $7M in 2010. In October 2013 the company went public, raising $251M. Then the platform shifted underneath it. In 2017 Apple's Intelligent Tracking Prevention cost Criteo about $25M in revenue. Eric Eichmann took over as CEO in 2016, Rudelle came back in 2018, and Megan Clarken followed in 2019. In June 2023 the French data authority fined the company EUR 40M, mainly for not verifying that people had consented to data processing. To outsiders, Rudelle is also a chess player. In June 2017 he and Veselin Topalov won an amateur-professional event at the Paris Grand Chess Tour, facing Garry Kasparov. He wrote a book in 2015, titled in English roughly "They told me it was impossible". Criteo reported US$1.94bn in revenue and US$145M in net income for 2025, with about 3,600 employees, and at the end of that year announced plans to move to the United States via Luxembourg. For a founder the thread is simple. Rudelle solved his own annoyance first, then each new company got bigger. The ones that lasted were built by someone patient enough to spend four years before the first product. ## Payfit (France) URL: https://startup.talk/france/payfit/ Founders: Firmin Zocchetto, Ghislain de Fontenay, Florian Fournier Founded: 2015 (incorporated 14 September 2015); Based in: Paris, Ile-de-France; Sector: Payroll and HR software (SaaS) In a high school in Laval, a small town in north-west France, two classmates built a social network for learning languages. It did not make them rich. It did make Firmin Zocchetto catch what he later called the startup bug, and he never really shook it off. His family came from Italian immigrants who had been entrepreneurs in construction, so working for yourself was not a foreign idea at home. Leaving university was another matter. After school Zocchetto enrolled at ESCP, one of the grand Paris business schools, and the plan on paper was a normal career. Then, while still a student, he started preparing a company with his old classmate Ghislain de Fontenay and a third partner, Florian Fournier. Balancing lectures with the company became impossible, and he dropped out. In an interview with Leaders League he was candid about the reception: for him the choice was obvious, he said, but for his family, a little less so. What almost nobody outside the founders' circle mentions is how ordinary the target was. Payfit did not chase a glamorous market. It went after payroll, the monthly chore that French companies dread because of the tangle of collective agreements, contribution rates and legal updates. The company was incorporated in September 2015, and the early money was small and fast: about half a million euros in January 2016 from Kima Ventures, TheFamily and Thibaud Elziere, then five million that October with Otium Venture and Xavier Niel. To handle the rules, the team even wrote its own programming language, called JetLang, to express payroll calculations. Then the cheques got big. Accel put in 14 million euros in 2017, Eurazeo and Bpifrance joined a 70 million round in 2019, and in January 2022 a 254 million euro round valued the company at 1.82 billion euros, France's 23rd unicorn. Growth had a price: the company later moved from pure software toward running payroll for clients as a managed service. The payoff arrived late. Reporting in July 2026 says Payfit reached profitability roughly ten years after its birth, with about 700 employees and 22,000 client companies in France, Spain and the UK, and recurring revenue near 80 million euros. Zocchetto is now the one giving advice. These days, he told Leaders League, he is the mentor. The lesson for a young founder is quiet: the unglamorous problem, attacked with a school friend you trust and a family that comes round slowly, can be worth a billion euros. ## Mirakl (France) URL: https://startup.talk/france/mirakl/ Founders: Philippe Corrot, Adrien Nussenbaum Founded: 2011/2012; Based in: Paris, Ile-de-France (and Boston); Sector: Marketplace software (SaaS) Two men who had sold a small company to a retail giant spent three years inside that giant, and then walked out to build what they had seen there. Philippe Corrot and Adrien Nussenbaum met in 2004. Two years later, in 2006, they founded Splitgames, a consumer marketplace. In November 2008 Fnac, the French retailer, bought it. That could have been the end of the story, a tidy exit and a comfortable job. Instead the pair were asked to build Fnac's own marketplace and led it for about three years. That is the quiet hidden detail of Mirakl's origins: the founders did not dream up a marketplace for retailers in a garage. They learned it from the inside, watching a big, established brand let outside sellers onto its shelves, and seeing what it needed. When they left Fnac they created Mirakl, in 2011 or 2012 depending on which record you read, with a simple promise. Any retailer or business could run a marketplace without building the technology itself. The money followed in careful steps. A 30 million dollar Series A led by Elaia came in November 2012, then 20 million dollars in 2015 led by 83North and Felix Capital, and 70 million in 2019 from Bain Capital Ventures. The pace changed in 2020 when Permira led a 300 million dollar round that valued the company above 1.5 billion dollars. A year later Silverlake led 555 million dollars at a valuation above 3.5 billion. The company set up in Paris and later in the Boston area, and reached roughly 750 employees across a dozen offices. In 2020 it launched dozens of new marketplaces, and when the pandemic hit it built StopCovid19.fr, a platform that helped distribute about 300,000 litres of hand sanitiser and 320,000 face masks in under a week. It kept buying capabilities, from invoice automation to retail media and, in 2025, a product aimed at shopping by AI agents. Business volume across customer marketplaces passed 8.6 billion dollars in 2023, and recurring revenue crossed 100 million dollars in 2021. Little is public about the founders' childhoods, so this is mostly a story of work. The takeaway is practical: the best product idea can come from the job you take after you sell your first company, if you pay attention to what the big customer cannot do for itself. They did not leave Fnac with a theory. They left with three years of watching the problem up close, and that turned out to be worth more. ## Vestiaire Collective (France) URL: https://startup.talk/france/vestiaire-collective/ Founders: Fanny Moizant, Sophie Hersan, Sebastien Fabre, Alexandre Cognard, Henrique Fernandes, Christian Jorge Founded: 2009; Based in: Paris, Ile-de-France; Sector: Pre-owned luxury fashion marketplace It began with a few thousand handbags, dresses and shoes that belonged to friends. In 2009, in Paris, Fanny Moizant and Sophie Hersan launched a website called Vestiaire de Copines, roughly the girlfriends' wardrobe, with about 3,000 pre-owned items offered by people they knew. Moizant had just finished a degree at the French fashion institute when she teamed up with Hersan, who had worked as a stylist. They were not building a tech company in the usual sense. They were turning a habit among friends, swapping and selling beautiful clothes, into a place where strangers could do it with confidence. The idea was ordinary, even cosy. What made it hard was trust: a Chanel bag sold by someone you have never met needs checking, photographing and authenticating. The wardrobe name did not travel. In 2012, when the site launched in the United Kingdom, it became Vestiaire Collective, a name that worked in any language. More co-founders had joined along the way, including Sebastien Fabre, Alexandre Cognard, Henrique Fernandes and Christian Jorge. Expansion followed: the United States in 2014, wider Europe in 2015, then Hong Kong, Singapore and Australia by 2017. The investors who came in tell the story of luxury taking resale seriously: Conde Nast, Kering, Tiger Global, SoftBank, Generation and others. In 2020 Vestiaire became the 11th French unicorn. A year later it earned B Corp certification, and in 2022 it bought the American resale firm Tradesy. Then came a bolder, more uncomfortable move. It announced a three-year plan to ban fast fashion brands, starting with 30 brands in 2022 and 33 in 2023, even though such items filled a lot of listings. The price of growing up showed in the leadership. Maximilian Bittner became chief executive, and in October 2025 Bernard Osta replaced him. In January 2026, after sixteen years, Fanny Moizant announced her own departure, leaving Hersan as the only original co-founder still at the company. A crowdfunding round in February 2024 raised around 3.5 million euros, more than three times its target, from customers themselves. Today it operates in over 70 countries. The lesson is understated: a business built on a favour among friends can become global, but only if you protect the trust that made the favour work, even when it costs you listings. ## Swile (France) URL: https://startup.talk/france/swile/ Founders: Loic Soubeyrand Founded: 2016 (as Lunchr); Based in: Montpellier, Occitanie; Sector: Employee benefits and meal vouchers fintech Loic Soubeyrand had already co-founded Teads, the video advertising company, when he went looking at something far more mundane: lunch. His new venture started in 2016 as Lunchr, an app to pre-order your meal and skip the queue. The first idea was small. What changed it was an observation about the bits of paper in people's pockets. Most meal vouchers in France were still paper, and the millions of salaried workers who use them were stuck with a clumsy system. Lunchr turned itself into a meal voucher business and launched a card on the Mastercard network. In March 2020 it was renamed Swile to signal that it was no longer only about lunch but about all the benefits a company gives its staff. The early money was modest. About 2.5 million euros came from Daphni in January 2017, then 11 million from Idinvest in 2018, and 30 million in February 2019 with Index Ventures and Kima Ventures, by which time it said it had 2,000 clients. In 2020 it raised 70 million, and in October 2021 a 200 million dollar round valued it above a billion, making it a French unicorn. Bold moves followed: a deal in 2021 to equip about 62,000 Carrefour employees, and acquisitions of Sweevana, Briq, a Brazilian benefits company and Okarito. The cost of speed was real. In 2021, Le Monde published testimonies from employees at several startups, Swile among them, describing toxic management, relayed by the Instagram account Balance ta start-up. That was a hard public moment for a company growing fast with a young workforce. And the accounts show the burn: in 2021 Swile lost about 41 million euros on revenue near 11 million. Then the arithmetic turned. Revenue reached 56.6 million euros in 2023, and about 171 million in 2024. The company says it became profitable in the last quarter of 2023, though reports differ on how the net result for 2024 should be read. It is based in Montpellier, and sponsors the ASVEL Lyon-Villeurbanne basketball club and Montpellier's football team. Little of Soubeyrand's early life is public. What the record does show is a founder who, after one success, chose a boring problem and paid attention to what people actually carried in their wallets. A second company can start from something as plain as a lunch voucher, as long as you are willing to rename it when it outgrows the name. ## Algolia (France) URL: https://startup.talk/france/algolia/ Founders: Nicolas Dessaigne, Julien Lemoine Founded: 2012 (8 October 2012); Based in: Paris, Ile-de-France (and San Francisco); Sector: Search and discovery API When Nicolas Dessaigne and Julien Lemoine agreed to start a company at the end of 2012, Dessaigne was 36 and Lemoine 31. Both had children and loans. Dessaigne later described that as a very risky situation, and the two gave themselves at least a year to find out whether it would work. Less than that, he argued, and you do not give yourself the chance to succeed. They had met years earlier at Arisem, a company later bought by Thales, and worked side by side for a few years. Lemoine, an EPITA graduate, left for the search company Exalead. Dessaigne, an engineer with a doctorate in computer science, followed a few years later. They overlapped there for only a few months. Then they decided to build something themselves, with Lemoine taking the technical lead because, as Dessaigne put it, he was far stronger at it. The first product was a technical success and a commercial flop. It was a search tool that worked offline inside mobile apps, and developers were not ready to pay for it. What nobody had planned for was what those developers asked next: could they have it online? A beta in March 2013 got a strong response. Around 1.5 million dollars from angels and funds, partly through one of the first beta testers, paid for the first two employees. Then came a rejection. Their first application to Y Combinator in 2013 failed. After a trip to the United States to visit clients in mid-2013, they were advised to apply again, and they got in for the winter 2014 batch. Y Combinator asked them to move to the United States, which opened the door to American investors and another 1.3 million dollars. Paying customers rose from 17 at the end of 2013 to 400 a year later. Funding followed: 18.3 million dollars in 2015 led by Accel, 53 million in 2017, and in July 2021 a 150 million dollar round valuing the company at 2.25 billion dollars. Algolia says it serves about 18,000 customers and handles more than 1.75 trillion searches a year. Dessaigne later became a general partner at Y Combinator. For a founder, the lesson is in the order of events: the flop was the product, the rejection was the invitation, and the thing that worked came from listening to the people who said no. ## Dataiku (France) URL: https://startup.talk/france/dataiku/ Founders: Florian Douetteau, Clément Stenac, Marc Batty, Thomas Cabrol Founded: 2013 (Paris); Based in: New York, USA (founded and R&D in Paris, Île-de-France); Sector: Enterprise AI / data science platform At twenty, Florian Douetteau made the kind of choice French parents fear. He was on the path to the École Normale Supérieure, the temple of French mathematics, and a speaker biography he later supplied to a conference says he walked away from the maths courses to work at a small search company that would become Exalead. He had been born and raised in Paris, the son of a system that sorts its brightest children early. He chose the noisy world of a startup instead. Exalead taught him two things. The first was how to build software that digs through enormous piles of data. The second was how to live inside a company that is always slightly short of money. He rose to run research and development there. Along the way he met Clément Stenac, a developer who had worked on VLC and Debian, two of the open-source projects that quietly run half the internet. Then came a detour. Douetteau left to become technical chief at IsCool Entertainment, a games company, where data meant players, clicks and money. What nobody outside the industry noticed was how much that job showed him. Companies were drowning in data and almost none of them had the people to use it. In 2013 he and three partners, Stenac, Marc Batty and Thomas Cabrol, started Dataiku in Paris. They joined the Agoranov incubator and then did something unfashionable. For about two years they grew on their own revenue, with no outside money. The first product, Data Science Studio, appeared in February 2014. The idea was plain. A bank, an airline or a maker of washing machines would have analysts who knew the business, engineers who could write code, and managers who signed the cheques. Dataiku gave them a single place to build and run models together. That turned a research toy into something a large company could buy. Money followed, and it followed from America. About €3 million came from French funds in 2015. Then FirstMark led a $14 million round in 2016, Iconiq led $101 million in 2018, and Google's investment arm CapitalG made Dataiku a unicorn in 2019. By 2021 the company had more than 1,200 staff and a $4.6 billion valuation. The centre of gravity moved to New York. The price came later. In December 2022, markets turned and Dataiku raised $200 million at $3.7 billion, a lower figure than a year earlier. In 2023 Douetteau said publicly that he was weighing an IPO. He still describes the firm as one that serves more than 500 large companies. The lesson is in the order of events. Douetteau did not start with a big idea. He started at twenty with a hard job at a small company, spent two decades learning where data goes to die, and only then built the thing. Leaving the prestigious path early gave him the long apprenticeship that made the company possible. ## OVHcloud (France) URL: https://startup.talk/france/ovhcloud/ Founders: Octave Klaba, Henryk Klaba Founded: 1999 (November, Roubaix); Based in: Roubaix, Hauts-de-France; Sector: Cloud computing / hosting The family was being watched. Before the Berlin Wall fell, the communist regime in Poland kept an eye on the Klabas, according to a portrait of Octave Klaba published in 2021. Octave was a teenager who already believed computers could change the world, and he grew up understanding that who holds your data decides how free you are. He arrived in Roubaix, in northern France, as a teenager with no French. Roubaix was a textile town past its prime. He studied engineering at the ICAM school in Lille, spent a short time at Alcatel, and in November 1999 he started a hosting company. His online handle at the time was Oles Van Herman. The company name came from its initials, and later got a second reading, 'On Vous Héberge'. He has said that on the internet 'if you're not paranoid, you get hacked', a line from a 2020 radio interview. What made OVH unusual was how it was built. Klaba made his own servers rather than buying branded ones. His father Henryk invented a water-cooling system for the data centres. Other family members worked in the company from the early days. Poland became the first foreign subsidiary in 2004, a nod to where the family came from. Then came the reveal that nobody on the outside expected. In the early 2000s the head of the French telecom Free lent OVH a cellar in Paris, and 10,000 servers ended up there. Klaba, according to Challenges, offered the owners of the building a deal. He would heat their flats using the warmth from his machines. The cheap, home-built approach turned a Roubaix start-up into Europe's largest hosting group. By 2024 revenue reached €993 million. The cost showed on the night of 9 to 10 March 2021. A fire destroyed the SBG2 data centre in Strasbourg and damaged its neighbour. Nearly four million websites went dark, and more than 100 companies later sought compensation. Seven months later OVHcloud listed in Paris. Its market value in 2025 was about €1.7 billion, below the figure hoped for at the IPO. Klaba, who stepped back as CEO in 2018, took the job again in October 2025. The lesson is that a founder's fear can become a design principle. A boy raised among watchers built a company around owning every layer: the servers, the cooling, the buildings. The same independence made it cheap, and the fire showed what it costs. ## Sorare (France) URL: https://startup.talk/france/sorare/ Founders: Nicolas Julia, Adrien Montfort Founded: 2018 (December); Based in: Paris, Île-de-France; Sector: Fantasy sports / NFT / Web3 gaming The first league to say yes was Belgium's. In 2018, two engineers with an unfinished card game for football fans got a licensing deal with the Jupiler Pro League. It was not Real Madrid. It was not even the French Ligue 1. But it was proof that a league would put its players' names on something built on a blockchain. Nicolas Julia and Adrien Montfort had met at Stratumn, a Paris blockchain company where Julia ran operations and Montfort led engineering, according to profile sites. They had spent years selling a technology that most clients did not understand. Julia has described getting into crypto companies from about 2015, and he was drawn to NFTs in 2018, long before the world had heard the word. They left Stratumn and started Sorare in December 2018. Julia, a graduate of the emlyon business school, handled the business. Montfort, with degrees in telecom engineering and data engineering, built the game. The idea was fantasy football where the cards are scarce, owned by the player, and can be sold. Money arrived in small amounts. About €550,000 of pre-seed came in May 2019 from investors including Xavier Niel's Kima Ventures. A $4 million seed followed in July 2020 with Partech and E.ventures, along with the former Germany international André Schürrle. Then Juventus signed in 2020, and with the club came Cristiano Ronaldo's card. The turn came in 2021, when NFTs became a global story. In February, Antoine Griezmann and Reddit's Alexis Ohanian joined a €40 million round. In September, Sorare raised €680 million at about $4.3 billion, a European record. Revenue that year was about €325 million, according to Les Échos. Mbappé, Zidane and, in November 2022, Messi became ambassadors. The fall-out was regulatory and financial. In July 2022 France's gambling regulator warned the company about its rules. In September 2024 the UK Gambling Commission prosecuted it, reportedly the first such case against a major Web3 platform. Sorare responded with stricter identity checks and self-exclusion tools, and in 2025 moved from Ethereum to Solana. A report in November 2025 said it might cut 35% of its staff to turn a profit in 2026. Nobody can say yet how it ends. What the early years do show is a pattern first-time founders can use. Julia and Montfort did not wait for a famous club. They took the small league that would sign, used it to approach bigger ones, and built each deal on the last one. ## Ankorstore (France) URL: https://startup.talk/france/ankorstore/ Founders: Pierre-Louis Lacoste, Nicolas d'Audiffret, Nicolas Cohen, Mathieu Alengrin Founded: 2019 (November); Based in: Paris, Île-de-France; Sector: B2B wholesale marketplace for independent retail The whole building heard them scream. In late 2019, four founders sat in a small office in Paris when the first order arrived on Ankorstore. Pierre-Louis Lacoste later told an interviewer that the noise carried through the walls. Lacoste knew exactly what a small shopkeeper feels, because he had been one. Before Ankorstore he ran his own shop, then spent seven years managing a clothing brand. He opened the first US concept store for the Basque brand Jean Vier in Los Angeles, held a senior role at the Canadian label Lolë from its Montreal headquarters, and then became country manager for Etsy in France. Each job showed him the same gap. Online platforms gave large retailers every tool, while the independent boutique had a phone and a catalogue. He has said Ankorstore was not a 'Eureka moment' but a long-held wish. The name comes from a retail term, the anchor store, the shop that pulls people down a street. He wanted independent retailers to be that anchor in their own town. He was not alone. Nicolas d'Audiffret and Nicolas Cohen had already built A Little Market, a site for handmade goods. Mathieu Alengrin came from Vestiaire Collective, where he was head engineer. The four started in November 2019 with a marketplace problem that kills most such businesses: no shops without brands, no brands without shops. They also had to convince investors that retail was not dead, he said. What made the difference was a rule. A shop could order from a brand with a minimum of €100, where classic wholesale might demand thousands. For a small boutique that was the difference between trying a new label and ignoring it. Investors saw it. Alven and Global Founders Capital backed a seed round of about €6 million. Index Ventures led €25 million in late 2020. In January 2022, just over two years after launch, a €250 million Series C led by Tiger Global valued Ankorstore at €1.75 billion. At that point it reported more than 200,000 independent retailers and 15,000 brands across 23 European countries, with almost 400 employees and plans to hire a thousand more. Rivals such as the American Faire press hard. Still, the founders show a useful habit. They did not chase a flashy invention. They took a problem they had lived behind a shop counter, sketched the fix in plain words, and gave it a name that explained the mission. ## Lydia (France) URL: https://startup.talk/france/lydia/ Founders: Cyril Chiche, Antoine Porte Founded: 2013; Based in: Paris, Île-de-France; Sector: Fintech / mobile payments and neobank Splitting a bill is the least glamorous moment in anyone's evening. The group sits around a table, someone reaches for a calculator, someone else says they will pay back tomorrow, and nobody remembers. Lydia began in that awkward minute. Cyril Chiche, the company's chief executive, and Antoine Porte, its technical chief, built an app that let one person send money to another with only a phone number. The app launched in 2013, and the story that follows is told through the product and the choices they made. The product's first years were quiet. Peer-to-peer transfers in France were slow and clumsy, run through bank websites that had not changed in a decade. Lydia worked on friction. It was easy to use, fast, and free for the sender, which suited students and young workers who owed small sums to friends. The turn came when Lydia stopped being only a way to repay a friend. It added a payment card, accounts, savings and subscriptions, and became what its makers called a financial super app. In January 2020 Chinese group Tencent led a €40 million Series B, a surprising backer for a French payments start-up. Then in December 2021, at the height of the fintech boom, an Accel-led $100 million round valued the company above $1 billion and made it one of France's unicorns. That valuation made life harder rather than easier. Revolut and N26 sold similar products across Europe, banks added instant payments, and fintech investors grew cautious after 2021. Lydia's answer was to widen its offer. A card called Sumeria and paid subscription plans aimed to turn millions of casual users into paying ones. One revenue estimate puts the company at about $87 million for 2024, though that figure comes from a tracking site rather than the company. The story is unfinished. What is clear is the method. Chiche and Porte picked a tiny, universal irritation, solved it better than the banks, and only later grew into the larger business that made investors take notice. A founder who wants a large market can begin with the smallest daily annoyance and earn trust one transfer at a time. ## ManoMano (France) URL: https://startup.talk/france/manomano/ Founders: Christian Raisson, Philippe de Chanville Founded: 2013 (company registered 26 July 2012 as Colibri SAS; site launched June 2013 as MonEchelle); Based in: Paris, Ile-de-France; Sector: DIY and home-improvement online marketplace The first name on the website was a ladder. MonEchelle.fr went live in June 2013 and sold what its name promised, plus about 30,000 other things for people who fix their own homes. The company behind it was not called MonEchelle at all. On paper it was Colibri SAS, registered in Paris in July 2012 with 150,000 euros. The two men behind it, Christian Raisson and Philippe de Chanville, had both gone through EDHEC business school. Raisson was not a lifelong tool person. His earlier jobs were at Deloitte, the consumer guide Guide Malin and the restaurant guide Gault Millau, a long way from drills and paint. What he and de Chanville spotted was a gap: DIY shopping in France was fragmented, and nobody online offered a deep catalogue to the amateur on a Saturday morning. Money came in small pieces. Business angels put in 300,000 euros in 2013. The historic shareholders added 150,000 more the next February, and about two million followed that September. That is a thin runway for a marketplace that has to win both sellers and shoppers. Then came the decision that mattered: in September 2015 they dropped the ladder and renamed the company ManoMano, a name that worked in Spanish and Italian as well as French. Spain and Italy opened that year, the UK in early 2016, Germany that November. What almost nobody remembers is how fast investors followed the expansion. Thirteen million euros came in March 2016, then 60 million from General Atlantic in 2017, then 110 million, then 125 million with Temasek in early 2020. By the end of 2020 the site claimed 1.2 billion euros of business volume and seven million active customers. In July 2021 a 355 million dollar round valued it at about 2.6 billion dollars. The bill arrived later. In October 2023 the company announced a restructuring with a plan to cut 230 jobs. Customer service moved to Madagascar and Portugal. Press investigations alleged that the platform encouraged fake positive reviews, which ManoMano denied, and freelance advisers took the company to court over their status. In April 2026, after more than a decade, the two founders stepped back from daily operations and Loic Derrien became chief executive. The lesson is hiding in the first week: a name is cheap to change and a business is not. They launched with a clumsy brand, learned fast, and changed the one thing that blocked growth before they spent real money on it. ## Owkin (France) URL: https://startup.talk/france/owkin/ Founders: Thomas Clozel, Gilles Wainrib Founded: 2016; Based in: Paris, Ile-de-France; Sector: AI for drug discovery and diagnostics (federated learning on medical data) Thomas Clozel was a hospital doctor when he decided the future of medicine would not be decided on the ward. He was an assistant professor of clinical haematology at Henri Mondor, in the Paris suburb of Creteil, surrounded by patients and by data that mostly stayed locked in filing systems. He came to medicine with an unusual family background. His parents, Jean-Paul and Martine Clozel, built the Swiss drug company Actelion and later started Idorsia. So he grew up near a model of how a scientist becomes an entrepreneur. He chose the clinic first. He trained in Paris and New York, and he has said that as a doctor he saw how advanced technology was becoming, while the benefits reached patients slowly. The itch was specific. Drug development, he argues, is not working well, and biology is too complex to be reasoned out by humans alone. In 2016 a friend introduced him to Gilles Wainrib, a researcher with a doctorate in applying artificial intelligence to biology. Clozel supplied the clinical eye, Wainrib the machine learning. They founded Owkin in Paris that August. The obstacle was built into their chosen field. Hospitals cannot simply hand over identifiable patient records, and privacy rules make that sensible. The company leaned on federated learning, where the model travels to the data rather than the other way round. In finance and industry that was established. In healthcare it was close to unused. What Clozel remembers best is not a funding round. In an interview he recalled the moment the first models analysing tissue slides began to work, and later a model called MesoNET that predicted survival in mesothelioma patients better than the classification pathologists use. That, he said, convinced him the approach could work. The money followed. In November 2021 Sanofi put in 180 million dollars, valuing Owkin at more than a billion. Swiss operations opened near Basel from October 2021, at the Idorsia site his parents had founded, and by 2022 two of its cancer diagnostic tools had European certification. In 2024 the Paris hospital network AP-HP renewed its partnership for at least five years. The takeaway is plain. A doctor's frustration, paired with someone who can build the tool, is a stronger starting point than a market chart. The friend who made the introduction turned out to matter as much as the investor who wrote the big cheque. ## Pennylane (France) URL: https://startup.talk/france/pennylane/ Founders: Arthur Waller, Quentin de Metz, Thierry Deo Founded: 2020; Based in: Paris, Ile-de-France; Sector: Accounting and finance software for SMEs and accountants The control over Pennylane does not sit where a cap table would suggest. The founders' preference shares carry 51 percent of the votes whatever the capital split, and they cannot be sold until the end of 2032. The biggest shareholder owns only about 14 percent of the capital. It is a quiet piece of engineering, and it tells you how the company thinks. Pennylane started in January 2020 in Paris, at a strange moment to launch anything. Arthur Waller, who became chief executive, built it with several co-founders, among them Quentin de Metz and Thierry Deo. The idea was to give small businesses and their accountants one shared financial operating system: bank feeds, invoices and cash flow in one place, so that the accountant stopped chasing the owner for receipts. The first money was modest. Four million euros arrived in May 2020. A 15 million series A followed in January 2021 and was topped up by another 15 million six months later, when Sequoia Capital joined. That year the team did something unusual for a software company. It briefly ran its own accounting service, then sold it to an independent firm in 2021, choosing to stay a platform that accountants could trust rather than a rival to them. That choice paid off. Fifty million euros came in January 2022. In February 2024, 40 million at a billion-euro valuation made Pennylane a unicorn. Seventy-five million more followed in April 2025 from Meritech and CapitalG, with annual recurring revenue then claimed at 60 million euros. By December the press reported about 115 million. The company stopped short of the story you might expect next. In January 2026 a record 175 million euro round led by TCV, with Blackstone Growth alongside, valued it above 3.5 billion euros. Yet the founders sold only about five percent of the company, and the voting shares keep control with them. For a first-time founder the lesson is double. Pick the side of the relationship you will serve, as Pennylane did when it handed its accounting service to a firm. And decide early how much control you will keep, because the terms you accept in the first round are the ones you live with in the tenth. ## Spendesk (France) URL: https://startup.talk/france/spendesk/ Founders: Rodolphe Ardant, Guilhem Bellion, Jordane Giuly Founded: 2016 (incorporated 1 August 2016); Based in: Paris, Ile-de-France; Sector: Spend management software for companies Rodolphe Ardant has said that the most difficult stretch of his career as a chief executive was the spring of 2020, when offices emptied and every company's spending changed overnight. In an interview that October, he said he was prouder of how the team held together then than of closing the Series B. He was not new to hard stretches. Before Spendesk he founded Wozaik, a digital advertising start-up that Solocal bought in 2013, then served as operating chief at the car-sharing company Drivy. After that he was a founder-in-residence at eFounders, the Paris start-up studio. That is where Spendesk began, in 2016, with Guilhem Bellion and Jordane Giuly. The company was registered on 1 August. The problem they picked was the dull one in every office. Employees paid out of pocket and waited for repayment, while finance staff chased receipts. Ardant wanted B2C payments-style simplicity for businesses, with virtual and physical cards, expense claims and invoices in one tool. He described the aim as building trust and autonomy inside companies. What many people do not know is that one of the three left. Giuly stepped away after about twelve months of conversations with investors around the Series B, reportedly because he did not see himself building the fast-scaling company Spendesk was becoming. It is a candid reminder that founders are not always the same people at 20 employees as at 200. The company did scale. It went from about 20 to 120 people in 2018 and 2019, passed 40,000 users by October 2020, and in January 2022 raised 100 million euros to become France's 26th unicorn. The cost showed: a net loss of about 18 million euros in 2020 on 8 million of revenue, and in September 2024 roughly 150 jobs were cut after heavy losses. The repair has been slow. In March 2025 Spendesk recorded its first profitable month, with annual recurring revenue of 52 million euros. For a founder the useful idea is Ardant's own: pride came from the team's resilience in the worst months, not from the funding headline. A cap table gets you to the starting line; people get you through the winter. ## Voodoo (France) URL: https://startup.talk/france/voodoo/ Founders: Alexandre Yazdi, Laurent Ritter, Gabriel Rivaud Founded: 2013 (incorporated 15 April 2013; predecessor Studio Cadet 2012); Based in: Paris, Ile-de-France; Sector: Mobile games publishing and consumer apps For the first four years, Voodoo barely worked. Gabriel Rivaud, who later ran the games side, has said the company struggled until it changed strategy in 2015. That sentence sits oddly beside what came later: revenue of about 380 million euros in 2020, up from roughly one million in 2016. It began with school friends. Alexandre Yazdi and Laurent Ritter had been friends since lycee. In 2012 they set up Studio Cadet, a small firm building apps for other people. It paid bills, but the plan was always larger. They wanted to be a publisher, the name on the product rather than the contractor behind it. Voodoo was incorporated in Paris on 15 April 2013, with Rivaud there from the start. Then came the lean years. Building apps for clients and publishing your own games are different trades, and the second is a lottery that most players lose. What nobody outside the company saw during that stretch was the grind of trying, and failing, to find a hit. The turn in 2015 was a change of method. Voodoo shifted toward hypercasual games, tiny titles that a player understands in seconds, made with a network of outside studios. Hundreds of partner teams joined, and the company tested ideas and scaled only what worked. In May 2018 Goldman Sachs invested about 200 million dollars, the biggest French tech fundraising since 2015. Tencent took a stake in 2020 at a 1.4 billion dollar valuation, and in July 2021 the Belgian holding GBL paid 266 million euros for 16 percent. Then the market turned. Hypercasual saturated, players left quickly, and Apple's tracking changes squeezed advertising. Voodoo cut its partner studios from around 700 to around 100 and moved up toward richer casual games. In December 2022 France's data regulator fined it three million euros for tracking Apple users who had opted out, a penalty upheld in 2024. The company bought BeReal in 2024 for a reported 500 million euros and reported 778 million dollars in revenue for 2025, up 16 percent. The quiet lesson: four years of near-failure with friends you trust is survivable. Voodoo's fortune came not from a better game but from a better way of choosing games. ## Believe (France) URL: https://startup.talk/france/believe/ Founders: Denis Ladegaillerie, Arnaud Chiaramonti, Nicolas Laclias Founded: 2005 (1 April 2005); Based in: Paris, Ile-de-France; Sector: Digital music distribution and artist services In early 2024 Denis Ladegaillerie did something few founders ever have to do. He sat on both sides of the table. The man who had taken Believe to the Paris stock exchange three years earlier announced that he, with the funds TCV and EQT, wanted to buy the whole company back and take it off the market. The price was EUR 15 a share, valuing Believe at about EUR 1.52 billion. Minority shareholders thought it was too low, and said so loudly. It is an uncomfortable position for a founder: the people he had invited in at the IPO were now telling him his own offer undervalued the thing he had built. To understand why it mattered to him, go back to April 2005, when Ladegaillerie, Arnaud Chiaramonti and Nicolas Laclias started a digital music company in Paris. Their idea was simple and slightly heretical. Independent artists had no good way to get onto the new streaming and download platforms, and the big labels had little reason to help them. Believe would be the pipe. The first years were slow. The company signed the French rapper MC Solaar for distribution in 2008 and, in 2010, its own label, Believe Recordings, signed its first artist. In 2015 it raised 60 million dollars from Ventech, TCV and XAnge and bought TuneCore, the American self-service distributor. A year later it paid EUR 10 million for the French label Naive Records. Growth did not come without bruises. The company was accused of aggressive copyright claims on YouTube, and in 2020 it faced a federal lawsuit in New York alleging large-scale infringement. After the invasion of Ukraine it kept operating in Russia but cut ties with sanctioned artists and labels, and Ladegaillerie said its investment there had been interrupted. The IPO came on 10 June 2021 and raised EUR 300 million. Revenue climbed to about EUR 880 million in 2023 and about EUR 989 million in 2024, though the company was still posting small net losses: EUR 18.1 million in 2023, EUR 2.12 million in 2024. What almost nobody expected was how hard the market would push back. Warner stepped aside, regulators cleared the path, but investors held out. In 2025 the consortium raised its offer first to EUR 15.30 and then to EUR 17.20, a figure an independent expert judged fair. By late June 2025 it held 94.99 percent of the shares, and the listing story was over. The lesson for a first-time founder is a quiet one. Going public does not end your obligations to the people who back you, and getting out can cost more than getting in. Ladegaillerie got his company back, but he paid the price the market insisted on. ## Aircall (France) URL: https://startup.talk/france/aircall/ Founders: Olivier Pailhes, Jonathan Anguelov, Pierre-Baptiste Bechu, Xavier Durand Founded: 2014; Based in: Paris, Ile-de-France; Sector: Cloud business phone systems (SaaS) For a company that sells telephones, Aircall started with an odd problem: nobody in a small business wanted to think about the phone. Landlines meant engineers, contracts and hardware. Sales teams wanted calls to show up inside the software where they already worked. That gap is what four people spotted in Paris in 2014: Olivier Pailhes, Jonathan Anguelov, Pierre-Baptiste Bechu and Xavier Durand. Their product was a cloud phone system for small and mid-sized companies, priced at roughly EUR 30 to EUR 50 a month and plugged into CRM and helpdesk tools. It was the kind of idea that sounds obvious once someone has built it. The biggest choice came in 2017. French software companies often stay home for years. Aircall instead opened in New York to chase American customers, and its revenue tripled that year. It is a decision that many European founders postpone and later regret postponing. Money followed the traction. In 2018 Draper Esprit, NextWorld Capital and Newfund put in USD 29 million. In May 2020, as remote work turned the phone into a distributed tool, a USD 65 million round brought total funding above USD 100 million, with Balderton and eFounders among the backers. Then in June 2021 Goldman Sachs led a USD 120 million round, making Aircall the 16th French unicorn. What the milestones show is a company that kept widening its footprint rather than changing its idea. It added offices in Madrid, Seattle, London and Sydney and, by 2022, annual recurring revenue passed USD 100 million. In 2024 it turned ten and opened an office in San Francisco. It now serves more than 24,000 customers. The next test is the same one facing every phone company: what a call is when software can answer it. In 2026 Aircall bought Vogent, a maker of AI voice agents, in May and Piper AI, a revenue-intelligence and coaching tool, in June. The bet is that the call stays at the centre of a sales team's day, and that the conversation can be understood, scored and sometimes answered by software. There is a lesson for a first-time founder here that has nothing to do with phones. Aircall did not need to reinvent the product to survive. It moved to the biggest market early, listened to where customers already worked, and bought its way into the next shift when the last one matured. ## Lemonway (France) URL: https://startup.talk/france/lemonway/ Founders: Sebastien Burlet, Antoine Orsini, Damien Guermonprez Founded: 2007; Based in: Montreuil, Ile-de-France; Sector: Payment institution and payment infrastructure for marketplaces and crowdfunding The number that nearly sank Lemonway's reputation was 100,000. It was the euro amount above which the company's own systems raised an alert about a suspicious transaction. In 2017 the French banking regulator looked at it and decided it was far too high to catch anything. The penalty was EUR 80,000 and a formal reprimand. The regulator also said the firm had failed to report suspicions in about twenty cases to Tracfin, the financial intelligence unit, including some linked to bitcoin. For a company whose whole business is being trusted with other people's money, it was a public bruise. The company that took it had begun quietly. Lemonway was created in 2007 by Sebastien Burlet, Antoine Orsini and Damien Guermonprez, and for several years it was a small fintech waiting on paperwork. In December 2012 it won its licence as a payment institution, the permit that lets a company hold and move funds for others. That made it useful to a new wave of crowdfunding sites and marketplaces that needed to pay many people without becoming banks themselves. By 2015 it had offices in Spain, Italy and the UK. A year later it was payment intermediary for 200 crowdfunding platforms and 1.3 million payment accounts in Europe, and a Business Insider piece called it an attempt at a European PayPal. In 2017 it counted two million customer accounts. Then came the sanction. Management said afterwards that compliance had become its priority and strengthened the teams doing it. In 2018 Lemonway sold its Dakar subsidiary, which had sold offers for diaspora communities, to refocus on Europe. That year Breega and SpeedInvest led a EUR 10 million Series A, and in 2019 a EUR 25 million round followed, taking total funding to about EUR 50 million. The repair work is visible in the numbers. Lemonway joined the FrenchTech120 in 2020, reported 65 percent growth in 2021, and in 2023 reached break-even on revenue of EUR 32.3 million. For a founder in a regulated business, the lesson is blunt. Growth and compliance are not rivals, but compliance that lags growth gets measured by someone else, and the bill arrives with a headline attached. ## H Company (France) URL: https://startup.talk/france/h-company/ Founders: Charles Kantor, Laurent Sifre, Karl Tuyls, Daan Wierstra, Julien Perolat Founded: 2024; Based in: Paris, Ile-de-France; Sector: Agentic AI (computer-use and web agents) Three months. That is how long it took for a company that had just raised USD 220 million to lose three of its five founders. In August 2024 H announced on LinkedIn that Daan Wierstra, Karl Tuyls and Julien Perolat were leaving, putting the company in the hands of Charles Kantor as CEO and Laurent Sifre as CTO. It was an awkward way to start. H had been unveiled in the spring as a Paris answer to the American AI labs, with a seed round big enough to rival companies many years older and backing from names reported to include Amazon and the Arnault family office. The pitch was frontier-scale research on agents, software that does not just answer a question but operates a browser or a computer to finish a task. The departures were attributed to operational differences. The stake was a large pile of other people's money and a promise to build something worthy of it. The remaining team kept shipping. In November 2024 H launched Runner H, an agent platform, alongside testing and web-browsing agents under the Tester H and Surfer H names. It was a clear signal: the company wanted to be judged on working software, not on its founding drama. But the shake-up was not over. In June 2025 Gautier Cloix, a former head of Palantir France, took over as CEO from Kantor, saying he wanted to steady the ship and scale in the way Palantir had. Other departures followed through the year. In May 2026 Laurent Sifre, the last of the founding pair, stepped down as CTO and took a non-operational role as head of the scientific council. What almost nobody predicted at the time of the round was how little the founding team would be intact by the second anniversary. A founder who raises at that scale in a hot field sets expectations that the company, not the cap table, has to meet. The lesson is a hard one for first-time founders. Money can assemble a brilliant team, but it cannot settle how the team will work together. Agree on who decides what before the cheque clears, because afterwards every disagreement is public. ## Hugging Face (France) URL: https://startup.talk/france/hugging-face/ Founders: Clement Delangue, Julien Chaumond, Thomas Wolf Founded: 2016; Based in: New York, USA (French seat in Paris); Sector: Open-source machine learning platform and model hub The first product was a chatbot for teenagers. It was not meant to change how artificial intelligence gets built. Hugging Face, founded in 2016 by the French entrepreneurs Clement Delangue, Julien Chaumond and Thomas Wolf, began as an app for bored teens to chat with, named after the hugging emoji. The app was the founders' idea. What happened next was not. When they open-sourced the model behind it, developers took notice of the model and ignored the app. The company followed its users: it became a machine-learning platform where people share models and code, and in 2017 it moved to the United States to raise money. Delangue's path to that moment began early. Press profiles describe him as a teenager who was one of the most visible French sellers on eBay at 17, and later a graduate of the ESCP business school who turned down a job at Google. He also co-founded Moodstocks, a Paris image-recognition startup that Google acquired in 2016. Those details come from profiles rather than a long interview I could read in full, so treat them as sketches, not a biography. What almost nobody saw coming in 2016 was how large the pivot would grow. In March 2021 Hugging Face raised a USD 40 million Series B. In May 2022 a Series C valued it at USD 2 billion on revenue of about USD 15 million. It also launched the BigScience workshop in 2021, a collaboration that produced BLOOM, an open multilingual language model. The real prize arrived in August 2023, when Google, Amazon and Nvidia put in USD 235 million at a USD 4.5 billion valuation. The company had effectively become neutral ground in the AI race: rivals with competing models still all pointed their users to its hub. Being the commons has costs. Researchers found more than 1,500 exposed API tokens belonging to big tech firms, a 2024 study reported at least a hundred malicious models, and questions about copyrighted training data keep coming. In 2024 it bought XetHub to improve how it stores huge files, and in April 2025 it acquired the French robotics company Pollen Robotics, a sign it wants to move from software into machines. The lesson for a first-time founder is humble. The product you set out to build may not be the one people want from you. Hugging Face survived by listening to what its users took and ignored, and by being willing to throw its own app away. ## Sigfox (France) URL: https://startup.talk/france/sigfox/ Founders: Ludovic Le Moan, Christophe Fourtet Founded: 2010 (some records say 2009); Based in: Labège (near Toulouse), Occitanie; Sector: Internet of Things / low-power wide-area network (0G) The first diploma Ludovic Le Moan earned was not in software. It was a vocational certificate as a turner-miller, a man who shapes metal on a lathe. He was born in Le Havre in November 1963 and grew up in the Mare-Rouge neighbourhood. His father was an electrician, his mother kept a shop. As a teenager he left the school system. What almost nobody connects to the Sigfox story is that a teacher then pushed him towards a classe préparatoire, and in 1988 he came out of ENSIMAG in Grenoble as an engineer. He moved to Toulouse, wrote software, and ran the local agency of an IT services firm until 2000. At 38 he started Anyware Technologies, a machine-to-machine company, and sold it eight years later to Wavecom. A second venture, a mobile content service called Goojet, ended up absorbed by Scoop.it. Two tries, one exit, one soft landing. Around 2010 he found the idea that would define him. Most connected objects, a water meter or a parking sensor, only need to whisper a few bytes a day. Why pay for a cellular plan? With engineer Christophe Fourtet he set up Sigfox near Toulouse to run a global network built only for that whisper. Money followed. Salesforce, Intel, Samsung, NTT, SK Telecom, Total and Air Liquide backed it, and in November 2016 it was valued at about €600 million. Revenue reached €60 million in 2019. Then the awkward part came out. In 2016 the company had talked of about seven million devices; in early 2018 it acknowledged the real number of connections was about 2.5 million. Cellular networks answered with NB-IoT and LTE-M, LoRa grew as an open rival, and Sigfox's demand for a large cut of operator revenue did not help. In late 2020 it sold its German network. Debt stood near €150 million. In February 2021 Le Moan stepped down as CEO in favour of Jérémy Prince, amid reports of internal criticism of his management. On 26 January 2022 Sigfox entered receivership. On 21 April the Toulouse commercial court chose UnaBiz, a Singapore firm, to take it over for about €25 million, keeping 110 of 174 jobs. Its US affiliates later filed for bankruptcy as well. A man who started on a lathe built a network that reached across continents, and a company that never turned a profit. The numbers he could count by contract and the numbers he could count by use were not the same, and the gap between them was the company's real balance sheet. Founders who measure the wrong thing can be sincere and still be wrong. ## Withings (France) URL: https://startup.talk/france/withings/ Founders: Éric Carreel, Cédric Hutchings, Frédéric Potter Founded: 2008 (June); Based in: Issy-les-Moulineaux, Île-de-France; Sector: Connected health devices (scales, watches, blood pressure, urine analysis) On 31 May 2018, Éric Carreel got back a company he had already sold. He had been born in Amiens in 1959 and trained as an engineer at ESPCI in Paris, graduating in 1984. After a stretch in radio research he founded Inventel in 1990 with the physicist Jacques Lewiner. Inventel built the Livebox for France Télécom, and Thomson bought it in 2005. He could have stopped there. Instead, in June 2008, he co-founded Withings with Cédric Hutchings and Frédéric Potter. The first product was a Wi-Fi body scale in 2009, billed as the first connected scale. A blood pressure monitor came in 2011, a baby scale in 2012. Ventech put in $3.8 million in 2010, and a $30 million round in 2013 brought Bpifrance, Idinvest and 360 Capital to the table. Then came the offer that looked like victory. Nokia agreed in April 2016 to buy Withings, reportedly for about $192 million. Hutchings moved to Silicon Valley to run Nokia Digital Health. The Withings name was phased out; by June 2017 it was gone. What happened next was costly. Nokia wrote down $164 million, almost the entire purchase price, in October 2017. An internal memo, reported at the time, said the business had struggled to scale. Nokia wanted to return to its telecom roots and began a strategic review. Carreel stepped in. In May 2018 he bought the health business back and relaunched Withings in September. The price was never disclosed. About 200 people came with it, mostly at the Issy-les-Moulineaux headquarters. Hutchings did not join; Carreel said he was working on another project. Asked by VentureBeat what had gone wrong, he declined: he had no comment on what was in the past. The second life was quieter and more medical. Body Scan won FDA clearance in 2023, a urine-testing device called U-Scan followed, and the company bought Impeto Medical and the fitness app 8fit in 2022. Maddyness reports the company reached profit in 2025, seven years after the restart. Not every move has pleased users: from October 2026, new accounts need a paid Withings+ subscription for automatic sync to outside apps, and customers complained. For founders, the lesson is in the order of events. The exit was not the end of the story; the buyer's retreat became the founder's second chance. He did not argue about why it failed, he went and fetched it back. ## Ynsect (France) URL: https://startup.talk/france/ynsect/ Founders: Antoine Hubert, Jean-Gabriel Levon, Fabrice Berro, Alexis Angot Founded: 2011; Based in: Évry, Île-de-France; Sector: Insect farming: proteins for animal feed, fertilizer Antoine Hubert did not start with insects. He started with soil. As an agronomy student, first in Rennes and then in Paris, he became fascinated by the ecology of the ground under our feet. He worked on organic farming, composting and waste processing in non-profit work, building toolkits and partnerships. Nobody pays well for that, and he knew it. What almost nobody knew was how the idea for Ynsect arrived. It was not a eureka moment on a farm. He took an existing study from the UN Food and Agriculture Organization on insect farming, a paper explaining that insects could meet the rising demand for protein, and decided somebody should build it at scale. In 2011 he founded Ynsect with Jean-Gabriel Levon, Fabrice Berro and Alexis Angot. The plan was industrial from day one: grow beetle larvae in tall vertical farms, extract protein for animal feed and fish farms, sell the leftovers as fertiliser. The first farm opened in Dole in 2015. In 2018 a much larger site was announced at Poulainville, near Amiens. A fertiliser authorisation came in 2020, and in 2021 Ynsect bought the Dutch company Protifarm. Investors came, famously including Robert Downey Jr.'s fund alongside Astanor. In 2023 a Series D brought in €160 million. Over its life Ynsect raised about €600 million. Hubert stepped aside as CEO in July 2023. Then the arithmetic turned. Costs doubled between 2022 and 2023. Insect meal costs two to ten times as much as fishmeal or soy, and shoppers were not buying insect food in numbers anyone could depend on. The company focused on animal feed. In September 2024 it entered safeguard proceedings; in March 2025, judicial reorganisation. In June 2025 almost 70% of its 200 jobs went. In December 2025 a court ordered liquidation. The public bill drew political attention. The economy ministry reportedly said about €148 million in public funds had gone in, and one politician called it public money thrown out of the window. A student who loved soil built a factory that grew insects, and ended up unable to make them cheap enough. Founders in climate technology should remember that a good thesis is not a price. The product must be cheaper than the thing it replaces, not just better for the planet. ## Exotec (France) URL: https://startup.talk/france/exotec/ Founders: Romain Moulin, Renaud Heitz Founded: 2015 (June); Based in: Wasquehal (Lille area), Hauts-de-France; Sector: Warehouse robotics and automation (Skypod) Romain Moulin did not begin in robotics glory. He began as a sales engineer. His first employer was a small French company that built automated forklifts. His job was to sell them. He later moved up to technical manager, then joined GE Healthcare, where he worked on robotic imaging machines and learned how to ship hardware. At GE he worked with Renaud Heitz, whom he had known for more than twenty years. In an interview he describes himself as someone with a strong passion for hardware, with Heitz leading the software. What pushed them was a market fact. Amazon had bought Kiva, the robot warehouse system, and kept it for itself. Every other retailer was stuck with rigid systems that took months to install. In June 2015 Moulin and Heitz launched Exotec in the Lille area to build a robot fleet for everyone else. The early money was small. Business angels put in about €340,000 in year one. They prototyped something called Exo, then the Skypod, a robot that drives on the floor and climbs racks up to 14 metres. In December 2016, 360 Capital and Breega backed them with about €3.3 million. Cdiscount agreed to work with them on design, and the first Skypod system went into a Bordeaux warehouse in early 2017. After that the curve bent. A €15 million round followed in 2018, €77 million in 2020, and in January 2022 Goldman Sachs Asset Management led $335 million at a $2 billion valuation, France's first industrial unicorn. A US headquarters opened in Atlanta in 2021, and Japan became a key market, with Uniqlo among clients. Exotec says it reached $1 billion in worldwide sales by 2024, with more than 200 client sites and 1,300 staff. Revenue in 2025 was reported at €340 million. In early 2026 it opened a 25,000 square metre headquarters in Wasquehal. At one E.Leclerc site in Seclin, orders per day rose by 50% and customer wait times fell by 73%. For founders, the quiet lesson is in the pair. A salesperson who had to explain automated forklifts to sceptical buyers, and a software architect he had trusted for two decades, built a product that customers could actually install. Know what buyers dislike about the current thing before you build the next one. ## PrestaShop (France) URL: https://startup.talk/france/prestashop/ Founders: Igor Schlumberger, Bruno Lévêque Founded: 2007 (project started 2005); Based in: Paris, Île-de-France; Sector: Open-source e-commerce platform Before it had a brand, it had a placeholder name: phpOpenStore. It started as a student project at Epitech, the Paris computer science school, with five students named in one account: Igor Schlumberger, Bruno Lévêque, Denis Roussel, Marc Rossi and Laurent Basso. The common thread is the aim: a free, open-source alternative to expensive proprietary shop software, available in French and English from the start. The name changed. Presto is the Italian word for quick; shop needs no translation. PrestaShop was born, and in 2007 Schlumberger and Lévêque created the company in Paris to build a business around it. The model was unusual. The software was free software from conception. Anyone could download it. Money came from add-ons: modules, themes and services sold in a marketplace. That meant the company was betting that merchants would build their shops on its code first and pay later. It worked, slowly and then fast. Between 2010 and 2012 the team grew from 17 to over 100. A second headquarters opened in Miami in 2011. By then the platform counted 100,000 merchants. In March 2014 the company raised about €6.75 million, reported abroad as a $9.3 million Series B. Open source brings its own pain. Version 1.7, the first stable release of the new branch, only arrived in 2016, and the company's free cloud offer from 2015 was gone by 2017. Alexandre Eruimy became CEO in 2018. The Miami office closed in 2019. Meanwhile Shopify and WooCommerce took attention. Even so, a 2019 BuiltWith report gave PrestaShop 32% of French online shops. In November 2021 PrestaShop joined the MBE Worldwide group, which later renamed itself Fortidia. Version 8 shipped in 2022 and version 9 in June 2025. The company says it joined the European group cyber_Folks in 2026. For founders, the lesson is in the order. Give the code away first, build a community of thousands of merchants and developers, then sell what they need next. The students who wrote the first lines could not know which of those merchants would stay. Many did. ## Shift Technology (France) URL: https://startup.talk/france/shift-technology/ Founders: Jérémy Jawish, Eric Sibony, David Durrleman Founded: 2014 (incorporation recorded as March 2013 in French registry data); Based in: Paris, Île-de-France; Sector: Insurtech / AI fraud detection and claims automation (SaaS) It started, in one telling, with an intern. Jérémy Jawish was working on mathematical models to catch insurance fraud inside an insurance company. Eric Sibony was an intern on much the same territory. Out of that meeting, as startup profiles describe it, came a company that would one day be valued above a billion dollars. Neither man's schooling or hometown is on the public record we found, so the story begins at the desk, not the family table. What they saw was a gap. Insurers checked suspicious claims largely by hand, and the startup later argued that only about 15% of fraud cases were ever detected. A claims handler cannot read two million files. A model can. So in 2014 they set up Shift Technology in Paris. The early years were the unglamorous part: selling software to a cautious industry that guards its claims data and does not like experiments. The first outside money came in 2016, a $10 million Series A led by Accel with Elaia Partners and Iris Capital. A year later came $28 million from Accel and General Catalyst. Investors were betting that insurers would eventually have to trust a machine to flag a lie. They did. In 2019 Bessemer led a $60 million round, and the company joined the government's Next40 list of French companies tipped to become global leaders. Then came the jump: in May 2021, Advent International led a $220 million Series D at a valuation above $1 billion, making Shift the thirteenth French licorne. By then it counted about 100 insurance customers in 25 countries and said it had analysed nearly two billion claims. Offices sat in Paris and Boston. The product had widened by then. Fraud detection was the doorway; claims automation, underwriting, subrogation and financial-crime tools came through it. A partnership with Sapiens followed in 2022, and the company recruited former US health regulator Seema Verma to an advisory board. Guidewire, the insurance software company, later invested to extend the Series D. Revenue has never been published. There is an odd modesty to the story. Shift did not invent a new market. It took a boring, expensive, well-understood problem, claims leakage, and applied a better tool. The first meeting was between a modeller and an intern, not a celebrity founder and a famous investor. For a first-time founder the lesson is quiet. The best idea is sometimes already sitting on the desk of the person doing the dull work, and the person next to them may be the co-founder. ## Meero (France) URL: https://startup.talk/france/meero/ Founders: Thomas Rebaud Founded: 2016; Based in: Paris, Île-de-France; Sector: On-demand photography marketplace turned AI photo-management SaaS In June 2019 Thomas Rebaud announced a number that made French tech sit up: $230 million for a photography platform. Headlines called it a unicorn. Rebaud explained the size of the cheque with a line that would age badly. If you want to make it happen quickly, he said, you have to invest in many things at the same time. Meero's idea was simple enough. Businesses needed photos, from estate agents to restaurants to event organisers. Freelance photographers wanted work. Meero connected them and used AI to sort and edit the pictures so delivery was fast. Early money came in 2017, including a round that drew Xavier Niel and backers tied to Bernard Arnault, then $45 million in 2018. The big round turned the company into a hiring machine. Meero grew to roughly 600 people. Everything depended on one assumption: that companies would keep ordering shoots. What almost nobody knew in 2019 was how much of that headline money had really arrived. In January 2022, Maddyness, reading public documents, concluded that Meero had received only about €130 million. That would put the valuation below a billion dollars, and the publication left Meero out of its unicorn count. The company that launched with a billion-dollar label never quite owned it. Then the world stopped travelling. Events were cancelled, shops closed, marketing budgets froze. Photo shoots are physical, and the model could not shrink quickly. Headcount fell from about 600 to 350, and then 72 further jobs went. Rebaud stepped down as CEO, keeping his shares and moving to chairman; Gaétan Rougevin-Baville, the COO and one of the first employees, took over. He told Les Echos that visual content budgets had been among the first to be frozen. The company tried to sell its photographer service and, according to reports, found no buyers. By February 2023 clients could no longer book shoots. Meero pivoted to software for storing and managing photos, the AI tooling it had built for its own marketplace. Rebaud's own history, his school and his hometown, is not in the public record we found, so Meero's story is told through its numbers and its choices. The lesson for a founder is uncomfortable. A very large round is a promise to grow fast, and fast growth built on a physical marketplace is hard to reverse. The smart part of Meero, the software, outlived the part that burned the cash. ## PhotoRoom (France) URL: https://startup.talk/france/photoroom/ Founders: Matthieu Rouif, Eliot Andres Founded: 2019-2020; Based in: Paris, Île-de-France; Sector: AI photo editing and generative image app In January 2024 a rumour about PhotoRoom ran ahead of the company. TechCrunch reported that the Paris app was raising $50 to $60 million at a valuation of up to $600 million. The founders would not comment. A month later the real figure was out: $43 million at $500 million, a little lower than the whispers, confirmed by CEO Matthieu Rouif. It was a striking moment for a company only about four years old, with roughly 50 employees. Rouif is the CEO. Eliot Andres, his co-founder, is the CTO. Their schools, hometowns and earlier jobs are not in the interviews we found, so the story has to be told through what they built. What they built was a phone app that did one annoying job well: cut an object out of its background. People selling on online marketplaces, small shops and casual users all needed exactly that. The product grew by being useful rather than clever. By early 2024 the app had passed 150 million downloads and, the company said, was processing about 5 billion images a year. It also sold access through an API and a web editor, which let other businesses bolt it onto their own tools. The money followed. A Series A of about $19 million arrived in 2022, with early backers including Kima Ventures, FJ Labs and Meta. Balderton Capital led the 2024 Series B, with Aglaé Ventures joining and Y Combinator participating. Total funding stood near $64 million. The part founders should notice is what they spent it on. Rather than rent someone else's image model, PhotoRoom said it trained its own from scratch, and claimed its custom design made generation up to 40% faster than rival visual AI tools. Launched alongside the round was Instant Diffusion, a tool for placing products in generated scenes. In May 2025 it bought a smaller company, Generate Banners, to add ad creation. None of this is a story of collapse or courtroom. It is a story of narrow focus. Competitors are far larger: Adobe, Canva and others have the distribution. PhotoRoom's bet is that a tool built for one painful job can stay ahead of a suite built for everything. The takeaway for a founder is plain. A single repeated task, done a million times a day, can fund the harder research later. ## Heetch (France) URL: https://startup.talk/france/heetch/ Founders: Teddy Pellerin, Mathieu Jacob Founded: 2013; Based in: Paris, Île-de-France; Sector: Ride-hailing / VTC mobility app The prosecutor did not like how relaxed they were. In December 2016 Teddy Pellerin and Mathieu Jacob stood in a Paris criminal court. More than 1,500 taxi drivers had joined the case as civil parties. The prosecutor asked for a €10,000 fine and a two-year ban on running any company for each founder, and a €300,000 fine for Heetch itself. She said the pair seemed too décontractés given what had happened to their drivers, many of whom had been taken into police custody and fined. Heetch had started in 2013 with a modest promise: rides home at night for people in their early twenties. It worked only from the evening until dawn, and by Pellerin's own account its customers were mostly students. Two-thirds of rides began in the Paris suburbs, where night buses thin out. The name came from hitch-hiking, and the company cultivated a pirate image. The first small rounds, €1.5 million in 2014 and again in 2015, came from backers including Kima Ventures, TheFamily and Alven. By 2015 it had 51 employees and about €2 million in revenue, slightly loss-making. Then the state moved. In the summer of 2015, Heetch, UberPop and Djump were named in a ban on apps that matched private drivers with riders. Uber and Djump stopped. Heetch kept running. Its drivers paid for it with custody and fines, while the founders launched a campaign, touche pas à mon Heetch, asking riders to speak up. In March 2017 the court ruled. Heetch and its two directors were convicted of complicity in illegal taxi activity, misleading commercial practices and illegal matching of clients with non-professional drivers. Damages ran to about €441,000, with fines on top, and each director paid €10,000, half suspended. The original service was suspended in France. Heetch relaunched with professional drivers, available at any hour. Investors stayed: €16.5 million in 2018 and about $38 million in 2019, led by Cathay Innovation and Total Ventures. The company pushed into Africa, where an Abidjan moto-taxi launch in 2019 ran into another ban and impounded vehicles. In 2022 it became the first VTC platform to take entreprise à mission status. Pellerin and Jacob's schools and early careers are not on the public record we found. For a founder, the case is a hard one. Defiance bought Heetch the market and a criminal record. The company survived only by becoming the thing it had refused to be: a business that fits the rules. ## Kyriba (France) URL: https://startup.talk/france/kyriba/ Founders: Jean-Luc Robert Founded: 2000; Based in: San Diego, California (R&D in Saint-Cloud, Île-de-France); Sector: Treasury, risk and working-capital software (SaaS) Most people who have heard of Kyriba assume it is American. The head office is in San Diego, the customers are global, and the chief executive since about 2023 is Melissa Di Donato, who replaced the founder. But the research and development teams remain in Saint-Cloud, west of Paris. That split is the quiet story of Jean-Luc Robert and the company he started in 2000. His childhood and schooling are not in the public record we found, so the story starts with the product: software for corporate treasurers. These are the people who decide where a company's cash sits, which currencies it holds and whether it can pay next week's bills. In the early 2000s, many did that on spreadsheets. Kyriba sold them a cloud service. Its revenue, according to its early investor Iris Capital, came mainly from France in the first years. Iris made its first investment in 2010, in the Series C, and came back several times, including in 2017. Something happened in those years that Iris calls a technological pivot, though the details remain private. What is certain is the result. Since 2010, Iris says, revenue grew 25 times and annual recurring revenue 30 times. The company expanded into the United States and Europe, added risk management, payments and working-capital products, completed its first acquisition and became profitable. In 2019 Bridgepoint took a majority stake in a deal that planned $160 million of growth capital and valued Kyriba at about $1.2 billion. Robert stayed on as an investor alongside Daher Capital and Iris. Iris sold its 2010 fund position then; its 2017 IrisNext investment has since returned more than ten times. The next transaction was bigger. In October 2024 Bridgepoint reinvested as majority owner and General Atlantic joined as a minority investor, in a deal reported to put in about $400 million of equity for a stake near 15% and value the business above $3 billion. Iris exited fully, saying the company's value in euros had grown more than a hundredfold. Robert handed over most of his shares and kept a board seat. The company now says it has more than 3,000 customers. For a founder, the pattern is patience. Kyriba took more than two decades, three generations of investors and one pivot to reach a multibillion-dollar price, and it kept its French engineering core the whole way. ## Mooncard (France) URL: https://startup.talk/france/mooncard/ Founders: Tristan Leteurtre, Damien Metzger Founded: 2016; Based in: Paris, Île-de-France; Sector: Fintech / spend management and business payment cards Every finance team knows the monthly ritual. Someone has paid for a train ticket, a client lunch, a software subscription, and now a shoebox of receipts has to be matched to a spreadsheet. Tristan Leteurtre and Damien Metzger built a company around that small, dull misery. Mooncard was founded in 2016 in Paris. The pitch was plain: give every employee a company card, let them snap a photo of the receipt on their phone, and have the accounting entry appear on its own. No advances out of pocket. No reimbursement forms. For a ten-person agency or a regional retailer, the saving was hours of tedious admin each month. The early years were the hard part. A card is a trust product. A finance director has to believe that a young company will not lose track of the money, and that employees will really photograph their receipts. Mooncard spent its first years proving both, one SME at a time. In January 2019 it announced a round of about five million euros to speed up the roll-out, and Leteurtre was by then describing himself as co-founder and CEO of a company with a clear product and thin margins for error. The reveal, if there is one, is how far a boring tool can travel. SMEs were the first users. Then the buyers got bigger and stranger. The company's own history now lists the French State's purchasing directorate as a client, alongside names like Air France, Indigo, Norauto, Vinci Airports, Loxam and Médecins sans frontières. A tool built to save a small accountant's evenings was now processing the spending of public bodies and national brands. Money followed in steps. About twenty million euros came in 2021 from existing backers such as Aglaé Ventures and Raise Ventures, as Mooncard pushed into other European countries. In 2023 it raised 37 million euros, bringing in Orange Ventures and the Canadian fund Portage, and used part of the money to add a feature that measures the carbon footprint of card spending. The company says it now serves about 5,000 client companies, with 200,000 cards in circulation across seven countries and roughly three billion euros of employee spending paid through them. It has not published a revenue figure, and leadership has since passed in the press to Pierre-Yves Roizot as chief executive, while Leteurtre is still presented as the founder. The lesson for a first-time founder is unglamorous. Mooncard did not invent a new kind of money. It picked one irritation that repeats every month in thousands of offices, and it kept sanding it down until even a government department was willing to sign. ## Matera (France) URL: https://startup.talk/france/matera/ Founders: Raphaël Di Meglio, Victor Prigent, Jérémy Krebs Founded: 2017 (incorporated 17 January 2017 as illiCopro); Based in: Paris, Île-de-France; Sector: Proptech / cooperative property management (syndic) A man buys an apartment in Paris, opens the folder from the building's property manager, and understands almost none of it. The charges, the meeting minutes, the invoices for works nobody explained. He is an owner, and he feels like a guest. The story goes that this was how one of Matera's co-founders met the problem, and it became the seed of a company. The company began as illiCopro, incorporated on 17 January 2017. Raphaël Di Meglio was the chief executive, Victor Prigent ran product, Jérémy Krebs ran technology. Their idea was to let a building's own owners run it, with a cooperative syndic model, and to hire experts for the hard parts such as accounting and law. For owners it promised charges that were 20 to 50 percent lower, a claim the company itself made. The early money was small. In January 2019 they raised 1.5 million euros. A year later, in January 2020, they raised ten million from Index Ventures and the Spanish fund Samaipata and changed the name to Matera, after the Italian town of houses carved into rock. The new name was meant to carry them into Spain and Italy. They then had about 800 buildings and 25,000 co-owners on the platform. What almost nobody outside the property trade saw coming was the price of a poster. In March 2020 Matera ran a campaign that mocked traditional syndics. It worked as marketing and as a provocation. Professionals complained, and trade bodies took the start-up to the Paris commercial court. In January 2022 the court found clear disparagement and ordered 70,000 euros in damages. Two years later the Court of Appeal threw out the claims that Matera was practising the profession illegally or misleading customers, but still convicted it of disparagement, while cutting the bill by 40,000 euros. Growth kept coming at a cost. Revenue reached 11.5 million euros in 2023 with a loss of 7.7 million, then 15.7 million in 2024 with a loss of 1.16 million, a sharp narrowing in one year. In 2023 and 2024 the company added rental management and a traditional syndic offer, a quiet admission that not every building wants to run itself. The lesson for a founder is double-edged. Anger at an incumbent can be the best brand you will ever have, and also the most expensive. Matera's founders discovered that you can be right about the model and still pay for the way you said it. ## Younited (France) URL: https://startup.talk/france/younited/ Founders: Charles Egly, Geoffroy Guigou, Thomas Beylot Founded: 2009 (opened to the public December 2011 as Prêt d'Union); Based in: Paris, Île-de-France; Sector: Fintech / consumer credit and instalment payments Charles Egly worked at a bank. When he needed a consumer loan, he went to his own employer, and the rate it offered him was above ten percent. He later told MoneyVox that everything started from a personal experience. A banker, shown the price of his own product, did not like what he saw. He knew something the average borrower did not. Peer-to-peer lending, in which individuals lend directly to each other, was already established in the United States and Britain. In France it did not exist. Egly and his co-founder Geoffroy Guigou, a former McKinsey and Poweo man, founded the company in 2009. Thomas Beylot is also named as a co-founder in several articles. The first years were slow and regulated. Money raised in 2010 and 2011, nine million euros in all, paid for the licence work. In 2011 the firm was approved as a credit institution, and it opened to the public that December as Prêt d'Union, the first French platform of its kind. By September 2015 it had around 27,000 clients, 210 million euros of loans and about 90 staff. The turn came in 2016. Egly and Guigou renamed the company Younited Credit, opened it to institutional lenders and went to Italy. The reason was scale: matching individual lenders to borrowers was a charming idea but a small market. Spain followed in 2017, Portugal in 2018, Germany in 2020. By January 2019 it had financed one billion euros in credit, and in 2022 it had granted three billion. The prize arrived in December 2022, when a sixty-million-euro round made it France's 28th unicorn. Revenue had been about 125 million euros in 2021. Then came the price of timing. When Younited listed on Euronext Paris on 20 January 2025, its valuation was far below the unicorn peak, at about 622 million euros. Egly was still the chief executive. In May 2025 the company said it would acquire the French neobank Helios through a share exchange. Founders often hear that a company is a solution to a problem they personally had. Egly's version is more specific: he was the problem's insider, with a bank badge and an unfair rate. That is a good test for any idea. Does it come from an irritation you have felt yourself, in your own wallet? ## Agicap (France) URL: https://startup.talk/france/agicap/ Founders: Sébastien Beyet, Clément Mauguet, Lucas Bertola Founded: 2016; Based in: Lyon, Auvergne-Rhône-Alpes; Sector: Fintech / SaaS cash-flow and treasury management Ask a small-business owner how much cash they will have in three months and many will open a spreadsheet. It is usually out of date, and somebody built it at night. Agicap was founded in Lyon in 2016 to replace that spreadsheet. The founders were Sébastien Beyet, Clément Mauguet and Lucas Bertola. Beyet had been a senior strategy consultant at Ares & Company. Mauguet came from finance and consulting and graduated from EM Lyon in 2013. They were not building a consumer app. They were building a tool for the finance director who must know on Monday if payroll is safe on Friday. The early years were modest and slow. The product linked bank accounts and accounting data into a single view of cash coming in and going out, and a forecast on top. It was sold as a subscription to SMEs in France first. In 2019 the company took in 2.4 million euros from BlackFin Capital and Kima Ventures. The following year Partech put in fifteen million. That is a small amount for a software company with ambitions, but it was enough to build the product and open in Germany, Spain, Italy and the Netherlands. The turn came in 2021. Greenoaks Capital, a California fund, invested 82 million euros, about 100 million dollars. For a Lyon software company, that was an unusual cheque from an unusual investor. The money bought speed. In 2022 Agicap reported 100 percent growth and announced plans to hire 600 people, 300 of them in Lyon. It was selected for the French Tech 120 programme in 2022 and again in 2023, and an investment bank listed it among 50 companies that could become unicorns within two years. The price of that speed arrived later. In November 2024 Agicap raised a Series C of about 45 million euros, with AXA Venture Partners and Accel among the investors, to push into mid-sized companies across Europe. It was a smaller cheque than the 2021 round and came with a sharper focus. The company has not published revenue. What it has published is a direction: from small firms and their spreadsheets to larger finance teams that need cash visibility across borders. The lesson for a founder is that a boring, urgent question can carry a company a long way. Agicap did not claim to reinvent finance. It answered one number, how much cash will we have, and kept moving up the ladder of customers who needed the answer. ## Malt (France) URL: https://startup.talk/france/malt/ Founders: Vincent Huguet, Hugo Lassiège, Jean-Baptiste Lemée Founded: 2013; Based in: Paris, Île-de-France; Sector: HR tech / freelance marketplace Before it was a marketplace worth hundreds of millions in annual business, Malt was a drinks evening for freelancers. The company was called Hopwork then. It held its apéros freelance, small meetups for people who worked alone, and it said openly that it was copying Airbnb: build a community first and the transactions would follow. Vincent Huguet was the chief executive. With Hugo Lassiège and Jean-Baptiste Lemée he started Hopwork in 2013. Huguet had already co-founded Dromadaire, an online greeting-card company, and Ooprint, a printing business, so he knew what it meant to build a web business from nothing. This time the product was people. The early years were a bet on trust. A freelancer needs a reason to believe that a platform will bring work. A client needs a reason to believe that a stranger on a screen is competent. The founders chose rules that irritated some freelancers and reassured buyers: no one could set a day rate below 125 euros, to avoid a race to the bottom. They raised five million euros in December 2016, partnered with Bpifrance in September 2017 and renamed the company Malt that November. The turn was the large company. Huguet told Maddyness in March 2025 that a decade earlier big firms did not really know what a freelancer was. Malt helped them learn. By February 2019 it said it had 100,000 freelancers and 15,000 corporate clients, and raised 25 million euros. In June 2021 it raised 80 million from Goldman Sachs Growth Equity, Eurazeo and its early backers ISAI and Serena. By January 2022 it claimed 300,000 freelancers, and a month later it joined the Next40. The price was competition and a change of cast. Alexandre Fretti, who had run the French business beside Huguet since 2020, left in 2024. Rivals such as Crème de la Crème and Comet crowded in, and Germany, its second market, proved slow and complex. Huguet said the company needed no new money unless it entered the United States, and that an IPO was not on his agenda. The numbers he gave for 2024 were an 800 million euro business volume, up 30 percent, and a promise of group-level profit in 2025. Malt does not publish revenue, only that volume, and takes about ten percent as commission. The lesson is old and easily forgotten. Malt did not start with a marketplace. It started with a room of people who worked alone and wanted to meet. The transactions came afterwards, because the community was already there. ## Lengow (France) URL: https://startup.talk/france/lengow/ Founders: Jérémie Peiro, Mickaël Froger Founded: 2009; Based in: Nantes, Pays de la Loire; Sector: E-commerce software (product feed and marketplace automation) Before it had a logo or a salesperson, the company had a word. Lengo, from Swahili: objective. Jérémie Peiro and Mickaël Froger added a W for the web and, on 1 July 2009, registered Lengow in Nantes, a city better known for shipyards and bicycles than for software. The objective they chose was unglamorous. Every online shop that wanted to sell on a price-comparison site or marketplace had to send its catalogue in a different format, with different rules, again and again. Mistakes meant rejected listings and lost sales. Peiro and Froger built a tool that took one catalogue and spoke every language. Money came in small pieces. In April 2010 they raised €210,000. In July 2011 came €1.2 million, with Alven Capital among the backers. Neither sum was enough to waste. Nantes was a long way from the Paris investors, and the founders learned to grow on modest cheques while merchants slowly moved from one sales channel to many. Then the market caught up with them. As Amazon, Google Shopping and a thicket of regional marketplaces became part of every retailer's plan, feed management stopped being a back-office chore and became a growth lever. In 2012 Lengow won an international e-commerce award. In 2013 it launched its own Ecommerce Day, an annual event that turned a small vendor into a name among French retailers. In September 2015 the company raised €10 million and reached for the United States and China. It reframed itself from a feed tool into an e-commerce automation platform, and in 2016 it joined an accelerator programme in China and opened an incubator of its own. Expansion was not painless. A 2013 press report claimed the company was investing in around fifteen European countries to stifle competition, a charge that sat awkwardly with its friendly image. In early 2021 the American private-equity firm Marlin Equity Partners took a majority stake, and in April 2022 Lengow bought Netrivals, a pricing and market-analysis tool, to widen what it sold. The numbers since show the strain of a maturing niche: revenue of €16.7 million in 2023 with a €10.8 million net loss, then €15.5 million in 2024 with a €1.77 million profit. The lesson is not about hockey-stick growth. Lengow found a boring, repeated pain, solved it for thousands of merchants, and survived long enough to be worth buying. For a first-time founder, the plain objective, the lengo, can be enough to build a company on, provided you are still standing when the market finally needs it. ## Veepee (France) URL: https://startup.talk/france/veepee/ Founders: Jacques-Antoine Granjon, Julien Sorbac, Mickaël Benabou Founded: 2001 (vente-privee.com, 30 January 2001; first company Cofotex 1985); Based in: La Plaine Saint-Denis, Île-de-France; Sector: Flash-sales e-commerce In 1996 a stock-clearance dealer bought a newspaper's old factory. Jacques-Antoine Granjon, then in his mid-thirties, paid for the former printing works of Le Monde in Plaine-Saint-Denis, north of Paris, a cavernous building that had once turned out the country's most serious daily. He would use it to store other people's unsold clothes. Granjon was born in Marseille in 1962 and raised in Paris, where he went to the Lycée Saint-Louis-de-Gonzague before the European Business School. At EBS he met Julien Sorbac. In 1985, the year Granjon graduated, the two friends started a wholesale company, Cofotex, reselling surplus stock. It was a trade of back rooms, phone calls and cash, and it lasted close to twenty years. What almost nobody saw coming was that the dull trade would become the foundation of an internet company. Brands hated clearance sales; discounting in public cheapened them. Granjon knew that from the inside. In early 2000, with a quarter-century of destocking behind him, he and Sorbac conceived a private online sale: members only, a few days, then gone. Vente-privee.com opened on 30 January 2001, with seven associates alongside him. The first years were slow. The site was an investment, not a hit. The turn came in 2004, with a sale for a well-known lingerie brand that showed labels could trust the format and shoppers would queue for it. By 2007 the American fund Summit Partners had bought 20 percent. Granjon, known for long hair and a casual manner, told one interviewer that long and clean beats short and dirty. Growth brought noise. In 2013 a Paris court ruled the vente-privee.com trademark partly void, a decision reversed on appeal in 2015. Between 2016 and 2018 the consumer authority investigated the firm over reference prices, a case the company contests. A 2011 American venture with American Express never took root. Still, the machine kept turning. Acquisitions followed: two Paris theatres, a Belgian site, and in 2016 the Spanish rival Privalia. In 2019 the name became Veepee, and in 2024 the group reported €3.2 billion in sales including VAT. Challenges estimated Granjon's fortune at about €1.3 billion in July 2026. The quiet lesson sits in that old printing plant. Granjon did not abandon his unglamorous first business to chase a new idea. He used what it taught him, the pain of brands and the habits of bargain hunters, as the raw material. The best opportunity may be the one you have been standing beside for twenty years. ## Platform.sh (France) URL: https://startup.talk/france/platform-sh/ Founders: Fred Plais, Ori Pekelman Founded: 2009 (as Commerce Guys); Platform.sh spun out 2015; Based in: Paris, Île-de-France; Sector: Cloud application hosting and developer platform (PaaS) A company that changes its name twice is usually telling you something. This one began as Commerce Guys in 2009, became Platform.sh in 2015, and in 2025 became Upsun. Each name marks a different idea of what the business was for. Commerce Guys started where many French web businesses of the period did, building online shops on open-source tools. The pain it saw there was familiar to any developer. Getting a new feature from a laptop to a live website meant a tangle of servers, configuration files and fragile staging copies that never quite matched production. Projects slipped, and engineers spent more time keeping environments alive than writing code. In 2015 the team stopped selling shop-building services and spun out the tool it had built for itself. Fred Plais, who had led the original company, became chief executive of Platform.sh, with Ori Pekelman as co-founder and strategy chief. The pitch was simple to say and hard to build: every branch of your code gets its own full copy of the live environment, data and services included, and you can throw it away when you are done. That idea found buyers. In May 2018 the company announced a $34 million Series C, large for a Paris infrastructure company that was not selling to consumers. In June 2022 it raised a $140 million Series D, taking disclosed equity funding past $170 million. The company now says it has raised more than $180 million and serves over 16,000 developers, with Adobe and UNICEF among its customers. The harder test was positioning. Platform.sh had become known as a hosting service, a crowded field where American names such as Heroku, Vercel and Netlify spent heavily on marketing. In 2025 the company chose a fresh start, rebranding as Upsun and recasting the product as a platform that gives both human developers and AI tools the exact real-world context they need to deploy reliably. The rebrand is a bet that the next wave of software will be written partly by machines, and machines need a faithful copy of the world to test against. Whether the bet pays off is not yet known. The human insight is in the repetition. Plais and his team did not start with a grand theory of cloud computing. They started with an annoying job they hated doing for clients, built a tool to end it, and kept renaming and reshaping the company as the problem grew larger. For a founder, an irritation you cannot stop thinking about is often the product. ## Lemlist (France) URL: https://startup.talk/france/lemlist/ Founders: Guillaume Moubeche Founded: 2018; Based in: Paris, Île-de-France; Sector: Sales engagement and cold-email software (SaaS) The money he put into the product was $1,000. Not a seed round, not a family loan: Guillaume Moubeche's own savings, spent in 2018 on a piece of software that would drop a personalised image into a sales email. He came from Paris, but not from money. His parents grew up on a farm, never earned degrees, and, he has said, made many sacrifices so he and his brother could have a good education. At university he took whatever work he could find, babysitting and bartending among it. He told an interviewer he struggled with authority and with outdated ways of doing things, and made himself a promise to become his own boss. He began with a lead-generation agency, sending cold emails for clients with a friend. The work taught him one thing quickly: generic messages got ignored. Replies came when an email looked written for one person. That was the itch. In January 2018 he started coding what became lemlist, and within about three months the first customers were paying. What almost nobody knew was how close the launch came to being someone else's. Without money for advertising, he posted in a Facebook group of cold-email experts to find beta testers. The post took off, and a competitor then announced it would copy the personalised-image feature. Moubeche rushed to launch on Product Hunt, unplanned, and reached number one Product of the Day. By December 2018 he had a hundred customers. Money came calling after that, though not at first. He pitched tens of venture funds and was refused every time. Once revenue passed $1 million, investors began writing to him weekly. He has said he came to see fundraising as time wasted on things that brought no direct value. His line in the interview was plain: bootstrapping is the best way to grow a SaaS business. In 2021, at around $10 million in annual recurring revenue, he says he sold 20 percent of the company in a $30 million cash-out at a $150 million valuation. He has since built the parent company lempire and quotes revenue of about $20 million after five years, with later podcast titles pointing to $45 million. These are his own figures. The lesson is in the order of events. The rejections came first, the offers came later, and he kept control in between. A founder who has been told no by every investor may be learning to need none of them. ## Ornikar (France) URL: https://startup.talk/france/ornikar/ Founders: Benjamin Gaignault, Alexandre Chartier Founded: 2014 (registered 7 January 2014; some press says late 2013); Based in: Paris, Île-de-France; Sector: Online driving school and driver insurance The first thing the driving schools did was sue. In April 2014, only weeks after Ornikar was registered in Paris, several trade unions took the start-up to court for unfair competition. In June the commercial court threw the claims out. It would not be the last time the establishment tried to shut the young company down. The two men behind it had known each other since school. Benjamin Gaignault and Alexandre Chartier met at a private lycée in Rueil-Malmaison. Gaignault went on to KEDGE Business School and graduated in 2011. He has been described as impatient by nature, someone who preferred building his own company to climbing a corporate ladder. Years later the two would say they had not had the idea of the century, only an answer to a real need in an existing market. That need was easy to see. Getting a French driving licence was slow and expensive, and prices were hard to compare. Ornikar offered the theory test online and, soon, lessons with partner instructors at lower prices. In November 2014 Partech Ventures and Elaia Partners put in €1.5 million, and early angels included Xavier Niel, Marc Simoncini and Jacques-Antoine Granjon. The pushback was fierce. In December 2015 a court ordered the company to stop giving driving lessons until it had received prefectoral approval. The approval arrived in March 2016, but the founders later said some traditional schools refused candidates who announced they had taken the theory test through Ornikar. Trade groups wanted the start-up closed. What the unions failed to see was that every attack made Ornikar better known. Press from Libération to the BBC and The New York Times covered the fight over regulation, and student-age customers did the rest. Funding followed: €10 million in 2018, €35 million in 2019 and, in April 2021, €100 million with the American fund KKR joining. By 2020 the company said 420,000 learners had trained on its platform, with lessons offered in more than 300 French cities. It added young-driver insurance in September 2020 and extended it to all drivers in June 2021. In 2025 its statutes added financing activities such as licence and car loans. The lesson is uncomfortable. Entering an old industry means being treated as a threat before you are treated as a customer. Gaignault and Chartier survived because they kept obeying the rules the courts set, got approved, and kept building while the incumbents were busy in court. ## Sea Group (Singapore) URL: https://startup.talk/singapore/sea-group/ Founders: Forrest Li, Gang Ye, David Chen Founded: 2009 (Garena, Singapore); rebranded Sea in 2017; Based in: Singapore; Sector: Gaming, e-commerce, digital financial services The first home Forrest Li made in Singapore was a single rented bedroom in a three-room public housing flat in Braddell. He shared it with his wife, whom he had met at Stanford. Between them they carried about US$100,000 in student loans. He was a long way from Tianjin, where he was born in the late 1970s, and from Shanghai Jiao Tong University, where he studied engineering. After graduating he worked at Motorola and Corning, then went to Stanford for an MBA. There he heard Steve Jobs tell graduates to stay hungry and stay foolish. Li has credited that speech with turning him toward starting a company. He did not start with a grand plan for a regional giant. In 2009 he founded Garena in Singapore, a gaming platform for Southeast Asia. The region had millions of young players, internet cafes full of them, and almost no credit cards. Selling games there meant building the payment rails yourself. The turn came in 2010, when Riot Games gave Garena the rights to publish League of Legends across Southeast Asia. A small company with a Braddell-sized balance sheet suddenly held one of the most played games on earth. Garena sold prepaid cards through shops, ran the servers and ran the esports. By 2014 it was valued at about US$1 billion. A year later Ontario Teachers' Pension Plan invested at a valuation above US$2.5 billion. Then Li did the thing that surprised the industry. In 2015 he launched Shopee, an online marketplace, using the money and the instincts gaming had given him. In 2017 the group raised US$550 million, renamed itself Sea and listed on the New York Stock Exchange, the first big US listing from a Southeast Asian tech company. The price came later. Sea raised US$6 billion in 2021, the largest such raise in the region at the time. Then the boom ended. In January 2022 Tencent sold a US$3 billion stake. The investment arm was shut down. Profit disappeared in 2022 and the company had to cut hard. It came back. In 2023 Sea posted its first profitable year since listing, US$162.7 million. For 2024 revenue reached US$16.8 billion and net income US$444 million. Free Fire peaked at more than 100 million daily players that year. Li, now a billionaire at around US$3.6 billion by Forbes' 2024 count, spends part of his time on a football club he bought from a state-linked owner and now leads the national football association. The lesson is not about size. It is about what a founder does with the narrow thing he has. Li had debt, one room and a gaming company in a region that could not pay online. He solved that one problem so well that everything else was built on top of it. ## Grab (Singapore) URL: https://startup.talk/singapore/grab/ Founders: Anthony Tan, Tan Hooi Ling Founded: 2012 (as MyTeksi, Kuala Lumpur); Based in: Singapore; Sector: Super-app: ride-hailing, food delivery, financial services As a boy, Anthony Tan worked on the assembly line at his father's company. His father, Tan Heng Chew, ran Tan Chong Motor, the Nissan assembler and distributor, so the son of the boss spent time among the cars, not above them. At 11 he was already trading X-Men comics for a profit. He was born in Kuala Lumpur in 1982, studied economics and public policy at the University of Chicago and then worked in supply chain and marketing at the family group. The obvious road led to the top of Tan Chong. He did not take it. The reason, by his own account, started with a complaint. At Harvard Business School a friend criticised Malaysia's taxis. Tan and his classmate Tan Hooi Ling wrote a business plan for a taxi booking app. It won second prize at the school's New Venture Competition in 2011. Second prize. Not first. Hooi Ling had her own constraint. A Malaysian engineer from the University of Bath, she had joined McKinsey, which paid for her MBA, and she owed the firm her return afterwards. She has described herself as an introvert who would rather stay out of the spotlight. In 2012 they launched MyTeksi in Kuala Lumpur, funded by a US$25,000 Harvard grant and Tan's own money. It became GrabTaxi, then Grab. The company moved to Singapore, and the fights began. Taxi drivers and motorbike taxi operators in Vietnam, Malaysia and Thailand pushed back. Regulators took years to catch up. In 2017 the company was hit by safety incidents, including the killing of a driver in the Philippines. The turn was a war with Uber. In March 2018 Grab absorbed Uber's Southeast Asian business in exchange for a stake, which left Uber with 27.5 percent at the time and Grab with the region. It added food delivery, payments and a digital bank licence with Singtel. In December 2021 it listed on Nasdaq through the biggest SPAC deal then seen. The price was paid in jobs. Grab cut about 360 people in 2020 and around 1,000 in 2023. Revenue reached US$2.80 billion in 2024, with a net loss of US$158 million, and the company reported its first full-year profit in early 2026. Tan once walked away from the family business to build something unproven. What he kept from that factory floor is harder to see. The founder who had watched cars assembled by hand built a company that moves people, food and money, and never stopped fixing the parts. ## Razer (Singapore) URL: https://startup.talk/singapore/razer/ Founders: Min-Liang Tan, Robert Krakoff Founded: 2005 (Razer brand traces to 1998 San Diego, subsidiary of Karna LLC, which went bankrupt in 2001); Based in: Singapore (also Irvine, California); Sector: Gaming hardware, software, fintech As a child, Min-Liang Tan pirated a game. Decades later, a lawyer and a billionaire, he paid for it. He put US$10,000 into the Kickstarter for Wasteland 2, the sequel to the game he copied, and has said it was to atone for the theft. The developers wrote him in as a non-player character. He was born in Singapore in 1977, the youngest of four. Two siblings became doctors. He went to Raffles Institution and Hwa Chong Junior College, then studied law at the National University of Singapore. By 2002 he was in the top 20 of his postgraduate class and working as an advocate and solicitor at the Supreme Court. It was the safe road, and everyone around him would have understood it. He took a different one. In 2005 he and an American, Robert Krakoff, relaunched a brand with a messy past. Razer had begun in 1998 in San Diego as a subsidiary of Karna LLC, built to sell a high-end gaming mouse called the Boomslang. Karna went bankrupt in 2001. The brand survived. Tan and Krakoff bought it and restarted, with backing from Li Ka-shing and Temasek. What nobody could have guessed was how far a mouse could go. The company named products after predators, snakes for mice and spiders for keyboards, and spoke to gamers as peers. The DeathAdder line later sold more than 20 million units. The company kept reaching. It bought the software of the Ouya microconsole in 2015, THX in 2016 and the phone maker Nextbit in 2017. On 14 November 2017 Razer listed in Hong Kong under the code 1337, gamer slang for leet. Shares closed about 18 percent higher on day one. Tan, then 40, was reported to be worth US$1.6 billion, the youngest self-made billionaire in Singapore. Then came the misses. The game store closed. Ouya shut. The Razer Pay wallet was closed in 2021. In 2020 the Federal Trade Commission found it had overstated the protection of its Zephyr masks, and the company refunded over US$1.1 million. In 2019 Kotaku reported accusations, from 14 former staff, that Tan had berated and threatened employees. In May 2022 Razer went private again, after revenue had reached US$1.62 billion for 2021. Krakoff had died the month before. The odd thing about Tan is how small the origin was. A boy copies a game, a lawyer buys a failed mouse brand, and the debt is finally paid in a crowdfunding page. Founders often build the thing they once could not afford. ## Carousell (Singapore) URL: https://startup.talk/singapore/carousell/ Founders: Quek Siu Rui, Marcus Tan, Lucas Ngoo Founded: 2012 (launched 2012; incorporated as Carousell Pte. Ltd. January 2013); Based in: Singapore; Sector: Online classifieds marketplace The first thing ever sold on Carousell was a used Amazon Kindle. It went for S$75. The people who built it had spent their own money on gadgets they could not easily resell. Quek Siu Rui, Marcus Tan and Lucas Ngoo were NUS students who had spent time as interns in Silicon Valley through the university's Overseas College programme. They had seen how small teams build things. Back home, they were frustrated by how hard it was to sell second-hand electronics online. They won Startup Weekend Singapore in 2012 with an early version of the idea, a snap-and-sell app that was first called Snapsell. They tried it on blog shop sellers using a mock-up. Then they did the unglamorous part. They walked through flea markets and asked sellers whether they would use it. Of 100 they approached, 70 said yes. The money was nothing. For the first year and a half, Quek has said, none of the three drew a salary. They took the train and ate cheap. In a 2020 profile Quek said starting the app was no walk in the park. What almost nobody knew at the time was how close they were to being rich early. In 2014, a news report said, a 26-year-old Quek turned down an offer of US$100 million for the company. He said no. That bet needed more years and more money. Series B brought US$35 million in 2016 and Series C US$85 million in 2018, co-led by Rakuten Ventures and EDBI. In 2019 it bought OLX Philippines and merged with Telenor's classifieds arm, which brought in Mudah and Chotot. Carousell reached a valuation of US$1.1 billion in September 2021 after a US$100 million round. It also paid a trust price. In 2018 police said more than 70 percent of Singapore's e-commerce scams happened on Carousell. The company added star ratings, fingerprinting to catch repeat offenders and a payment-holding system built with DBS, Stripe and Visa. A data leak later exposed more than 2.6 million users. Still, the company is growing into luxury resale. In September 2025 it opened a pre-owned luxury boutique in Singapore. There is a plain lesson in the 70 out of 100. The founders asked strangers in person before writing much code. They hung on to three equal partners, a Kindle and an answer. ## Lazada (Singapore) URL: https://startup.talk/singapore/lazada/ Founders: Maximilian Bittner, Pierre Poignant Founded: 2012; Based in: Singapore; Sector: E-commerce marketplace In 2011 Max Bittner packed up his young family and left Germany for Singapore. He had a backer, Rocket Internet, and an idea borrowed from a market he already knew: sell everything online, the way Amazon did. There is not much public record of his early life. He is said to have been born in Munich. He studied at University College London, worked as an analyst at Morgan Stanley, then as a consultant at McKinsey, and went on to an MBA at Northwestern. A 2014 CNBC profile described him as a consultant turned entrepreneur. The bet was bold for the time. Southeast Asia had hundreds of millions of people and little online retail. Roads were bad, credit cards were rare and cash on delivery was the norm. Lazada launched in 2012 as an Amazon-style retailer, added third-party sellers in 2013 and began operating in Singapore in 2014. It raised over US$685 million from investors such as Tesco and Temasek. The turn arrived in April 2016, when Alibaba took a controlling stake. In 2017 it raised that to 83 percent with another US$1 billion. A German-founded startup now ran on Chinese capital and logistics. The founder did not stay. Bittner stepped down as CEO in March 2018, and a relay followed: Lucy Peng and Pierre Poignant in 2018, Chun Li in 2020 and James Dong in 2022. He later became chief executive of Vestiaire Collective, the luxury resale site. Lazada launched LazMall for authentic brands that year and reached 50 million annual active buyers by September 2019. The cost showed up later. Counterfeit goods dogged the platform. The Philippine IP office tied 69 percent of its reports in the first half of 2023 to Lazada. A January 2024 Singapore retrenchment drew criticism from the union, which said it had not been consulted. Another round of cuts, about 5 percent of staff across Southeast Asia, followed in 2026. Shopee became a tough rival. Lazada still operates across the region and in September 2025 opened a programme letting Tmall merchants sell directly to Southeast Asia. The lesson of Bittner is about being first and being the founder. He moved a family across the world to start something others would run. A company can outgrow the person who started it, and the founder has to decide whether that is a loss or a launch. ## PropertyGuru (Singapore) URL: https://startup.talk/singapore/propertyguru/ Founders: Steve Melhuish, Jani Rautiainen Founded: 2007 (company registered as AllProperty Media Pte Ltd around 2006); Based in: Singapore; Sector: Proptech / online property marketplace In 2006 the condo Steve Melhuish was renting in Singapore went up for sale. He had been in the country about eighteen months, and he knew almost nothing about its property market. Finding a new home meant newspaper classifieds and agents who each held a slice of the picture. He was 47 by the time he told this story to Malay Mail, and he was blunt about it. Many people think you have to be passionate to start a business, he said. He was clueless. What he did have was a plan, and an introduction. An investor friend who had heard the plan said he knew someone at INSEAD with a similar idea: Jani Rautiainen, a Finnish entrepreneur already chewing on the same problem. The two teamed up. Their first company carried a dull name, AllProperty Media Pte Ltd, registered in Singapore. The brand people would remember, PropertyGuru, came with the 2007 launch. The early years were a fight against copycats. Property portals are easy to imagine and hard to keep alive, and Melhuish later said the company's growth came from paranoia, because so many rivals had appeared and vanished. He kept expanding anyway: Malaysia, Thailand, Indonesia, then fifteen mobile apps. Then came the money. In 2016 PropertyGuru raised S$175 million from investors including TPG Capital and Emtek Group, which that year was the second-largest raise by a Southeast Asian tech firm. That same year it agreed to buy Ensign Media, its biggest deal yet. By then the site held about 90 percent of the Singapore market and drew 14 million visitors a month. A quieter story ran alongside. In early 2015, a man named Hari Krishnan, LinkedIn's first employee in Asia, used PropertyGuru to find his own apartment in Singapore. He had known Melhuish and Rautiainen for years, and he called himself a happy user. He joined as president and chief business officer that November and eventually ran the company as CEO. Rautiainen left the board in 2019. The company listed on the NYSE in March 2022, then went the other way: in August 2024 EQT agreed to take it private in an all-cash deal valuing it at about US$1.1 billion, completed by December. Krishnan left in mid-2025, and Lewis Ng took over. The lesson sits in the origin. Neither founder started with a calling. One started with a rental that was about to be sold, the other with a half-formed idea he was lucky enough to compare against someone else's. Irritation, plus the right introduction, built a market leader. ## Trax (Singapore) URL: https://startup.talk/singapore/trax/ Founders: Joel Bar-El, Dror Feldheim Founded: 2010; Based in: Singapore (R&D in Tel Aviv); Sector: Retail analytics / computer vision for shelf recognition Before he built a company that photographs store shelves in seventy countries, Joel Bar-El sold toys. He imported them from Czechoslovakia and walked them into shops around Tel Aviv, a student trying to pay for life with a partner who would later become his wife. He has called it a learning experience. He grew up in Tel Aviv, close to the sea; his father was a lifeguard. At Tel Aviv University he was drawn to astrophysics, started in the physics faculty, and found it too hard after the first year. So he rearranged his path, taking master's-level courses early and mentoring children through the Perach programme, which paid half his tuition. He graduated in 1996. What almost nobody pictures when they hear about Trax is how slow the start was. Bar-El and Dror Feldheim, a Swiss-Israeli, founded it in 2010 in Singapore, with research in Tel Aviv. The idea: take photographs that field representatives snap of store shelves on a phone, and let image recognition count what is there, and what is missing. The founders made a risky early decision. They went for a multinational as their first client, rather than small ones. Procter & Gamble was the first to approve the concept; Coca-Cola and the rest of the consumer goods industry followed. Bar-El later admitted the company paid a price for that choice. The first recurring client arrived only in mid-2013, three years in. Early money came only from family offices and wealthy individuals, not venture capital. That first round, around 2012, totalled about US$7.5 million. The technology took a while to break through, he said. When it did, it grew fast. Warburg Pincus led a US$64 million round in 2017; Boyu Capital and GIC came in by 2019; in 2021 SoftBank Vision Fund 2 led a US$640 million round at a valuation reported between US$2 billion and US$3 billion. Trax bought Lenztech, Shopkick and Planorama in 2019. In May 2024 Bar-El and Feldheim stepped back from running the company, staying on the board. Bar-El has since started Swish, a computer-vision basketball company, using the same idea on a court instead of a shelf. The pattern in his life is a man who keeps selling small things until a bigger one fits. For a founder the useful lesson may be patience: a first big client can cost you years, and still be the reason everyone else listens. ## Nium (Singapore) URL: https://startup.talk/singapore/nium/ Founders: Prajit Nanu, Michael Bermingham Founded: 2014 (Instarem launched Aug 2015; rebranded Nium 2019); Based in: Singapore; Sector: Fintech / cross-border payments infrastructure The idea for Instarem came from a bachelor party. In 2013 Prajit Nanu needed to send money overseas to help pay for a friend's celebration, and the transfer was slow and expensive. He saw, in his words, a huge gap in the cost efficiency of cross-border remittance. And he decided to build his own fix. He had come to this from sales. Before Instarem he held sales roles at the outsourcing firms TMF Group and WNS Global Services. He started the business from his apartment with a colleague, Michael Bermingham, who became chief operating officer. Founded in 2014, it took until 2015 to go live. When the first payment went through that year, the company had three employees and an outsourced technology team. The name was a contraction: instant remittance. Tiny seed rounds followed. In January 2015 came US$500,000 from Global Founders Capital, then US$5 million for a Series A in 2016 with Vertex Ventures among the backers. Those small cheques mattered later. The worst moment came in early 2020. When the pandemic hit, investors backed out of a funding round. Vertex Ventures, which had been there from the Series A, stayed and supported the company. Nanu has also talked about the pull of family, recalling his daughter sitting on his lap during a client call when she fell ill. Balancing work and home, he has said, is a continuing challenge. The company survived and changed shape. In 2019 it rebranded as Nium, turning a consumer remittance app into infrastructure that banks and businesses build on. In July 2021 a US$200 million Series D made it a unicorn, with a valuation put at about US$2 billion by early 2022. Then came the cost. Nium cut about 10 percent of staff in October 2023 and lost several senior executives that year. A planned US listing slid from 2025 to the end of 2026, which Nanu tied to strengthening the leadership team. In 2024 the Brunei Investment Agency led a US$50 million Series E. For a founder, the lesson is the size of the trigger. Nothing grand: one irritating transfer, a spare room, three people. What counted was who stayed when the money left. ## Advance Intelligence Group (Singapore) URL: https://startup.talk/singapore/advance-intelligence-group/ Founders: Jefferson Chen (Chen Lanjie) Founded: 2016; Based in: Singapore; Sector: AI and fintech (Atome buy-now-pay-later, ADVANCE.AI risk and identity, Ginee e-commerce tools) Long before Jefferson Chen ran a Singapore fintech valued near US$2 billion, he started a nonprofit. At Tsinghua University he founded Technology & Education: Connecting Cultures, known as TECC, a student group built around bringing technology and education across borders. It later earned him the UN's Fritz Redlich Alumni Award, and he is described as its youngest recipient. Chen studied computer science at Tsinghua and graduated with first-class honors. That detail matters, because the company he would build was about teaching machines to judge people. His next move was across the Pacific. He went to Stanford for an MBA and co-founded its Chinese Entrepreneur Organization. Then came money: Goldman Sachs, where he worked on IPOs and mergers for Asian companies, and Farallon Capital, where he headed private investing for Greater China at a firm managing about US$30 billion. Later he became a partner at GSR Ventures, which is also an investor in his company. So Chen arrived at founding as an investor who had watched many companies from the outside. In 2016, by most profiles, he started Advance Intelligence Group in Singapore. It grew into three units. ADVANCE.AI sells identity verification and risk tools to businesses. Atome offers buy-now-pay-later to shoppers. Ginee helps online sellers run their shops. The backers read like a Who's Who of Asian growth money: Warburg Pincus, SoftBank Vision Fund 2, Northstar Capital, Pavilion Capital. Reported funding runs from more than US$500 million to US$616 million. In 2019 it was valued around US$400 million; by 2023 above US$2 billion. The group says it works across more than fifteen countries. The trade-offs of this story are what the public record does not show: the early hard months, who said no, what it cost. Chen has given few interviews, and the sources are thin on that side. What the record does show is a founder whose habits repeat. At Tsinghua he made a group to link technology and education; at Stanford a group for fellow entrepreneurs; in Singapore a company built on data and credit. He keeps building the institution first, then the product. For a first-time founder, that is a quiet kind of lesson: collect your people early. ## Ninja Van (Singapore) URL: https://startup.talk/singapore/ninja-van/ Founders: Lai Chang Wen, Shaun Chong, Boxian Tan Founded: 2014; Based in: Singapore; Sector: Logistics / last-mile delivery His parents thought he had become a delivery driver. In a sense he had. Lai Chang Wen, a former derivatives trader at Barclays, was driving parcels around Singapore himself in 2014, one of three young founders of a logistics company they barely understood. It did not start that way. Lai, born in 1987, went to Raffles Institution and Raffles Junior College, then studied finance at SMU. He turned down a government scholarship because he always wanted to go into business. As a student he bought used phones and resold them at a markup, and put together bicycles from parts. In 2010 he co-founded Marcella, a made-to-measure menswear brand born from his own trouble finding clothes that fit. It opened its first store that year and later ran five shops. His mother called leaving a five-digit salary for fashion a real shock. Then he left fashion too. The problem that changed everything was a van. Marcella had to deliver clothes, and deliveries were a mess. With co-founders Shaun Chong and Boxian Tan, Lai concluded Marcella was not going to go far, and the three turned the pain into the product. They had no logistics experience. To learn, they paid couriers to talk about their jobs and mailed parcels to one another. They wrote algorithms to plan better routes. The name came from the way Singapore's armed forces slang referred to food trucks that turned up in restricted military areas. The cost was personal. The founders took 90 percent pay cuts and worked 22-hour days; Lai has said he even cleaned the office toilet. He later admitted that as a trader he had saved so little that his bank account could not top up his EZ-link card. Monk's Hill Ventures led a US$2.5 million Series A in March 2015. Later rounds brought DPDgroup, Grab and Alibaba. By September 2021 the company raised US$578 million and crossed a valuation of US$1 billion, with about one million parcels a day moving by 2019. The price came later. In 2023 Lai said he would hold off on an IPO until profitability improved. There were two rounds of layoffs in 2024, and Vietnam operations were suspended in September 2024 over delayed salaries. The takeaway sits in the first scene. His parents saw a driver; he saw a company. Knowing nothing was the starting condition, and doing the job yourself, for a while, was the school. ## Patsnap (Singapore) URL: https://startup.talk/singapore/patsnap/ Founders: Jeffrey Tiong, Markus Haense, Guan Dian, Ray Chohan Founded: 2007; Based in: Singapore; Sector: Innovation and IP intelligence software (patent analytics, R&D) Jeffrey Tiong was an intern in Philadelphia when he first got fed up with patents. His job at a small company called BioConnect Systems was to research and evaluate them. The government websites that held the records were hard to use, and the documents themselves, he later said, were rather abstruse. A biomedical engineering student from the National University of Singapore had a simple, irritated thought: someone should make this easy. That irritation is the whole origin story. There is no famous family firm behind it, and little public record of his childhood. What there is, is a student who turned a complaint into a product. Back in Singapore he built a search engine for patents with help from NUS Enterprise. The Media Development Authority's iJam programme gave him S$55,000. It was a small cheque for a hard job. Patent data sits in dozens of national offices, in different formats and languages. Making it searchable for a company, not a lawyer, meant years of unglamorous work. Tiong brought in Markus Haense as technology chief, Ray Chohan for new ventures and Guan Dian to run Asia Pacific. The turn came when the buyers changed. Patsnap stopped being a tool for patent attorneys and became a tool for the people who make products. Engineers and R&D teams could check whether an idea already existed before spending money on it. Subscriptions ran about US$20,000 to US$30,000 a year for typical clients, around US$5,000 for startups with limited use, and up to US$500,000 for giants such as Dyson and DJI. By 2021 the company counted more than 10,000 customers, among them Spotify and Xiaomi, and had over 700 staff. The investors followed. Vertex Ventures, an early backer, led the Series A. In 2018 Sequoia and Shunwei Capital led a US$38 million Series D. In March 2021 SoftBank's Vision Fund 2 and Tencent led a US$300 million Series E that took the valuation above US$1 billion. The quiet cost of a company like this is time. It took well over a decade from a Philadelphia internship to a unicorn round, and almost none of that decade made headlines. Patsnap also depends on patent offices it does not control, and on customers who can cancel a subscription. Tiong's lesson is a small one and easy to miss. He did not set out to build a unicorn. He set out to fix the thing that annoyed him most as a student, and he kept fixing it until other people were willing to pay for the fix. ## Funding Societies (Singapore) URL: https://startup.talk/singapore/funding-societies/ Founders: Kelvin Teo, Reynold Wijaya Founded: 2015 (Feb 2015); Based in: Singapore; Sector: SME digital financing, fintech (peer-to-business lending) In its first month, Funding Societies received about S$3 million in loan applications. It lent out roughly S$250,000. The gap says almost everything about the business: demand was never the problem. Trust was. The founders, Kelvin Teo and Reynold Wijaya, were Harvard Business School students when they started in February 2015. Reports later described the company being launched in about 100 days while the two were still finishing their MBAs. An early account says they put in around US$400,000 of their own money alongside about US$1.2 million from backers. Little is public about where either grew up. Their bet was simple to state and hard to do. Small businesses in Southeast Asia could not get bank loans, and people who wanted a return had few places to put their money. A platform could match the two, as long as it picked borrowers well. The first month showed how picky it had to be: for every dollar disbursed, about twelve dollars of applications did not become loans. What helped was borrowed credibility. In April 2016 DBS Bank partnered with the platform. That year it launched in Indonesia as Modalku and became one of six licensed SME lending platforms in Malaysia. Sequoia India and Alpha JWC led a S$10 million Series A in August 2016, and SoftBank Ventures Korea led a US$25 million Series B in April 2018. The company kept adding countries and partners. It teamed up with Razer Fintech on merchant financing in 2020 and entered Thailand in February 2021. A larger Series C, reported at US$144 million with SoftBank Vision Fund 2 involved, followed, and Maybank invested in 2024. The platform says it has disbursed more than US$2.6 billion to small businesses in over 5.1 million transactions, though that figure is the company's own. The cost of this kind of lending is risk. Every country has its own regulator, its own credit-data gaps and its own repayment culture, and a lender to small firms takes the first hit when they struggle. Funding Societies operates as a licensed platform rather than a bank, which shapes what it can do alone. The lesson sits in that first month. A founder who sees S$3 million of demand and S$250,000 of supply should not panic. The ratio is the product. ## Coda Payments (Singapore) URL: https://startup.talk/singapore/coda-payments/ Founders: Neil Davidson, Paul Leishman Founded: 2011; Based in: Singapore; Sector: Digital goods payments, out-of-app monetisation for games and apps In 2011, the only way to buy digital content online in much of Asia was a credit card. Most people did not have one. That was the starting point of Coda, which says it began that year in Indonesia. Two names appear in the company's own telling of its origin: Neil Davidson and Paul Leishman. Little else is public about where they came from or how they met. One directory also lists Bobby Choi as a third founder. What is clear is the problem they picked. Game publishers wanted Southeast Asian players, and those players wanted to pay with what they actually had: mobile airtime, convenience-store vouchers, bank transfers, e-wallets. So Coda built a patchwork. Each country meant a new set of local payment methods, a new partner and a new regulator. It was dull work that big payment companies largely skipped, which is why it became valuable. A publisher could plug in one integration and reach players across many markets. The company raised a Series A in 2013, and in 2016 TechCrunch reported a round of about US$2 million, around the time Codapay added bitcoin processing for merchants in the region. In 2022 it took an investment from Apis Partners. A very large Series C figure is quoted in one database, but it conflicts with other records and should be treated with care. The turn was the app-store squeeze. As publishers grew tired of the fees charged by Apple and Google, Coda pushed toward out-of-app purchases: web stores, purchase links and direct payments that let a studio keep more of each sale. The company now describes itself as an out-of-app monetisation and commerce company, working with more than 300 publishers across over 70 markets. In February 2023 Shane Happach was named chief executive, and Davidson moved to executive chairman. In 2025 Coda bought Recharge, a European prepaid payments platform. It says it now has 400-plus payment channels, more than 120 million active users and over 600 staff in 20-plus locations. Those are the company's own figures. Coda's story is not one dramatic night. It is a founder pair who stayed on one unfashionable problem for more than a decade, until the world's biggest platforms started to look like the problem too. ## Aspire (Singapore) URL: https://startup.talk/singapore/aspire/ Founders: Andrea Baronchelli, Giovanni Casinelli Founded: 2018; Based in: Singapore; Sector: SME finance, neobank, spend management Before Aspire, many small businesses in Southeast Asia lived in a mess. They combined incumbent banks, spreadsheets and several fintech providers, and none of them talked to each other. Approvals ran across multiple platforms with no single source of truth. That was how customers described life before the product, and it is the problem Aspire was built on. The company was founded in 2018 in Singapore by Andrea Baronchelli and Giovanni Casinelli, and one profile describes it as a Y Combinator alumnus. Beyond their names and roles, very little is public about where they were born or what they did before. That is unusual for founders at this scale, and it leaves the story to be told through the product. Aspire started narrow. It offered working capital loans to small and medium-sized businesses that banks tended to overlook. Then the founders listened to what customers kept needing. They wanted a place to hold money in several currencies, pay suppliers, issue corporate cards, collect from customers and send invoices into their accounting software. So Aspire added cross-border accounts, cards, payables and receivables, and integrations with Xero, QuickBooks and NetSuite. The approach worked. A US$32.5 million Series A arrived in August 2019, and a reported US$158 million Series B in September 2021. In February 2023 Aspire closed an oversubscribed US$100 million Series C, co-led by Lightspeed and Sequoia Capital SEA, with PayPal Ventures, Tencent and LGT Capital Partners in the round. It came while venture funding was slowing across the region. By then, annualised total payment volume had tripled in twelve months to US$12 billion, and the company served over 15,000 businesses. The cost of growing a financial product is the weight of regulation and credit risk. Cards and loans must be underwritten, and each country has its own rules. The company said it would spend the new capital on product, regional expansion and hiring. The lesson for a first-time founder is to start with the customer's mess, not the category. Aspire did not begin as a neobank. It began as a loan, and the bank grew out of everything borrowers asked for next. ## Carro (Singapore) URL: https://startup.talk/singapore/carro/ Founders: Aaron Tan, Aditya Lesmana, Kelvin Chng Founded: 2015; Based in: Singapore; Sector: Used-car marketplace, financing, subscription Aaron Tan was 13 when he started making money. He built websites and sold them. It began in Singapore, where he grew up, with a kid who had worked out that something he could make was something others would pay for. Years later he studied computer science at Carnegie Mellon, where he met Aditya Lesmana and Kelvin Chng. He then worked as a venture capitalist in the United States. In that job he met car-related companies such as Beepi, Uber and DriveShift and saw momentum in the American used-car market. Back home the picture was different. Southeast Asia's used-car market was opaque, with middlemen and little way for a buyer to know whether a deal was good. In 2015 he returned to Singapore and brought his classmates with him. He has described it as selling them the opportunity: a chance to build an algorithm that could clear up the market, rather than carry on with what they were doing. Carro was founded that year. The early model was not a listings site. Carro bought cars, inspected them and sold them on, which meant taking inventory risk to earn trust. Over time it layered on financing and, in 2019, Singapore's first car subscription service. The turn came in 2021. By March, revenue was about US$300 million, roughly 2.5 times the prior year, and the company said it was profitable. In June, SoftBank Vision Fund 2 led a US$360 million Series C that took the valuation past US$1 billion. At the time Tan told CNBC he planned to go public within 12 to 18 months. A Temasek and PNB-led round of more than US$100 million followed late that year, with Sime Darby, Shinhan and Mirae Asset joining. The listing did not arrive on that schedule. In September 2025 Carro raised US$60 million from Japan's Cool Japan Fund, and Tan said at an event that the company was preparing to list but had not chosen where. The lesson is old but still true. Tan did not invent a new market. He noticed a broken one while working somewhere else, then went home and recruited friends who trusted him enough to follow. That trust, more than the algorithm, is what he had to sell first. ## Zilingo (Singapore) URL: https://startup.talk/singapore/zilingo/ Founders: Ankiti Bose, Dhruv Kapoor Founded: 2015; Based in: Singapore; Sector: Fashion e-commerce and B2B supply-chain marketplace The idea, by the founder's own account, arrived on a holiday. Ankiti Bose was in Bangkok when she noticed how many small shops had no presence online at all. Racks of clothes, hand-lettered signs, and nothing a buyer in another country could find. She was an analyst at Sequoia India at the time, a young woman whose job was to judge other people's companies. Her family had moved from city to city because of her father's job at a state-owned oil company, which left her used to starting over among strangers. In December 2014, at a house party in Bengaluru, she met a neighbour named Dhruv Kapoor. He would build the technology. In 2015 the two started Zilingo in Singapore, a name that plays on the word zillion. The first backer was Sequoia India, the firm she had been working for. Series A followed in September 2016 with US$8 million, then US$18 million in 2017 and US$54 million in 2018. What nobody expected was where the money would be made. The consumer site was a shop window. The real business was a tool for sellers, then a marketplace where those sellers' goods reached trade buyers. By 2018 the seller-management tool was profitable. By September 2019, about 80 percent of revenue was B2B. That year it raised US$226 million, with Temasek joining, and its value approached US$1 billion. Bose was in her twenties, and one of very few women to run a company of that size in Asia. Then the pandemic hit. In 2020 Zilingo closed its US and Australian offices, cut about 12 percent of its staff and lost its CFO. The fall came in 2022. In March, whistleblowers reported payments approved by Bose to companies with no obvious link to Zilingo's business. She was suspended on 31 March and dismissed on 20 May, after an independent forensic firm looked into the books. In June the two co-founders proposed a management buyout. It did not save the company. Zilingo's technology was sold to a Swiss firm in January 2023, and the company went into liquidation a month later. The story did not end quietly. In 2024 Bose filed a US$100 million defamation suit against an investor and accused Kapoor and a former COO of fraud, intimidation and harassment. Those are her claims in court, not findings. For a founder, the lesson is uncomfortable. A company that grows from a spark on a Bangkok street to a near-billion-dollar valuation in four years can outrun its own controls. The money was real. The checks on it were not. ## Circles.Life (Singapore) URL: https://startup.talk/singapore/circles-life/ Founders: Rameez Ansar, Abhishek Gupta, Adeel Najam Founded: 2014 (company origins); consumer launch 2016; Based in: Singapore; Sector: Digital telco / telecom software More than 100 meetings. That is how many investors Abhishek Gupta says the Circles.Life founders sat across from before their first major round came together. One investor, he has said, showed interest within half an hour. The other hundred or so did not. The founders, Rameez Ansar, Gupta and Adeel Najam, had met while working in places like McKinsey, BCG, Temasek, Huawei and Bain Capital. They were frustrated by what they called static and restrictive telco plans, and they wanted to give power back to the customer. The problem was that nobody had heard of them. A digital-only telco was a new idea in Singapore, and a new brand with no stores had to persuade people to hand over their phone numbers. They lacked trust, awareness and resources. Hiring was hard too, because a startup nobody knew was asking for good people to leave safe jobs. They started small, working from the BLK 71 startup hub. A wholesale deal with the telco M1 in July 2015 gave them the network they could not afford to build. The public launch came in 2016. Then they got loud. One stunt was a vending machine that gave S$50 for S$3. Another was a vandalism-style campaign, Youtiao666, that got people talking. The company says it signed over 100,000 customers in its first year. The turn came when investors did start to listen. Sequoia backed the company in 2019, EDBI and Founders Fund followed in June of that year and Warburg Pincus in 2020. With the money Circles took the model abroad: Taiwan, Australia, Indonesia, Japan. By June 2022 it said it held more than 7 percent of Singapore's mobile market. A SPAC deal with Bridgetown was reported that July at around US$2.5 billion. The price of speed was a map full of exits. Indonesia's Live.On customers moved to XL Axiata's AXIS brand in February 2024. Taiwan was scheduled to close in July 2024. The Australian business shut in January 2025 and its customers went to Amaysim. The listing talks did not end in a listing. The lasting business may be the software. Circles turned its own telco platform into a product other operators could buy. A founder reading this should notice the numbers in the opening: one yes in half an hour, a hundred slow noes. The story of the company is mostly what happened after the person in the middle of that spread finally said yes. ## Thunes (Singapore) URL: https://startup.talk/singapore/thunes/ Founders: Peter De Caluwe Founded: 2016 (predecessor TransferTo founded 2005); Based in: Singapore; Sector: Cross-border B2B payments / fintech Most startups begin with an empty page. Thunes began with a split. In 2005 a company called TransferTo started moving mobile airtime across borders. By 2016 it had grown into two different businesses. One kept selling mobile top-ups and became DT One. The other, with Peter De Caluwe among its co-founders, took the name Thunes and went after a bigger, harder market: moving money between businesses in different countries. There is little public record of De Caluwe's early life, and this account does not guess at it. What is documented is the problem he said he set out to solve in 2016, the inefficiency of cross-border payments between businesses. A payment from one country to another passes through banks, wallets and local schemes, each with its own rules, each with its own licence. Thunes' answer was to build one network and connect to the rest. The partners were payment companies, banks and wallet providers, not consumers. The company raised its Series B in 2020, led by Helios Investment Partners with Checkout.com and GGV Capital. It bought Limonetik, a Paris-based online payments platform, in 2021, and took a majority stake in the compliance-technology firm Tookitaki in April 2022. The turn was investor appetite in a harder market. In July 2023 Thunes raised US$72 million from Marshall Wace at a valuation reported above US$900 million. In April 2025 it raised US$150 million more, led by Apis Partners and Vitruvian Partners, the largest round in its history. The company has not disclosed that valuation. In June 2025 it said it would acquire Tilia, and it now holds money transmitter licences in all 50 US states. The leadership moved too. Floris de Kort ran the company as CEO until the end of October 2025. De Caluwe, who had spent about 18 months as Deputy Chairman, came back to the job, saying his focus would be continuity and global expansion. The quiet part of this story is the one a founder should notice. Thunes is not famous. Few people have used it directly. It makes money by being the plumbing under other people's apps, which means every new country is another licence, another bank and another year of patience. Sometimes the business worth building is the one the end customer never sees. ## Homage (Singapore) URL: https://startup.talk/singapore/homage/ Founders: Gillian Tee Founded: 2016; Based in: Singapore; Sector: Healthtech: home care and caregiving marketplace Gillian Tee had already done the thing most founders dream of. She had built Rocketrip, a travel startup backed by Y Combinator, which went on to raise nearly US$30 million. Then she did something that does not fit the usual startup legend. She moved home. After about 15 years abroad she returned to Singapore to be closer to her family. What they do show is what home looked like. Singapore was ageing fast. Families who wanted care for a parent had few clear options: an agency with a phone number, a foreign domestic worker, a nursing home. Prices were opaque and trust was thin. In 2016 she started Homage, an app and marketplace where families could book prescreened caregivers and nurses. The first money was small. A seed round of US$1.2 million came from 500 Startups, Golden Gate Ventures and SeedPlus. In July 2018 a US$4.15 million Series A followed, co-led by Golden Gate and HealthXCapital. Care is a harder business than software. A late app does not hurt anyone. A caregiver who does not show up does, and so the work is vetting and training as much as technology. The Series B was led by EV Growth. The turn came in 2021, when the pandemic made home care urgent. On 3 September Homage announced a US$30 million Series C, led by Sheares Healthcare Group, the healthcare arm of Temasek. The round was oversubscribed, and Homage said its total funding passed US$45 million. It reported more than 6,000 prescreened care professionals and revenue that had more than tripled in a year. Business outside Singapore grew over 600 percent that year, with operations in Malaysia and Australia. The company had to prove it could carry the same trust across borders, in countries with different care rules and a different workforce. The quiet insight is in why she came back at all. A founder with a respected exit-in-waiting went home for family reasons, and found that the same reason, caring for someone, was a market. The best problems are sometimes the ones that arrive at your own door. ## ShopBack (Singapore) URL: https://startup.talk/singapore/shopback/ Founders: Henry Chan, Joel Leong, Lai Shanru, Josephine Chow Founded: 2014; Based in: Singapore; Sector: Cashback, rewards and shopping platform Six people left the same employer in 2014 to build a company that gave money back to shoppers. Henry Chan and Joel Leong had worked at Zalora, the fashion retailer, where their work involved partnerships with brands and merchants. They had also both been through the NUS Overseas Colleges programme as students, which put them in start-up cities while still undergraduates. They quit stable jobs, brought four former Zalora colleagues with them, and began ShopBack in August 2014. The idea was simple to explain. Online stores pay affiliates a commission when a shopper arrives through a link. ShopBack shared part of that commission with the shopper, as cashback. For the store, it was a sale it might not have had. For the buyer, it was a discount. The early money was thin. A seed round of over US$500,000 came in 2014, and in March 2015 East Ventures and Accel-X led a US$600,000 round. About US$15 million more followed before the Series A. The turn came in 2017 and 2018. Credit Saison led a US$25 million Series A in November 2017. By early 2018 the company reported annualised sales of over US$500 million across more than 5 million users, though that is sales through its links, not its own revenue. In April 2019 Rakuten Capital and EV Growth led US$45 million more. Then it acquired its way into new markets and products. Seedly, a personal-finance community, came in May 2018 and was sold again in October 2020. In April 2020 it bought Ebates Korea. In November 2021 it bought the buy-now-pay-later firm hoolah. The price showed up in trust. On 17 September 2020 ShopBack suffered a data breach and told customers eight days later. For a company built on a shopper handing over an account to earn a few dollars back, that gap mattered. It now says it has about 30 million users in 13 countries. The founders began with a small observation: the money was already moving, and someone was keeping all of it. ## Zenyum (Singapore) URL: https://startup.talk/singapore/zenyum/ Founders: Julian Artopé, Frederik Krass Founded: 2018 (conceived Aug 2018; launched in Singapore Oct 2018); Based in: Singapore; Sector: Dental care, clear aligners, consumer health Julian Artopé once wore braces and hated them. Years later, that memory of pain and discomfort became the starting point of a company. He has described the idea in plain terms. Plenty of people want clear aligners. Existing options cost so much that straight teeth had become something of a luxury. In an interview with the South China Morning Post he said exactly that: the demand was there, the price was the wall. There is little public detail on where Artopé grew up or what he studied. What the record does show is speed. He brought in Frederik Krass as co-founder and conceived Zenyum in August 2018. By October, the invisible braces were on sale in Singapore. That is two months from idea to customers, in a field where most dentists measure change in years. The model was simple to say and hard to run. Zenyum 3D-printed transparent aligners, sold them at a flat price, and relied on partner dentists to check and fit them. The pitch was that a smile should not cost a small fortune. Rivals included Invisalign, with its global brand, and trust was the real hurdle: a patient has to believe that a cheaper aligner will not harm their teeth. Money followed. A seed of about US$2 million arrived in April 2019, then a US$13.6 million Series A in November 2019 from RTP Global, Sequoia India, TNB Aura and Enterprise Singapore's Seeds Capital. In June 2021, L Catterton led a US$40 million Series B. Total funding is estimated at over US$50 million. The company did not stay a one-product firm. It added electric toothbrushes, called ZenyumSonic, and a more advanced aligner for complicated cases, ZenyumClear Plus. Data providers now record an acquisition or merger in early 2026, though the buyer and terms are not clear from public listings. The lesson here is quiet. Artopé did not need a new technology. He needed a customer's grievance, one he had personally lived, and the nerve to ship a product within weeks. A founder who remembers what the pain felt like tends to price the cure differently. ## Endowus (Singapore) URL: https://startup.talk/singapore/endowus/ Founders: Gregory Van, Samuel Rhee Founded: 2017; Based in: Singapore; Sector: Wealth management, digital advisory platform Endowus likes to say it is aligned only with the best interests of its clients. Samuel Rhee, its co-founder and chairman, put it that way in October 2025, in a statement marking a milestone the company once would have found hard to imagine: more than US$10 billion in client assets. The line sounds like marketing until you remember what it is aimed at. In Singapore, much of ordinary wealth advice had long been sold through products that paid the seller. Endowus was founded in 2017 on the opposite idea: charge clients an open fee, take no hidden commissions, and let people see what they pay. Very little is public about the early life of Gregory Van, the chief executive, or of Rhee, and the company's own releases name neither a hometown nor a school. What the record shows is how the firm grew once the idea was in the market. It built a platform for both retail investors and the wealthy. It gained access to money Singaporeans cannot easily move around, such as retirement funds, and drew in family offices and institutions. In August 2023 it raised a US$35 million Series C and later extended it to US$52.5 million, by which time it said it managed over US$5 billion. Then it went across the border. The group now includes Carret Private, a Hong Kong multi-family office, and Hong Kong client assets tripled in a year. In October 2025 it closed a round of more than US$70 million led by Illuminate Financial, with Prosus, Citi Ventures and Asian family offices taking part. Total funding passed US$130 million. By the company's own count, it serves more than 100,000 clients, has delivered over US$1 billion in net investment gains for them, and says it saves them over US$50 million a year in fees. Those are company figures, and the release does not state revenue or profit. The lesson for a first-time founder is about whom you work for. Endowus won by choosing one side of a market where the incentives were tangled. Choosing a side is slow, and it costs you the easy sales. It also gives customers a reason to stay. ## StashAway (Singapore) URL: https://startup.talk/singapore/stashaway/ Founders: Michele Ferrario, Freddy Lim, Nino Ulsamer Founded: 2016 (launched Jul 2017); Based in: Singapore; Sector: Digital wealth management, robo-advisory In 2016, Michele Ferrario typed two words and a city into Google: robo-adviser, Singapore. He found nothing. Ferrario is originally from Italy, with a background in finance, and he was working in Singapore with a good salary. He was saving well, but the money was sitting in cash. He wanted something more sensible: a diversified portfolio built on low-fee exchange-traded funds. He asked several Singapore banks. They mostly offered mutual funds with high fees. He had heard that in the United States, robo-advisers were growing fast. Here there seemed to be none. He later said the search was the moment: he Googled it and could not find anything. Before that, he had been chief executive of Singapore-based e-commerce platform Zelora, so he was no stranger to building a company. What they do show is who joined him. Freddy Lim took the investment chair as chief investment officer, and Nino Ulsamer became chief technology officer. The three spent about nine months building before they launched in July 2017. The product was meant to be plain. A saver answers questions, gets a portfolio matched to their risk level, and pays a modest fee. Ferrario has said the aim is to help many people build wealth through lower fees and regular investing. The backing followed. Square Peg led a US$16 million Series C in July 2020. Then in April 2021, Sequoia Capital India led a US$25 million Series D, with Eight Roads Ventures and Square Peg returning. The platform had passed US$1 billion in assets under management that January. Part of the new money was set aside to buy back up to US$3 million of employee stock options, and the company was preparing a Thailand launch. A further Series D+ of about US$12.67 million shows up in company filings. The cost of a business like this is trust. People hand over their savings to an app only after they believe in the people behind it, and every market has its own regulator. There is a plain lesson in that first search. Ferrario did not need to predict the future. He needed to notice that something he wanted did not exist, and that he was not likely the only person who did. ## Bolttech (Singapore) URL: https://startup.talk/singapore/bolttech/ Founders: Rob Schimek Founded: 2020; Based in: Singapore; Sector: Insurtech, embedded insurance platform Rob Schimek is described in the press as an insurance veteran, and the label fits a man who spent his working life inside the industry before starting bolttech. Before 2020 he was chief commercial officer at FWD. One profile notes he trained as an insurance man, not a coder. That is the quiet irony of the company. Bolttech is called an insurtech, yet its founder came from the insurance side, not the software side. His bet was that insurance was a poor product to buy and a good one to bundle. A person buying a phone, a car loan or a flight will accept a cover if it appears at the right moment, without a form or a call. Bolttech was founded in 2020 in Singapore with Richard Li as chairman. The idea was to be the pipe: one platform that links insurers on one side with retailers, banks, telecoms and device-makers on the other. The early backing was large. Reports put the Series A at between US$180 million and US$210 million, and in May 2023 a Series B valued the company at US$1.6 billion. In December 2024 it announced a Series C expected to exceed US$100 million, led by Dragon Fund with Baillie Gifford and Generali. It closed six months later at US$147 million and a US$2.1 billion valuation, with Sumitomo Corporation and Iberis Capital joining existing backers such as Tokio Marine and MetLife. The numbers it gave are about reach. Annualised quoted premiums were roughly US$65 billion in 2025, up from about US$55 billion in 2023. It counts about 700 distribution partners, over 230 insurers and 6,500 products, with names such as Apple, Allianz, AXA, Progressive and Samsung on its list. The cost of this scale is that quotes are not sales, and partner and insurer counts have not clearly risen since 2023. The next test is Africa and North America. For a first-time founder the point is blunt. Schimek did not leave his trade to chase software. He took what he knew about a slow industry, found the place where it was clumsiest, and put a platform there. ## Biofourmis (Singapore) URL: https://startup.talk/singapore/biofourmis/ Founders: Kuldeep Singh Rajput, Wendou Niu Founded: 2015; Based in: Boston, US (founded in Singapore); Sector: Digital health, remote patient monitoring, AI care at home In August 2023, Kuldeep Singh Rajput quietly stopped being chief executive of the company he had co-founded eight years earlier. A spokesperson confirmed it to STAT. There was no farewell announcement. The news came about a month after Biofourmis laid off 120 people worldwide, 48 of them in the United States. That was a hard fall for a company that had become a unicorn only the previous year. Biofourmis had been founded in 2015 in Singapore, and search coverage says Rajput started it after leaving a PhD programme there, with Wendou Niu as co-founder. Little else about his childhood or education appears in public reports. The idea was to watch patients at home. Wearable sensors and software would track signs of trouble and flag them before a hospital stay became necessary. It was a pitch health systems and drug companies wanted to hear, and money followed. A US$300 million Series D in 2022, backed by SoftBank, General Atlantic and CVS Health, valued the company at US$1.3 billion. A further round of about US$20 million came some months later. Total funding was reported between US$445 million and US$464 million. Then the math caught up. Layoffs in July 2023 were described by the company as finding the right structure for what came next. A month later the founder left. A committee of executives ran the company while the board searched for a new chief. Ben Wanamaker of General Atlantic took a board seat, joined by Omar Ishrak, former Medtronic chief, and Sachin Jain of SCAN Health Plan. In October 2024, Biofourmis merged with CopilotIQ. One account said the deal joined CopilotIQ's consumer-facing business with Biofourmis's enterprise health-care customers; another called it an all-stock deal. The combined company was reported to be led by CopilotIQ's chief executive, David Koretz. Rajput reportedly transferred his shares to existing investors and started another health-tech company. What a founder can take from this is uncomfortable but useful. A big valuation is a promise to grow into it, and when growth slows the first person asked to carry that promise is often the one who made it. ## Spenmo (Singapore) URL: https://startup.talk/singapore/spenmo/ Founders: Mohandass Kalaichelvan Founded: 2019 (Y Combinator lists June 2019; Summer 2020 batch); Based in: Singapore; Sector: Spend management, business payables and corporate cards (fintech) Before Spenmo, Mohandass Kalaichelvan spent his working life inside two places that teach you how to size up a market. He started as a management consultant at Boston Consulting Group. Then he moved to Grab, where he led regional strategy and helped launch financial products across Southeast Asia. Little is public about where he grew up or studied, but that résumé explains a lot about the bet he made next. Spenmo was founded in Singapore in 2019 and went through Y Combinator's Summer 2020 batch, in the middle of a pandemic that shut offices and sent finance teams home. Small companies in the region were still paying suppliers by bank transfer, chasing paper receipts and closing their books late. The company set out to put payables, cards and spending controls in one tool. The pitch it made to customers was practical, not poetic: save time. The company has said its users save over 50 hours and US$10,000 a month, and claims books close 90 percent faster. Those are the company's own figures. What it was really selling was a quieter finance department. Investors bought in fast. Seed money came from Y Combinator, Flash Ventures, Rocket Internet, Iterative Capital and the XA Network. In September 2021 Insight Partners led a US$34 million Series A, with Addition, Salesforce Ventures and Alpha JWC among the backers. By then the team had grown past 60 people. Early in 2022 came a Series B reported at US$85 million, led by Tiger Global, and press reports put the valuation near US$500 million. That speed carries a price. Money raised at a peak has to be justified in a market that turns. Spenmo competed with other regional players such as Aspire and Volopay, and with global cards from Brex. A product that is easy to explain is also easy to copy, and a payables company holds other people's cash and trust, which means compliance has to keep pace with growth. In June 2023 Kalaichelvan stepped down as chief executive and moved into an advisory role, keeping his board seat. Justin Choi, a former private equity investor, was named to succeed him. The thing worth taking from Kalaichelvan's path is the order of his experience. A consultant learns to read a market; an operator at a regional giant learns how people actually pay in it. He combined the two and picked a problem that was boring, universal and sat next to money. Boring problems tend to have the most patient customers. ## Ezbuy (Singapore) URL: https://startup.talk/singapore/ezbuy/ Founders: He Jian, Wendy Liu, Xue Bin, Qiao Lizhi Founded: 2010 (as 65daigou; rebranded ezbuy in 2016); Based in: Singapore; Sector: Cross-border e-commerce, buy-for-me service The first website for 65daigou was not coded. It was drawn. The founding team sketched the interface in Paint, the plain Windows drawing program, and only later paid developers to build it properly. The company was named for Singapore's dialling code, 65, plus the Chinese word for a personal shopper. It began in 2010 as a favour turned business. Ezbuy's cofounders were a few Chinese students and young graduates in Singapore. After a trip to China, one of them, He Jian, was struck by how much more choice was on Chinese shopping sites. He suggested the same thing for Southeast Asian shoppers: tell us what you want, and we will buy it, inspect it, ship it and handle the returns. Wendy Liu said yes. They started from his apartment, and in the early days they packed and delivered orders by hand. Liu's route there was unusual. She is an only child who moved from China to Singapore at 17 and has described herself as having been quite a little princess at home. She graduated from NTU with first-class honours in electrical engineering and a business minor, taking extra e-commerce modules, then did a Master's at NUS while working full-time. That meant studying by herself after hours. What nobody outside the business paid much attention to was how hard the model was to scale. They owned no marketplace. They stood between a customer and sellers on someone else's site, and Taobao accused them of scalping. Every order was a small act of trust. The company turned a corner by becoming a hub rather than a middleman. IDG Capital and CGC Capital backed a Series A in 2015. In 2016 it rebranded to ezbuy and took a US$20 million Series B led by Vision Knight Capital, then added Prime membership and a quick-delivery service. By 2018 it said it had more than 3 million loyal fans and 10,000 sellers across six marketplaces, and raised a US$17.6 million pre-Series C. Then came the squeeze. Lazada pushed hard at the same shoppers, and in 2018 NASDAQ-listed LightInTheBox bought ezbuy for a reported US$86 million, by one reported figure. He Jian became LightInTheBox's chief executive. The founders' path is useful because it is so unpolished. They began with a favour, a flat and a drawing program. Customers were not buying software; they were buying someone's promise to stand behind an order. ## Fazz (Singapore) URL: https://startup.talk/singapore/fazz/ Founders: Tianwei Liu, Hendra Kwik Founded: Xfers 2015; Fazz Financial Group formed Mar 2021; Based in: Singapore; Sector: Payments, SME financial services, lending (fintech) In early 2020, two founders who had both been through Y Combinator started talking about a merger. Tianwei Liu ran Xfers in Singapore. Hendra Kwik ran Payfazz in Indonesia. Their companies solved different halves of the same problem, and a year later they announced one group. Liu's route was an engineer's. He studied computer and electronic engineering at the National University of Singapore, graduating in 2010, and worked as a software engineer, reportedly including a stint at Amazon's Lab126 hardware lab. The company's own account says that he saw the trouble of moving money first hand, which led him to start Xfers. Little else about his childhood or family is public. Xfers began around 2015 as payment infrastructure: a way for businesses in Singapore and later Indonesia to send and receive money without the friction of bank forms. That is unglamorous work. Regulators set the rules, banks control the rails, and each market has to be licensed separately. Payfazz had built the other piece, a network of agents who offer financial services to micro and small businesses and to people who have never had a proper bank account. In March 2021 Payfazz put US$30 million into Xfers and the two formed Fazz Financial Group. Kwik became group chief executive and Liu his deputy; each kept running his own company. The merger had a plain logic. Xfers had payments technology and a Singapore licence. Payfazz had reach into small Indonesian merchants. Together they described a platform for payments, financing and digital assets aimed at the businesses that banks serve poorly. Series B backers included Insignia Venture Partners and B Capital, with Tiger Global in the round. In September 2022, the group raised a US$100 million Series C to build lending and an all-in-one app, and said it served more than 150,000 businesses. The cost is the one every combination pays. Two teams, two regulators, two cultures and a lending book that is only as good as the economy. Liu also went on to lead StraitsX, a stablecoin company, which suggests his own attention moved toward digital assets. Fazz's recent revenue and profit are not public. For a founder, the useful detail is the timing and the humility of it. Liu and Kwik did not wait to be acquired. They spoke as equals, in the middle of a pandemic, and chose to be one company with a bigger map. ## Syfe (Singapore) URL: https://startup.talk/singapore/syfe/ Founders: Dhruv Arora Founded: 2017 incorporation; launched July 2019; Based in: Singapore; Sector: Digital wealth management, robo-advisory, brokerage Dhruv Arora's mother gave him the advice that millions of Indian parents give: keep your money in the bank. He did. Then he realised, in his own words to Fortune, that the savings did absolutely nothing. That quiet discovery sits at the centre of Syfe. Arora grew up in India and later became part of the early team at Grofers, the grocery delivery startup now known as Blinkit, where he worked on product and growth. In 2008, soon after the global financial crisis, he joined UBS as an investment banker in Hong Kong. There he noticed something that bothered him. Doctors, lawyers and consultants, people who were very good at their own work, were building wealth without knowing how to manage it. Money sat idle. His mother's rule, scaled up to a professional class. He started Syfe in Singapore, incorporated in 2017 and launched to the public in July 2019. The plan was a robo-adviser with a human touch: portfolios built from low-cost funds, with advisers who used to work at big banks on hand for a conversation. Only a few months later Covid hit and markets lurched. Arora has said the pandemic acted as a catalyst for a shift in investor behaviour. What nobody expected was that fear would send savers toward investing rather than away from it. Valar Ventures, Peter Thiel's firm, led an US$18.6 million round in September 2020 and kept backing Syfe through later rounds. A US$27 million Series C1 followed in August 2024, and a second tranche in 2025 brought the Series C to a reported US$80 million. The company has kept widening its reach. It operates in Singapore, Hong Kong and Australia, and in April 2025 it bought the Australian broker Selfwealth for A$65 million. Syfe reports about US$10 billion in client assets as of 2025 and told Fortune it reached profitability in the fourth quarter of 2025. The price of this kind of business is patience. Fees are small, and a customer must trust a stranger with savings. Competing with bank-owned platforms means being clearer, not louder. Arora's idea began as a family rule he stopped following. Some of the best products begin as a quiet disagreement with advice you were given by someone who loved you. ## Airwallex (Singapore) URL: https://startup.talk/singapore/airwallex/ Founders: Jack Zhang, Max Li, Lucy Liu, Xijing Dai, Ki-Lok Wong Founded: 2015; Based in: Singapore and San Francisco (founded in Melbourne, Australia; moved headquarters to Hong Kong in 2018); Sector: Cross-border payments and global financial infrastructure The idea for Airwallex began in a coffee shop in Melbourne. Jack Zhang, a software engineer, and Max Li, an architect, had put money into the cafe. When they tried to pay overseas suppliers, they found how slow and expensive it was to move money across borders. The irony was that Zhang had spent years designing foreign exchange trading platforms for two of Australia's biggest banks, NAB and ANZ. He knew what the plumbing looked like from inside. In 2015 they founded Airwallex in Melbourne with three others: Lucy Liu, Xijing Dai and Ki-Lok Wong. The five put in about US$1 million of their own money. Liu and Dai were University of Melbourne alumni. Little is public about where any of them were born. The early years were the unglamorous kind. The team had to build rails to local banks around the world. In 2018 the company raised about A$80 million, then one of the largest startup rounds in Australia. It moved its headquarters to Hong Kong that year, and it now lists Singapore and San Francisco as its headquarters. So Airwallex is a Melbourne-born company that grew up in Asia, and calling it a Singapore startup is only partly accurate. It did not stay quiet. Soon after the 2018 round, NAB cancelled transactional banking services for Airwallex customers. The company also turned down a roughly US$1 billion acquisition offer from Stripe. In 2019 Hong Kong police froze about US$26 million amid a money-laundering probe. A court dismissed the suspicions in 2021 and US$18.2 million was released. Then it took off. Valued at US$1 billion in March 2019, it raised again at US$5.5 billion in 2021. Annualised revenue reached US$720 million by March 2025, up 90 percent in a year. In October 2025 the company reported annual revenue of US$1 billion and US$235 billion in transaction volume. A December 2025 round led by Addition valued it at US$8 billion. The pressure has not eased. In January 2026 Australia's financial crime regulator ordered an audit over anti-money-laundering compliance, and US lawmakers raised questions about data and China links. For founders, the lesson sits in the order of events. The company began as a personal irritation, and they stayed with it through a bank cutting them off and an acquisition offer they refused. A payments company has to earn trust again every year. ## Glints (Singapore) URL: https://startup.talk/singapore/glints/ Founders: Oswald Yeo, Seah Ying Cong, Looi Qin En Founded: 2013 (founded Aug 2013; formal launch often dated 2015); Based in: Singapore; Sector: Talent and recruitment platform / HR tech In 2014, Oswald Yeo and his co-founders did something most hungry startup founders never do. Investors offered them more money than they had asked for, and they said no. They had aimed for about US$400,000, and they turned away roughly US$250,000 of it, saying they did not want to raise so much just yet. It was a strange thing for people in their early twenties to do, and that is exactly what they were. Yeo, Seah Ying Cong and Looi Qin En had met at Hwa Chong Institution, one of Singapore's elite schools, and had gone on to universities abroad. Then they paused their studies to chase something that began very small. In August 2013, the platform had just 38 members. The idea was to help friends find internships at young companies, and in the early days it was little more than a way of linking people through a spreadsheet. That modesty ran through the first years. By April 2015, Glints had more than 7,000 active members, and a seed round of US$475,000 led by East Ventures had given the three founders room to work. But a bigger problem than money was a market that did not yet care. Young graduates in Southeast Asia wanted jobs, employers complained they could not find good people, and nobody had built something that served both sides across borders. What nobody outside the company saw was the cost of being that young and that close. In 2017, Looi left. He said it was down to creative friction. He went back, finished his degree, and later joined Boston Consulting Group. A company founded by school friends had lost one of its three originals while it was still small. The turn came when Glints stopped being a place to find internships and became a hiring engine. It charged employers only after a successful hire, which made it easy for startups to try. In 2018 the listed Singapore recruiter HRnetGroup invested. In 2019 it raised US$6.8 million to enter Vietnam and Hong Kong. In 2021, Japan's PERSOL led a US$22.5 million round, then the largest into a career platform in Southeast Asia. In 2022 came US$50 million more, bringing the total to roughly US$80 million. Glints today runs recruitment, a training academy and a community for young professionals across several markets. It has not gone public, and its revenue is not published. The odd lesson sits back in 2014. Yeo's team treated money as something to be earned in steps, not grabbed in the first excited week. They took less than they were offered, built for a year, and went back when they could show members and hires. A founder who can say no to cash that is on the table is usually a founder who knows exactly what that cash is for. ## Ohmyhome (Singapore) URL: https://startup.talk/singapore/ohmyhome/ Founders: Rhonda Wong, Race Wong Founded: 2016; Based in: Singapore; Sector: Proptech / real estate services Before she ran property deals, Race Wong stood on stage. From 2002 to 2009 she was half of 2R, a Cantopop duo she formed with her sister Rosanne. It is not the usual start for a company that would end up on the Nasdaq, and she has spoken about it less than most founders speak about their old jobs. She was born in Kuala Lumpur and raised in Singapore. As children, she and her sister Rhonda moved back and forth between the two countries, and the sisters say that restlessness is what drew them to property. Home was something they had to keep re-finding. Later Race studied for a Master of Business in marketing at the University of Newcastle in Australia, finishing in 2013. In 2011 she was a marketing manager at Maybank Kim Eng in Hong Kong and licensed to deal in equities. She told one interviewer that she wanted to put her energy into something that gave people more value. In 2016, with Rhonda as CEO and Race as COO, she started Ohmyhome. The pitch was plain. Buying or selling a flat in Singapore is the biggest money decision most families make, and agents take a hefty cut. An online platform could cut some of that cost and add mortgages, renovation and legal help on the side. By 2023 the company said it had supported 11,000 transactions. Then came the unusual choice. In March 2023, in a slow market for new listings, Ohmyhome became the first Singaporean company to list in the US that year. It sold about 3.6 million shares at US$4, raising roughly US$11 million. The sisters said they needed capital and picked a public listing over a private round, partly for the brand recognition it would bring with developers and customers. What the sisters could not control was the scrutiny that followed. The prospectus disclosed that a transaction with a family member made up much of 2022 revenue, and it carried auditors' substantial doubt about the company's ability to continue. Without that transaction, first-half 2022 revenue growth was 2%, not the 45.8% quoted elsewhere. After listing, the shares fell far below the US$4 offer price. In 2025 the company consolidated its shares ten to one and sold its main operating subsidiary for a single US dollar, after waiving about S$19 million owed to it, to concentrate on digital marketing. Race said of her sister, 'It never crossed our minds not to work together.' That loyalty built the company and carried it into a harsh public market. The lesson for founders is plain. Going public gives you a brand and a bigger audience, but it also puts every related-party line in your accounts under a lamp. ## Seedly (Singapore) URL: https://startup.talk/singapore/seedly/ Founders: Tee-Ming Chew Founded: 2016; Based in: Singapore; Sector: Personal finance community / fintech Money questions are awkward to ask out loud, and that is where Seedly found its footing. The Singapore app began as a personal finance service for millennials and young people across Southeast Asia, and it let them track spending and ask strangers what to do with their first salary. The company started around 2016, when it took seed money from East Ventures and NUS Enterprise. It also came out of the second batch of DBS's HotSpot pre-accelerator, a programme run by the country's largest bank for very early teams. A bank that was happy to nurture a startup that would explain budgeting to its future customers says something about how seriously Singapore took young people's financial literacy. The public record on the people is thin. Tee-Ming Chew is named as co-founder and CTO, and Ming Feng Yeap held a marketing role. Beyond that, the founders' schools, childhoods and early jobs are not clearly documented in what we could read, and rather than guess, this profile sticks to what can be checked. What is clear is the arc. In May 2018, Seedly became the first acquisition by ShopBack, the e-commerce cashback company that was already among the best known startups in Southeast Asia. The terms were not disclosed. The whole team moved across, and Seedly kept operating as a business under ShopBack. It was an exit for the founders, their team and the investors, about two years after the company began. Then the app changed hands again. In October 2020, CompareAsiaGroup, a financial comparison company based in Singapore and Hong Kong, bought Seedly from ShopBack, with plans to expand the community beyond Singapore. The product that began as a student project had been through two corporate owners in under three years. There is a quiet lesson in that for anyone building a free community product. The audience is the asset, and big companies buy audiences. Seedly's founders did not wait for a perfect business model. They built something that young people actually used, got into a bank's accelerator, and sold to a buyer who needed trust with a younger crowd. The harder question, which the next owner faced, was how to earn money from people who came for free advice. ## 99.co (Singapore) URL: https://startup.talk/singapore/99-co/ Founders: Darius Cheung, Yan Phun, Chong Ming Hwee Founded: 2014 (began as Homey); Based in: Singapore; Sector: Property portal / proptech Darius Cheung has described himself at 28 as very young and naive. By then he had already done something most founders dream of. His mobile security company, tenCube, had been sold to McAfee for US$25 million. tenCube began in 2005 with two co-founders, Varun Chatterji and Rishi Israni. Its main product came from a simple pain, when Chatterji lost a phone and the data on it. Cheung, who had studied engineering at the National University of Singapore and taken part in its Overseas Colleges programme, had been interested in building businesses since the early internet boom, when he was about 17 or 18. The three sold when they felt they needed a bigger company behind them to serve the large mobile carriers. After the sale he stayed on at McAfee for two years. Then he tried again. He co-founded BillPin, a bill-splitting app. The team shelved it because people simply did not split bills often enough to support a mobile wallet. It was a small defeat after a big win. What almost nobody remembers is what 99.co started as. It was Homey, a roommate-matching site he built with Yan Phun and another friend from school. Only when the team looked closely at how renting and buying worked did they see how broken the property market was, and they turned Homey into a portal. The company launched as 99.co in 2014 with seed money of about US$560,000, then got US$1.6 million from Sequoia and Facebook co-founder Eduardo Saverin. The early money brought its own trouble. Cheung has said easy funding created pressure to grow faster than the team was ready for. The company then had to win over property agents who did not trust a new site, and to rebuild its product for Indonesia, where he has called the experience a rude shock. It worked anyway. In 2019 a US$15.2 million Series B, led by Allianz X and MindWorks Venture, funded expansion. Between early 2019 and 2022, sales grew more than eightfold. Traffic reached 30 million a month, and a US$52 million Series C pushed total funding past US$80 million. Cheung now invests in other founders, including Carousell and Glints, and says the people and the culture matter more than the product. A man who had already sold a company, and failed with another, built his best-known one out of a roommate app that few people remember. ## Singlife (Singapore) URL: https://startup.talk/singapore/singlife/ Founders: Walter de Oude Founded: 2017; Based in: Singapore; Sector: Insurance / insurtech At 47, Walter de Oude was about to do something that no one in Singapore's insurance industry had done in years. He was proposing to fold a business almost fourteen times his own size into a company he had started only three years earlier. He was born in South Africa and had lived in Singapore for more than fifteen years. Before he built his own insurer, he ran HSBC's insurance business in the city, which meant he knew exactly how slow and crowded a market dominated by a few big names could be. In 2017, he set up Singapore Life, called Singlife, and it was described as the first local insurer licensed by the Monetary Authority of Singapore since 1970. The early customers were not ordinary households. The company grew mostly through a digital platform and aimed first at private banking clients and wealthy individuals, a narrow niche where a new brand could earn trust. In 2018 it bought the Singapore portfolio of Zurich Life, a first sign that de Oude would grow by acquisition. The bigger move came in 2020. Singlife agreed to merge with Aviva Singapore in a deal valued at about S$3.2 billion. Before the merger, Singlife had around S$800 million in assets. Aviva's Singapore business held about S$11.8 billion, and came with 1.5 million customers and a large network of financial advisers. After the deal TPG held 35%, Aviva 25% and Sumitomo Life 20%. The combined business, called Aviva Singlife, became the largest insurer in Singapore, and de Oude became deputy chairman. He told Reuters he hoped to expand across Southeast Asia and eventually list the company in Singapore, though he gave no timeline. The path was not clean. Aviva decided in 2023 to leave the venture and sell its 25.9% stake and two debt instruments to Sumitomo Life, a reported S$1.4 billion deal, though the figure differs between sources. De Oude had moved to a non-executive founder director role on 1 January 2022, handing day-to-day management to Pearlyn Phau. In February 2026 the company launched physical gold-backed investment-linked policies. How big the founder's stake ever became is not public. What is public is the sequence: a former bank executive started small, aimed at a narrow group of rich clients, bought a portfolio, then bought something far larger. For anyone building in a market with entrenched giants, the lesson is that you may not win by out-competing them. You can sometimes win by buying their business. ## Sunseap (Singapore) URL: https://startup.talk/singapore/sunseap/ Founders: Frank Phuan, Lawrence Wu Founded: 2011 (Sunseap Leasing); Sunseap Group holding company later; Based in: Singapore; Sector: Renewable energy / solar The price talks happened at home. Over dinner, Frank Phuan's father would haggle with him about what a solar panel should cost, because the father had become the supplier to his son's company. It was a strange arrangement, and it began with an argument. Frank graduated from Nanyang Technological University in 2001 with a materials engineering degree and walked into his father's business. His father had started selling solar panels in 1979, when almost nobody in Singapore did. Frank saw what he thought was a flaw. The firm sold panels once and the money was gone. Why not sell the electricity the panels produced and be paid every month? His father said no, and gave reasons. Singapore had little space for panels, and government incentives did not yet cover the high costs. Frank did not get his way inside the family firm. He kept the idea, and ten years later acted on it. In 2011 he and fellow NTU graduate Lawrence Wu set up Sunseap Leasing. What almost nobody sees from outside is the legal wrinkle that shaped it. A startup could not simply sell electricity in Singapore, because it held no licence. So Sunseap leased the equipment and charged customers by how much energy they used each month. The workaround became a business model. It was expensive. Rooftops needed big bank loans, and investors had to be persuaded that a long payback was worth waiting for. Frank has said he sometimes doubted himself, because success meant a smaller slice of his own company. He told The Peak that in this job you are always $10,000 from being a millionaire and one dollar from being broke. He also kept Sunseap separate from his father's business, even while buying panels from it. The turn came as Singapore's buildings began to fill with panels. By 2022 Sunseap had equipped more than 3,000 buildings, including public housing and company rooftops, with customers such as Apple among them. In February that year, EDP Renewables bought a 91 percent stake in a deal worth S$1.1 billion. Phuan stayed on as chief executive and Wu as chief financial officer. The price he paid was a reduced share of the company he built, the very thing he had worried about years earlier. The lesson is in the dinner-table detail. Frank's first idea was rejected by the person he most respected in the industry. He did not fight the rejection. He went around it, and the supplier relationship that began as friction stayed alive. A good idea can be stopped by an authority you love, and still be right. ## Ascenda (Singapore) URL: https://startup.talk/singapore/ascenda/ Founders: Kyle Armstrong Founded: 2013-2014; Based in: Singapore; Sector: Loyalty technology / fintech Most people who have redeemed card points for a hotel night never think about the software behind the button. Ascenda built that software, and its story starts under a different name. The company was known as Kaligo Solutions before it became Ascenda. The name change is itself a signal about where the business moved. Kaligo was about travel redemption: letting someone turn the points on a bank card into flights and rooms. Ascenda widened that into a broader loyalty platform for banks, card issuers and other companies that hold rewards on behalf of their customers. The man most closely tied to it is Kyle Armstrong, listed as founder and chief executive. What can be said is about the business. Loyalty programmes are expensive for banks to run, and many of them are not their core skill. Ascenda sells the plumbing: a catalogue, a redemption engine, partner offers and the integration into the bank's own app. In July 2023 it raised US$7 million in growth equity led by Claret Capital Partners, a modest but telling cheque for a company that appears to have grown without large venture rounds. It later announced a loyalty-as-a-service offering, aimed at changing the economics for financial institutions, and a partnership with Cardless on co-branded card programmes. Those announcements describe a company moving from selling travel inventory to selling a whole operating model. For a first-time founder, the useful point sits in the name change. Kaligo started with one product and one use, and the company kept the customers and the integration work while changing what it sold on top. Small, well-defined infrastructure can outlast the first product built on it. Ascenda is a reminder that the quiet business behind the consumer brand is sometimes the one that lasts. ## Doctor Anywhere (Singapore) URL: https://startup.talk/singapore/doctor-anywhere/ Founders: Lim Wai Mun Founded: 2016 (idea); launched 2017; Based in: Singapore; Sector: Digital health / telemedicine In 2016 Lim Wai Mun was handing out meals to elderly people. He was not thinking about a company. He noticed that many of the seniors could not get to a doctor at all. He was born and raised in Singapore and studied mechanical engineering at the National University of Singapore. He started in private equity at Standard Chartered, then spent close to a decade at Temasek, where he helped set up Pavilion Energy, an LNG company. It was a solid, respectable career and had nothing to do with medicine. The meal rounds changed his plan. His first idea was simple: connect homebound seniors with doctors who would visit them for free. He gave up a year to try, taking an unpaid sabbatical to build the early service. The first version did not work. Doctors were already stretched by their own patients, and it was hard to get them to take on extra visits for nothing. What almost nobody expected was that the failure pointed to the real business. Instead of serving only seniors, he turned the service into basic care for anyone, delivered first by video. Telehealth in Singapore in 2017 was new, and he met scepticism and outright rejection from patients and from parts of the healthcare industry. The problem was trust. Patients wanted to know the doctor was real and the advice safe. The answer was to open physical clinics, so a screen consultation had something solid behind it. Growth followed. In August 2021 the company raised S$88 million in a Series C led by Asia Partners, and in December that year it added US$38.8 million and bought a listed clinic group. Lim was named EY Entrepreneur of the Year in Singapore in 2022. By late 2023 and 2024, Doctor Anywhere reported more than 2.5 million users in six countries and had raised another US$40.8 million. The price was the pivot itself: the founder's original, charitable idea was dropped. The seniors who inspired it were not the customers who built the company. For a founder, the lesson is plain. A moving, personal reason can start a company without being the model that sustains it. Lim kept the reason and let the model change. ## Wiz.ai (Singapore) URL: https://startup.talk/singapore/wiz-ai/ Founders: Jennifer Zhang, Jianfeng Lu, Tony Zhu Founded: 2019; Based in: Singapore; Sector: Conversational / voice AI The early testers hung up. That is how Jennifer Zhang's first talkbots met the public: a person picks up the phone, hears a machine, and ends the call. The idea behind them had started somewhere much smaller. Zhang has said she watched her mother struggle with online customer service, stuck in a system built for fluent English speakers. In Southeast Asia, over 630 million people speak more than 1,000 languages and dialects, and she thought a machine should adapt to them, not the other way around. Her route to that idea was unusual. She majored in film at Keio University in Japan and Zhejiang University in China, then took a Master of Public Policy at the University of Southern California. In the US she was a columnist for an entrepreneur magazine, and her interviews included Mark Zuckerberg and Jessica Alba. She then spent nearly three years in venture capital, as Managing Partner of GCC Capital in Beijing and General Partner of Cybernaut Zfounder Ventures. What almost nobody sees is why she left. She felt venture capital was too passive. She wanted to build instead of watching others do it. In early 2019, with two partners, Jianfeng Lu and Tony Zhu, both researchers in language technology, she co-founded Wiz.ai and chose Singapore as the base. The first stretch was trial and error. Beyond the hang-ups, Southeast Asian speech mixes languages in the same sentence, and Singlish is its own challenge. The company worked at making the bots sound and listen like locals. In 2020 it raised US$6 million in a pre-Series A led by GGV Capital, and saw 200 percent growth that year. Hillhouse Capital led a US$20 million Series A, and the team grew from three people to almost 300 across 17 countries. In November 2025 the company closed a Series B in the tens of millions of US dollars, led by SMBC Asia Rising Fund. The idea she carried from her mother survived every redesign. A founder can borrow a problem from a family kitchen and still build a company around it. The trick is to keep listening to the person who first had the problem, even after the bot has learned to talk. ## YouTrip (Singapore) URL: https://startup.talk/singapore/youtrip/ Founders: Caecilia Chu, Arthur Mak Founded: 2017 (founded); 2018 (app launch); Based in: Singapore; Sector: Fintech / multi-currency wallet and payments When Caecilia Chu was five, her father tried to borrow money to start a business. It was hard for him to get the loan. She has remembered it, and it shaped what she built decades later. She grew up in public housing in Hong Kong. Her father was a postman and her mother a kindergarten teacher. Money was not abundant, but scholarships took her to top American universities, including the Wharton School, and later an MBA at Harvard Business School. She worked at McKinsey, Citi and two Chinese fintech firms. On paper it was a strong climb. In 2017 she left the corporate track. With her schoolmate Arthur Mak, she started YouTrip in Singapore. It began as a two-person operation. What they wanted to fix was an everyday irritation: foreign exchange fees and opaque conversion rates paid by anyone who travels. A product built around a multi-currency wallet and card could beat the banks on price. The launch answered the question quickly. Within 48 hours of going live, YouTrip had 20,000 users. That is an unusual start for a fintech with no large marketing machine. Investors followed: a pre-Series A in 2019 led by Insignia Ventures Partners, US$30 million in Series A in 2021, and a US$50 million Series B in October 2023 led by Lightspeed. Earning came from merchant fees of 1 to 3 percent on overseas and international purchases, plus interest on user balances. In 2022 YouTrip recorded its first net profit and launched YouBiz for small businesses, which onboarded over 3,000 local SMEs in its first year. It stayed profitable in 2023 and 2024. The cost of this path is mostly hidden. The sources do not describe the personal price of the early years, and we will not invent it. We know that the founders kept the Series B money untouched into 2025, planning a Series C to hold a cash buffer, and that Chu said an IPO could be a few years away. By mid-2025 YouTrip had about 250 people and processed about US$10 billion a year in transactions, eight years after Chu left her job. For first-time founders, the lesson sits in the five-year-old memory. Chu did not start with a market map. She started with something she had watched a parent struggle with, and picked a product that removed that struggle for strangers. ## Oddle (Singapore) URL: https://startup.talk/singapore/oddle/ Founders: Jonathan Lim, Pua Yong Xiang, Jeremy Lee Founded: 2014 (launched March 2014); Based in: Singapore; Sector: Food and beverage technology / restaurant ordering and marketing software In September 2020, while restaurants across Singapore sat empty, three founders were doing something they did not want to talk about. Jonathan Lim, Pua Yong Xiang and Jeremy Lee had hired an adviser and begun speaking to potential buyers for Oddle, the company they had started six years earlier. Bloomberg reported it, and the founders stressed that talks were early and they might simply keep the business. It is not hard to see why selling was on the table. Oddle had launched in March 2014 as a subscription tool that let restaurants take orders online under their own brand, not under someone else's. The pitch was good. The numbers were slow. For the year ended March 2018, revenue was S$1.8 million, up 57% on the year before, but the company lost S$2.3 million. Across three years it had raised only about S$5 million from backers such as SPH Ventures, Wavemaker Partners and Quest Ventures. Small money for a company trying to out-run giant delivery apps that charged restaurants heavy commissions. The early model had also needed rethinking. Oddle started as a white-label e-commerce provider for food and beverage operators, and over time widened into something larger: e-shops, reservations, QR ordering, payment terminals and managed marketing. The idea was to help a restaurant own its customers, not rent them from a delivery platform. Then came the part that nobody could have planned. The pandemic shut dining rooms and sent owners hunting for any way to sell direct. Oddle's restaurant partners went from US$21 million of sales through the platform in 2019 to US$144 million in 2021. The company reported US$18 million in revenue for the year ended March 2021, which it described as 17 times the year before. The founders who had been shopping the business now had a business worth keeping. In May 2022 Oddle closed a US$5 million round led by Altara Ventures, and launched Oddle Eats, a food discovery app. It said it served more than 5,000 restaurants, cafes, bars and hotels, from Les Amis Group to a local PizzaExpress franchise, and planned to push deeper into Malaysia, Taiwan and Hong Kong. The lesson is uncomfortable and useful. Six years in, with small revenue and steady losses, these founders reached for the exit door. They were not wrong to look; they were wrong only about what the market was about to do. Staying alive long enough, on thin money, kept the company in the room when the room suddenly changed. The founders never claimed to have seen it coming, and that is the honest part: sometimes a company is saved by a decision to wait one more year. ## Matrixport (Singapore) URL: https://startup.talk/singapore/matrixport/ Founders: Jihan Wu Founded: 2019; Based in: Singapore; Sector: Crypto financial services / digital asset wealth management In 2011, Jihan Wu was a young financial analyst at a private equity firm in China, and he made a strange request of the people closest to him. He asked family and friends for money. About 100,000 yuan, in total. He used it to buy 900 bitcoins. Wu was born in 1986 in Chongqing and went to Peking University, graduating in 2009 with degrees in economics and psychology. A safe career lay ahead of him. Instead he fell into Bitcoin, and the people who lent him the money were backing a bet on something most of them could not have understood. He then co-founded Babite, China's first Bitcoin community website, and translated Satoshi Nakamoto's white paper into Chinese. Not every early bet worked. In 2012 he invested in the mining firm ASICMINER, and he lost money on hardware from Canaan Creative. The lesson he took from burning cash on other people's machines is not recorded, but what he did next is. In 2013 he co-founded Bitmain with the chip engineer Micree Zhan, and the company grew into the world's largest maker of bitcoin-mining chips, reporting US$2.5 billion of revenue in 2017. The success had a cost. Wu backed raising bitcoin's capacity, a stance that helped lead to the August 2017 split that created Bitcoin Cash. Critics gave him a nickname, a pun on his given name, and he denied having that much influence. Later reports describe a power struggle with Zhan. In 2019 Wu stepped down as co-CEO, and he left Bitmain entirely in 2021. By then he had already started over. In 2019 he founded Matrixport, a crypto financial services company, in Singapore, because China had banned crypto trading. He now holds Singapore citizenship. In August 2021 the firm raised about US$100 million at a valuation above US$1 billion, bringing its total to about US$129 million. A year later he was reported to be seeking another US$100 million at US$1.5 billion. Through Bitdeer he also bought a Singapore vault for S$40 million in 2022. Hurun counted his 2018 wealth at 16.5 billion yuan, around US$2.4 billion. But the quieter detail is the first one: a salaried analyst asking his family for a small pile of cash for something he believed in. Everything after rests on that early willingness to be wrong in front of the people who love you, and to keep going. ## Gushcloud International (Singapore) URL: https://startup.talk/singapore/gushcloud-international/ Founders: Althea Lim, Vincent Ha Founded: 2011; Based in: Singapore; Sector: Influencer marketing / talent management and creator economy In November 2018, two founders wrote a bigger cheque than they had ever received. Althea Lim and Vincent Ha paid US$8.2 million to get back control of the company they had started. Three years earlier, they had sold 62% of Gushcloud to a South Korean firm for US$7.5 million. To understand why, go back to the start. Lim and Ha had earlier co-founded a digital agency, The Barnett Group, and in 2011 they began Gushcloud as a social sharing platform. It was officially launched in February 2012. Then the business found its true calling: representing the young bloggers and video creators that brands could not reach through ordinary advertising. Gushcloud signed talent, sold their reach to advertisers and grew fast. Growth had a price. In 2014 the blogger Xiaxue publicly accused the company of unethical practices, and Gushcloud answered point by point. In 2015 there were new allegations, tied to a telco campaign, that paid bloggers wrote negatively about rivals. The telco and Gushcloud first denied it, and later apologised. For a company whose product was trust in a face on a screen, it was a rough few years. Still, the sale to Korea's Yello Digital Marketing Global in 2015 gave the firm money and a push abroad. Between 2015 and 2018 it grew from three markets to nine, and more than 300 people worked there. In 2017 YG Group, the Korean entertainment house, invested US$3 million, with Japan's D2C Inc also joining. What is easy to miss is how the 2018 buyback changes the story. The founders did not hold on tight and refuse to sell; they sold, learned what it felt like to answer to a majority owner, and then paid more to take the company home. That is a different kind of control. After that they kept building: DSTNCT, a marketing agency, was acquired in February 2019, Ribbit, a micro-influencer network, launched in April, and Russell Simmons, co-founder of Def Jam Recordings, became President that September. A raise of about US$11 million followed, though other reports say US$15 million. MODA Creative was added in 2020. The closing thought is simple. Money from a strategic investor can speed a company up, and the same money can hold it. Lim and Ha took it, used it, and then decided that the stake was worth more to them than the price. Many founders sell and never get the chance to ask what the sale cost them. ## MoneySmart (Singapore) URL: https://startup.talk/singapore/moneysmart/ Founders: Vinod Nair Founded: 2009 (about six years before a 2015 article); seed funded 2011; Based in: Singapore; Sector: Personal finance comparison and fintech marketplace By 2015, Vinod Nair's company was drawing about 750,000 visits a month in Singapore and 650,000 in Indonesia, and he was describing his content as BuzzFeed meets personal finance. It sounded breezy. The road to that sentence was not. Nair had already built one startup before MoneySmart, a real estate search engine. That experience did not make fundraising easy. He later admitted that finding seed money for MoneySmart was hard. The idea was to help ordinary people compare financial products, starting with mortgages, at a time when little consumer information on loans and cards was available online. Banks held the data and the leverage. Investors wanted proof that a website could earn money by sending customers to them. The proof came slowly. A seed round of S$589,000 from Golden Gate Ventures and Singapore's National Research Foundation arrived in 2011, two years after founding. With that, Nair added credit cards and other products, and wrote headlines meant for people who find money boring. One of them compared retirement in Singapore to a big exam you did not study for. The second real test was expansion. In 2014 he entered Indonesia by acquiring KreditAja, rather than starting from nothing. He told TechCrunch he planned to move slowly and localise, with two new markets the following year and the Philippines, Hong Kong, Thailand and Malaysia on his list. Rivals were not small. CompareAsia had landed a US$40 million Series A from Goldman Sachs, a figure that dwarfed MoneySmart's own US$1.96 million round in October 2015, led by Singapore Press Holdings' investment arm with Convergence Ventures and OPT SEA joining. The next round was larger: S$14 million in 2017, led by Japan's Kakaku.com, the owner of a major price-comparison site. By then Nair's patient, content-led approach had found a backer who understood it. The revenue and profit numbers are not public, so the story ends on something softer. Nair's path reminds first-time founders that a second company does not automatically get an easier cheque. What helped him was a plain business: tell people the truth about money, and make the telling readable enough that they come back. Investors came later, once the audience was already there. ## Tech in Asia (Singapore) URL: https://startup.talk/singapore/tech-in-asia/ Founders: Willis Wee Founded: 2010; Based in: Singapore; Sector: Media / tech news, events, jobs and data The conversation that changed Willis Wee's early career did not happen in a boardroom. It happened at a McDonald's near the Singapore Management University campus, with Willson Cuaca, a co-founder of East Ventures. Wee was a student. He has since pointed to that talk as a turning point. He had launched Tech in Asia on 1 August 2010, in his third year at university. The gap was obvious to him: global media barely paid attention to the startups of Asia. He could have gone to Facebook or LinkedIn after graduating. Instead he chose to build his own business. At the start, it was a media site, and then more. He developed Techlist, a platform linking investors and startups in the region. From 2012 the company ran conferences each year, mostly in Singapore, Tokyo and Jakarta. In 2015 Facebook co-founder Eduardo Saverin backed it, and about a year before its subscription launch it raised US$6.6 million in a round led by Hanwha Investment and Securities. Then came the part that rarely makes the conference stage. In 2018, the company laid off staff, and Wee later spoke on a podcast about the mistakes that led to it. Around then he also launched a marketing agency and a paywall subscription (around 2017-2018). A media business in a small, crowded region has to find revenue from readers, advertisers and events, and none is easy. In January 2024, SPH Media completed its purchase of Tech in Asia. Wee said he hoped to gain from a partnership with The Business Times. But the years after were not smooth. In July 2025 the company wound down its Indonesian-language product and cut 18% of its workforce. On 31 March 2026, Wee and COO Maria Li resigned, Terence Lee took over operations, and from April the company was merged into SPH Media's Features and Lifestyle division. There is something poignant in the arc. A student who skipped the safe job built a regional voice in under sixteen years, and then handed it over. The lesson for founders is plain: the conversation with the right person, even over fast food, can matter more than a business plan. And a founder should know early what he is building toward, because the exit may arrive on someone else's schedule. ## ViSenze (Singapore) URL: https://startup.talk/singapore/visenze/ Founders: Oliver Tan, Chua Tat-Seng, Li Guangda, Roger Yuen Founded: 2012; Based in: Singapore; Sector: AI visual search / visual commerce Oliver Tan has said he was away on business when Lee Kuan Yew died, and that he cried at a memorial service in San Francisco, far from home. It is a small detail, but it explains a lot about a man who would spend a decade selling Singapore-made artificial intelligence to the world. Tan did not come from a laboratory. He took an MBA at Royal Holloway, University of London, and worked on the business side of media and internet companies: business development at e-Cop, later Quann, and at Pacific Internet, and then a spell as managing director of MNC Media in Indonesia. He was the person who knew how to sell. What he needed was something to sell. It came out of a university. In 2012, researchers at the NUS-Tsinghua NExT centre, among them professor Chua Tat-Seng and Li Guangda, had built technology that could look at a picture and find similar things. Tan joined with Roger Yuen, who became chairman, and the four of them started ViSenze. The founders put in the first money themselves. The pitch was concrete. A shopper sees a handbag, snaps it, and the retailer's site shows her something she can buy. In 2014 Rakuten Ventures led a US$3.5 million Series A. Two years later came US$10.5 million more. What almost nobody saw was the awkward part of being the front man. Tan has said that when he pitched to customers and investors in China, his partners discovered his Mandarin was not as strong as they had assumed. The company's biggest buyers and a good share of its ideas lived in Chinese-speaking markets, and its salesman was working in a second language he had overestimated. He laughs about it in interviews now. He has also said he used to work so hard he forgot to eat. The turn was the 2019 Series C of US$20 million, co-led by Gobi Partners and Sonae IM, which took total funding to about US$34.5 million. By then ViSenze was selling to retailers and smartphone makers, and Tan was describing the best moment of his startup life as the day investors committed money: in his words, "Let's go make it come true!!!" The price of staying independent in visual search was steep. Google, Amazon and Pinterest were all building the same thing for free. In 2025, reports say, ViSenze was acquired by Rezolve AI, and Tan went on to lead Rezolve's Asia-Pacific business. The lesson is in his own advice: do not chase the most original idea, he has said, aim to be much better than the original. A university can give you the science. It cannot tell you that your pitch has a language problem, and it takes a founder honest enough to say so. ## Mighty Jaxx (Singapore) URL: https://startup.talk/singapore/mighty-jaxx/ Founders: Jackson Aw Founded: 2012; Based in: Singapore; Sector: Collectibles / art toys / licensed merchandise The email was a mistake. In 2015, somebody at Warner Bros. sent Jackson Aw an internal message that was never meant for him. In it was the name and contact of the studio's global head of toys. Aw, by his own recollection, did not delete it. He wrote to her. He was a toy collector who had never stopped being one. Aw loved designer art toys as a polytechnic student, and after he graduated he became, by his own description, a full-fledged addict. At 18 he had gone to Shenzhen on a whim to look around toy factories and see how things were made. In 2012 he turned the hobby into a brand. He and an artist made a figure called Hell Lotus, and the first run was unveiled at Singapore Comic Con. One account says it was paid for with a S$20,000 loan and that he made 200 pieces. His parents, one profile notes, were hesitant, and had to be talked round. The next stretch was lean. The same profile says he went into debt and slept on the floor of his studio. Mighty Jaxx made the toys, boxed them, and shipped them, essentially a one-person operation. The business was alive, but it was not yet respectable. Nobody with a big licence trusted a tiny Singapore shop. What almost nobody knew was how the first big door opened. It was that accidental email. A stranger at a studio replied, a conversation began, and the small firm began making figures from the DC universe. Collaborations followed with F1, Hasbro and Toei Animation, and the XXRAY line, built on the artist Jason Freeny's dissected-toy designs, became a signature. Money followed the licences. A S$2.15 million pre-Series A came in 2019, then S$3.2 million in April 2020, then a US$10 million round led by Tencent in 2021. In March 2022, East Ventures led a US$20 million Series A+ at a reported valuation above US$200 million, bringing total funding to US$34.8 million. EY named Aw its Entrepreneur of the Year for consumer and lifestyle products in 2021. By then the company had offices in London and Los Angeles and had started to make NFTs and tech-enabled collectibles. The cost of the climb was visible in how he talks about it: debt, the studio floor, a hard first few years. It is hard to find a cleaner lesson in being ready. Aw did not get lucky with the email. He got lucky that, when it arrived, he had a factory network, a product and the nerve to hit reply. ## Beam Mobility (Singapore) URL: https://startup.talk/singapore/beam-mobility/ Founders: Alan Jiang, Deb Gangopadhyay Founded: 2018; Based in: Singapore; Sector: Micromobility / shared e-scooters Alan Jiang had seen the numbers before he saw the scooter. Working for Uber across Asia, he noticed that roughly four in ten trips were shorter than three kilometres. A car, he came to think, was the wrong tool for those. He had come to this by an unusual path. Jiang studied economics at Yale, worked as a consultant at Exeter Group, and in December 2012 joined Uber, where he ran launches in China, Malaysia and Vietnam until early 2017. Then he joined ofo, the bike-sharing company, as head of Southeast Asia. He had now been inside two of the most aggressive growth machines in modern transport. By May 2018 he was out of ofo, a company that would later collapse. Months later he had a new one. He had a ready-made partner: Deb Gangopadhyay, a Yale classmate who had built a marketing automation stack at his own startup, Push Nova, and who became Beam's CTO. In October 2018, Beam announced a US$6.4 million seed round from Sequoia India, Founders Fund, ZhenFund and Class 5 Global. The plan was Singapore first, then Malaysia, Australia and South Korea. What nobody could easily plan around was the rulebook. Governments across Southeast Asia changed their minds about e-scooters repeatedly. Singapore tightened rules; other cities banned and unbanned. In 2020, Beam raised US$26 million from Sequoia India and Hana Ventures, targeting South Korea, Australia, Malaysia, New Zealand and Taiwan, and it made a point of becoming Climate Neutral Certified after measuring and offsetting its 2019 emissions. Then came the big cheque: US$93 million in 2022, led by Affirma Capital, with operations in 35 cities. By 2023, the company was talking publicly about growth and profitability, a rare pairing in a business that had burned through investor money everywhere. In July 2025, Beam agreed to merge with Neuron Mobility, creating one of the largest scooter operators in the world with more than 100 cities. Neuron's Zachary Wang would be global CEO and Jiang CEO of Asia. A founder who had once run a market became part of one bigger one. There is a useful instinct in his path. Jiang did not start from a scooter. He started from a trip-length chart and a question about the right vehicle. Founders in crowded categories could do worse than start from a number only they have seen. ## Coinhako (Singapore) URL: https://startup.talk/singapore/coinhako/ Founders: Yusho Liu, Gerry Eng Founded: 2014; Based in: Singapore; Sector: Crypto exchange / digital assets In December 2014, two young Singaporeans flew home from California with something unusual in their luggage: a personal cheque from Tim Draper. Yusho Liu and Gerry Eng had just finished Tribe 4 of Adam Draper's Boost VC accelerator, a programme for bitcoin start-ups, and Draper senior, the famous venture capitalist, had decided to back them himself. The idea was simple and, in 2014, risky. Bitcoin was in a long slump after the collapse of Mt. Gox. Most people in Singapore had never owned a coin. Liu and Eng wanted to make buying one as easy as using a bank app, and they called the company Coinhako. The records say little about their childhoods or their schools. What the record does show is what they chose. They went to the accelerator in California, which treated bitcoin as the next internet, and then came back to a small, conservative financial centre that was still deciding how to regulate it. What almost nobody remembers is how long they had to wait. For years Coinhako operated in a grey zone where there was no clear licence. Singapore's Payment Services Act came into force in 2020. In November 2021, Coinhako said it had in-principle approval from the Monetary Authority of Singapore for a major payment institution licence, calling itself the first local, non-bank crypto exchange to get there. The timing was better than anyone could have known. A month later, in December 2021, the SBI-Sygnum-Azimut Digital Asset Opportunity fund made its first-ever investment in Coinhako, a round that was three times oversubscribed. The money was meant for expansion across Southeast Asia and for Coinhako Prive, a service for institutions and wealthy individuals. Then the market turned. In 2022, the crypto industry went through failures that took down large exchanges. A local exchange with a regulator's blessing was now a selling point, not a cost. Filings seen by data providers point to about US$10 million raised that year. In early 2026, reports said SBI Holdings intended to buy a majority of Coinhako Group, with the deal pending. The tale holds one instruction for founders. Liu and Eng chose a field with no rules, took the money of one early believer, and then did the slow, dull work of being ready when the rules arrived. The patience, not the pitch, was the product. ## Sleek (Singapore) URL: https://startup.talk/singapore/sleek/ Founders: Julien Labruyere, Adrien Barthel Founded: 2017; Based in: Singapore; Sector: Fintech / digital corporate services, accounting Two men walked into accounting firms in Singapore pretending to be customers. They were not. Adrien Barthel and Julien Labruyere were quietly doing market research, and what they found shocked them: firms earning margins of around 40 percent from clients who were unhappy. Both came from outside the accounting world. Labruyere had built a business-management career and then co-founded 4xLabs, a fintech working on foreign exchange. Barthel had spent six years in e-commerce, and was chief marketing officer at Luxola, the beauty retailer that LVMH bought in 2015. Between them they knew software, marketing and payments. They did not know accountants. That turned out to be an advantage. Their frustration with how slowly company registration worked was the start of Sleek, founded in Singapore in 2017. The idea was an all-in-one digital service: incorporate a company, keep the books, run payroll, file taxes, all on a platform rather than through paperwork and phone calls. What the interviews do not say is how hard it was. Accounting is regulated work with deadlines. Small business owners are cautious, and the firm they hire holds their financial records. Sleek had to prove that software plus a team of professionals could be more reliable than the person down the corridor. It convinced investors first. A Series A of about US$14 million, led by White Star Capital and Jungle Ventures, was later extended to US$25 million with EDBI joining. Sleek grew from Singapore into Hong Kong, the United Kingdom and Australia, selling the same promise in four rulebooks. In June 2025, it announced a US$23 million Series B led by Australia's Ellerston Capital, with Japanese software firm Money Forward as a backer. The company said it supported more than 16,000 SMEs a year, and that it would spend the money on automation and on deepening its presence in Australia and the UK. The most useful part of the story is the method. Before they wrote a line of code, the founders sat in the customer's chair at the competition, and left with a number. Forty percent margins and unhappy customers is not a technology problem. It is an opening. ## Atlassian (Australia) URL: https://startup.talk/australia/atlassian/ Founders: Mike Cannon-Brookes, Scott Farquhar Founded: 2002; Based in: Australia (Sydney; US hub in San Francisco); Sector: Enterprise software / team collaboration (Jira, Confluence, Trello) Most people tell the Atlassian story from the IPO bell. Start earlier, with an email that almost nobody answered. In 2001 Mike Cannon-Brookes, a university student at UNSW, sent a note to his classmates asking if anyone wanted to start a technology company. One person wrote back: Scott Farquhar. That single reply is the whole origin of a business now worth tens of billions. The reason they gave later was strikingly modest. The two have said they wanted to earn what a big employer would pay a graduate, about A$48,000, without working for anyone else. No grand mission. A salary, and the freedom to skip the interviews. Cannon-Brookes was not a typical Sydney kid. Born in New Haven, Connecticut, to a British banking executive and his wife, he was six months old when the family moved to Taiwan, then Hong Kong at three. He was sent to boarding school in England before landing at Cranbrook in Sydney, and then UNSW on a co-op scholarship. What almost nobody remembers is that this was not his first company. Before Atlassian he had co-founded The Bookmark Box, a bookmark-management tool, with classmate Niki Scevak, and sold it to Blink.com in 2000. He had already learned that a small web product could find a buyer. Atlassian began in 2002 on about A$10,000 of credit-card debt. They had no investors and no sales staff. The early business was supporting other companies' customer service teams, which demanded round-the-clock availability. Frustrated by the bug-tracking software on offer, they built their own, called Jira, and then made the unusual decision to stop doing services and sell it. Confluence followed in 2004. The rule they set was that the software should be bought, not sold. A customer could find it, try it and pay on the website. That choice kept costs tiny and let them run for years without outside money. When they finally took capital, in July 2010, it was US$60 million from Accel Partners, and by then revenue was heading toward US$100 million. The price of growth came later. A Confluence flaw in 2023, a data exposure in 2019, and layoffs of about 500 in 2023 and roughly 10% of staff in March 2026 showed that a company this big is no longer a two-student experiment. Farquhar stepped back as co-CEO in August 2024, leaving Cannon-Brookes alone at the top. The company reported US$5.22 billion in revenue for fiscal 2025, though still a net loss of US$257 million. The lesson is in the first line of the story. The company was started by someone who had already sold one small thing, and who asked a room of people for a partner, got one answer, and took it. ## Canva (Australia) URL: https://startup.talk/australia/canva/ Founders: Melanie Perkins, Cliff Obrecht, Cameron Adams Founded: 2013 (1 January, Perth); launched August 2013; Based in: Australia (Sydney); Sector: Design software / SaaS Somewhere in San Francisco in 2013, Melanie Perkins heard the same word again and again. No. More than 100 investors turned her and Cliff Obrecht down. The reasons were as plain as they were unfair. They were from Australia, far from the money. They were a couple, which made some investors nervous. And neither was a trained engineer. Perkins came from Perth. Her mother was an Australian-born teacher; her father a Malaysian engineer of Filipino and Sri Lankan heritage. At Sacred Heart College she trained seriously in figure skating and, almost as an aside, ran a small business selling handmade scarves. She started at the University of Western Australia, studying communications, psychology and commerce, and tutored other students in graphic design. That job taught her something she would build a company around: even people who needed design software could barely use it. She left university at 19. Hidden inside that decision is a fact that rarely gets told, because it came before Canva. In 2007 she and Obrecht started Fusion Books, a web-based yearbook publisher with a drag-and-drop editor and template library. It grew across Australia and later into New Zealand and France. Canva, launched in August 2013, was essentially the yearbook idea turned loose on everything else. The road to launch ran through those rejections. The founders needed a technical partner and an investor who would take the risk. Help came from Lars Rasmussen, the Google Maps co-creator, who advised them and introduced Cameron Adams. An investor suggested the team pick one leader, and Obrecht put forward Perkins. Then the numbers moved. More than 750,000 people used Canva in its first year. By 2017 it was profitable with 294,000 paying customers. In January 2018 it raised A$40 million from Sequoia, Blackbird and Felicis at a A$1 billion valuation, and by 2021 it was valued at US$40 billion. Success has had a price. A 2019 breach exposed data on about 139 million users. In 2026, with AI serving costs rising, Canva cut its growth outlook from 30% to 20%, and in November it laid off 1,271 people, about a fifth of its staff. It still counts more than 265 million monthly users and US$4 billion in recurring revenue. What a founder can take from it is simple. Perkins heard the same no about three things, geography, partnership and background, and none of them was about the idea. She kept the idea, changed the audience, and went back. ## Afterpay (Australia) URL: https://startup.talk/australia/afterpay/ Founders: Nick Molnar, Anthony Eisen Founded: 2014 (founded October in Molnar's Rose Bay home; design work began April 2014); Based in: Australia (Sydney); Sector: Fintech / buy now, pay later The idea for Afterpay did not come from a bank or a venture fund. It came from a jewellery salesman in Sydney who noticed that young people wanted to buy things and did not want a credit card. Nick Molnar was born in 1990 and went to Moriah College in Sydney. While still at school he was selling jewellery on eBay, and became the top Australian jewellery seller on the platform. That is the hidden piece that explains the rest. Before he was a fintech founder he was a teenager watching exactly how customers paid. He went on to a Bachelor of Commerce at the University of Sydney. Along the way he persuaded Ice.com, a US online jeweller, to let him launch an Australian version, iceonline.com.au, which he grew to A$2 million in annual revenue. The business taught him two lessons. Shoppers hesitate at the checkout, and many will repay early if the terms feel fair. In April 2014, he worked with a product manager and an engineering manager in Melbourne to design a no-interest payment product. There were two options on the table: pay thirty days after delivery, or split the bill into four instalments over up to sixty days. In October 2014 he and his neighbour Anthony Eisen, a former chief investment officer at Guinness Peat Group, founded the company in the living space of Molnar's home in Rose Bay. They chose the four-instalment model. Then came a smart choice. Instead of paying for retailers to sign up, Molnar encouraged shoppers to ask their favourite brands to offer Afterpay. Demand came from the customers. The company listed on the ASX on 4 May 2016, raising A$25 million, and later took money from Matrix Partners and Tencent, which paid A$300 million for about 5% in 2020. The hard years came with the scale. In 2019 the regulator AUSTRAC ordered an external audit of its anti-money-laundering compliance. Critics argued the service could push students and young people into stress and debt. Governments began planning to regulate buy-now-pay-later under credit laws. In August 2021, Block, then called Square, agreed to buy it for about A$39 billion in stock. The deal closed on 31 January 2022. By 2020 the company had never reported a profit, and Molnar's own fortune later slid from A$2.67 billion in 2021 to A$1.11 billion in the 2025 rich list as the sector re-priced. The takeaway is that Molnar sold a product long before he built a company. He understood the buyer's hesitation because he had spent his teens watching it. ## SafetyCulture (Australia) URL: https://startup.talk/australia/safetyculture/ Founders: Luke Anear Founded: 2004; Based in: Australia (Sydney; originally Townsville); Sector: Workplace safety and operations software (iAuditor / SafetyCulture platform) Long before he built software for safety, Luke Anear spent about seven years looking for people who were hurt on the job, or said they were. He worked as a private investigator for insurance companies, checking workers' compensation claims for fraud. It is an odd start for a man who now runs one of Australia's most valuable private software companies, and it is the fact that explains the company's name. Anear spent years seeing what happened when workplaces went wrong, after the injury and the claim, when someone had to work out what had been missed. SafetyCulture began in about 2004. The idea was close to a simple promise: replace paper checklists and clipboard inspections with a phone. The product that carried it, iAuditor, let a frontline worker walk a site, tick boxes, take photos and file a report on the spot. The lean years are quiet in public. What is clear is the scale at which the business later stood. In 2021 it raised US$73 million at a US$1.6 billion valuation, with Insight Partners and Tiger Global leading. Within months it was valued at A$2.2 billion after a further A$99 million round. Anear said at the time that the company had been cash-flow positive for the previous 18 months, a small remark that tells you a lot about how it was run. He framed the money as fuel for turning a checklist app into a wider operations platform. That is the second turn in the story. A tool that begins with one narrow job, inspecting a site, can become the place where all the routine work of a frontline team lives. The lesson is a modest one. Anear built a company around a problem he had seen from the wrong side for seven years. He did not need to invent the pain. He had been paid to watch it. ## Culture Amp (Australia) URL: https://startup.talk/australia/culture-amp/ Founders: Didier Elzinga, Doug English, Rod Hamilton, Jon Williams Founded: 2009 / 2010 (Forbes Australia); Based in: Australia (Melbourne; offices in US, UK, Germany); Sector: HR tech / employee experience Culture Amp, the company that now measures how workers feel, began as a product that nobody wanted enough. The founders, in Melbourne around 2009, started in the performance-review space. They tried to get traction and could not. Didier Elzinga has said, in the company's own telling, that the team brought in a fourth founder, Rod Hamilton, and from there Culture Amp took the shape it has today: a platform for employee surveys and organisational feedback. The pivot was the whole company. The hidden part is Elzinga's earlier life, which was not corporate software at all. He studied at the University of Adelaide and trained as a software engineer. In the mid-1990s he started as an entry-level engineer at Rising Sun Pictures, an Australian visual-effects studio, and rose to become its chief executive. The Hollywood connection is real: Rising Sun worked on film effects, and a related business he founded, Rising Sun Research, won a Technical Academy Award. He moved from making wizards fly on screen to asking what makes people want to show up on Monday. That earlier career matters. A creative studio is a place where a dozen strong personalities hold a project together by feel. The pattern is clear: he went where he could build a culture on purpose. The turn came when the company opened its first international office, in San Francisco, in 2015, alongside a US$6.3 million Series A led by Felicis. New York and London followed. It acquired Zugata in 2019 and Disco in 2021, and in July 2021 closed a Series F led by Sequoia Capital China and TDM Growth Partners, reported at about US$1.5 billion in valuation, although the press figures for the size of the round differ. The recent years have been harder. In February 2026 a planned liquidity event was reported to be on hold during a tech sell-off, and in June 2026 a former AI specialist claimed he was dismissed after making complaints. The company launched AI Coach in 2025 and acquired Orgnostic in 2024. The lesson is in the pivot. The first idea did not work, and the team was honest enough to change it. ## Zip Co (Australia) URL: https://startup.talk/australia/zip-co/ Founders: Larry Diamond, Peter Gray Founded: 2013 (incorporated June 2013 as zipMoney); Based in: Australia (Sydney); Sector: Fintech, buy now pay later The business plan for what became one of Australia's biggest fintechs was drawn on butcher paper in a hotel bar. Larry Diamond had an idea: a credit line that lived inside your phone and let you pay for a bike, a couch or a pair of shoes at checkout, without a plastic card. What he lacked was someone who understood consumer credit. So he found Peter Gray, a credit specialist, and the two began meeting at a North Sydney hotel. For roughly six months, they talked, drew and argued. Diamond later said that although Zip was just an idea, a bond was formed. Diamond had trained in information technology at the University of Technology Sydney, graduating in 2004. What he had was a belief that the bank card was a clumsy way to borrow small amounts. The company was incorporated in June 2013 as zipMoney. Friends and family backed the first loan book, a leap of faith with no history to point to. Diamond has since credited those early investors for believing in the team rather than the numbers. In December 2013 the first customer signed up. It was not a giant retailer. It was a bicycle shop called Chappelli Cycles. Then came the part nobody plans for. Listing on the ASX in September 2015 changed the company's trajectory, as Diamond has put it, and growth turned feverish. By 2017 Zip was onboarding about 1,200 new customers a day. A A$260 million debt facility arrived, A$200 million of it from NAB, and Westpac put in A$40 million of equity. Diamond has been candid about the cost. His hardest problem, he said, was moving from a small, close-knit team to a structured organisation, and he called hyper growth incredibly chaotic. Being the driving force of a company for the first time was new to him too. Expansion followed in a rush. Pocketbook, PartPay, Quadpay in the United States for A$430 million, Spotii, Twisto. Then the bill. Zip pulled out of Singapore and the United Kingdom in 2022 and shut Pocketbook the same year. In May 2026 the High Court sided with Firstmac in a fight over the very name Zip. What survived was the American bet. For the year to June 2025 Zip reported revenue of about A$1.08 billion and cash earnings of A$170 million, with the US arm contributing most of the profit. The lesson is a modest one. The company that conquered a continent started with two people, some butcher paper, and a willingness to sit in the same bar for six months until the plan was good enough to ask strangers for money. ## Rokt (Australia) URL: https://startup.talk/australia/rokt/ Founders: Bruce Buchanan, Justin Viles Founded: 2012 (Rokt rebranded from Rocklive 2013); Based in: Australia (Sydney); major operations in New York, USA; Sector: Ecommerce advertising technology At sixteen, Bruce Buchanan was running a handyman business. He was not doing it for fun. His mother had been ill with leukemia for about five years, and when she died he was in Year 10. His childhood had been scattered even before that. His parents came from the United States and the United Kingdom, and he went to school in both countries before his mother moved him to the New South Wales Central Coast after the marriage ended. After she died, he had few options. He kept going to school, did odd jobs, looked after his younger sister, and lived on a friend's floor for a year, with some help from his father. He has described the period as one that made him grow up fast. What almost nobody sees behind the polished chief executive is how he started. He had loved computers since he saved up for an early Apple. A part-time job at a CAD and CAM business led to him buying it, paying the seller back over two years. Its manufacturing clients were closing around him. When he finally exited, the money bought a house and an MBA at UNSW. That road led to Jetstar, where he rose to chief executive. From there he saw a problem in online retail. Shoppers at checkout were being shown nothing useful, and merchants were leaving money on the table. He met Justin Viles, whose company, Rocklive, brought live events into pubs and clubs and had drifted into digital marketing. Buchanan bought it through a Singapore company, split the equity evenly with Viles, and cut about 80 percent of what existed to chase one thing: ecommerce. He paid for the first year himself. There was no cash flow. Traction took about six months. He has called the early years rough. A Series A arrived at the end of 2013. The Series B, around 2016, was harder than any of it, because growth was good but not yet dramatic. Then the market moved. Most of the retailers a global network needed were in the United States. For about two years he flew two weeks on, two weeks off, before moving his family there. Today Rokt works with names like Uber, Macy's and Albertsons, bought mParticle in 2025, and was valued at US$3.5 billion in a secondary sale that year, with an IPO being prepared for 2026. The pattern in his story is the same at every step. He kept buying and fixing something small, using his own money, before anyone else would. Early on, hardship taught him to rely on himself, and that habit became the company. ## Linktree (Australia) URL: https://startup.talk/australia/linktree/ Founders: Alex Zaccaria, Anthony Zaccaria, Nick Humphreys Founded: 2016; Based in: Australia (Melbourne); Sector: Creator economy, link-in-bio software The night before his sister's wedding, Alex Zaccaria's phone would not stop. A user in the United States had posted his little side project on Product Hunt, and sign-ups were flooding in. Over that day, roughly three to four thousand people joined. The servers struggled. Nobody had planned for this. The product was small. Alex, his brother Anthony and their friend Nick Humphreys ran a Melbourne digital agency that looked after musicians and entertainers. They posted announcements on Instagram, which allowed only one link in a bio. They were tired of changing it. So they asked a developer who had recently joined the agency, and who had little work at the time, to build a page that held several links. The prototype took about six hours. They gave it to the agency's own clients first. About ten people a day signed up. It was a feature, not a company, and they rebuilt it properly before any real attention arrived. Nick, who has been described as the creative guy, shaped how it looked. Anthony later took the commercial side. Then the wedding week. After that day, sign-ups stayed above 800 a day. The company spent nothing on marketing. People just left the link in their bios, and creators made about 45,000 YouTube videos explaining it. Alicia Keys became the first major celebrity user, and her agency brought in Pearl Jam, the Red Hot Chili Peppers and The Killers. Qantas and Red Bull arrived without a sales call. The near-death moment came in 2018, when Instagram flagged the links as spam and banned them. Users rallied, and Instagram reversed course and apologised. Alex has since admitted that early on they worried about Instagram copying or blocking them. Later, infrastructure and scaling problems kept coming, in his words, in all kinds. By August 2019 they reported about US$3 million in annual recurring revenue, 2.8 million users, and no outside money. Then came investors. A US$10.7 million Series A in 2020, US$45 million in 2021, and a 2022 round at a reported US$1.7 billion valuation. The user count had passed 8 million by October 2020 and neared 16 million by March 2021. What stands out is how ordinary the origin was. Nobody set out to build a platform. They fixed an annoyance for their own clients, and then stayed ready when strangers noticed. ## Employment Hero (Australia) URL: https://startup.talk/australia/employment-hero/ Founders: Ben Thompson, Dave Tong Founded: 2014 (platform launch; earlier KeyPay); Based in: Australia (Sydney); Sector: HR, payroll and workforce software There were weeks when Ben Thompson could not be sure the payroll software company he had started would make payroll. He has since said that nothing teaches cash flow discipline quite like that pressure. He was a lawyer first. Thompson went to school and university in Armidale, New South Wales, graduating in economics and law. He became an employment lawyer, and after working abroad he came home and spent time at his father's workplace, a firm called Employment Innovations. That is where the idea arrived. Clients told him the same thing repeatedly. The hardest part of running their business was not the product or the customers. It was employing people. He could advise one employer at a time. He could see that was a ceiling. What almost nobody sees about the early days is how crowded they were. Thompson was running four companies at once, from a single floor above Wynyard Station in Sydney. Employment Innovations. A business called Power2Motivate. The software, and a fourth. He has called the decision to stop doing all of it the turning point. He had also adopted a strict ninety-day planning habit as early as 2005, borrowed from the Rockefeller Habits, and he still uses it. The platform launched in 2014, first as payroll and then as a wider home for hiring, pay, benefits and compliance. It went remote-first early, which paid off when COVID arrived in 2020 and he turned the company's energy to webinars and resources for employers. Then the money got big. The company bought his own side business for A$43 million, then spent A$112 million on a Canadian rival, Humi. At one point Thompson publicly fired a broadside at investor Hostplus, claiming the super fund wanted to put the company out of business. Forbes Australia has put the company's value above A$2 billion, and it counts more than 1,700 staff. The quiet lesson sits in the earliest detail. He did not invent a new market. He listened to what clients said was hard, noticed his own advice did not scale, and then stopped doing everything else so that one idea could grow. ## Immutable (Australia) URL: https://startup.talk/australia/immutable/ Founders: James Ferguson, Robbie Ferguson, Alex Connolly Founded: 2018 (as Fuel Games; renamed Immutable 2019); Based in: Australia (Sydney); operations in San Francisco, USA; Sector: Web3 gaming, blockchain infrastructure Their parents did not love the video games. Robbie and James Ferguson spent hundreds of hours on RuneScape and League of Legends as boys, which frustrated the adults in the house. James has said it showed him how much of life would one day be digital. Both brothers were headed for professional careers. Robbie, the younger, finished top of his school and started a double degree in computer science and law. He never finished it. What it did give him was a friend, Alex Connolly, who became Immutable's third co-founder and chief technology officer. James, five years older, was working through commerce and law when he took a year off and went to San Francisco to learn to code. He now calls himself a recovering law student. In 2017 CryptoPunks launched, and the brothers saw something. If a digital item could be owned and traded, a game could work the same way. They built Etherbots, which they describe as the first multiplayer competitive blockchain game. It launched in February 2018. Robbie was 21. The game reportedly earned about US$1 million in its first week. That result did more than pay bills. It gave a pair of young founders credibility to raise money. In July 2018, as Fuel Games, they began pre-selling Gods Unchained, a trading card game, after a US$2.4 million seed round led by Continue Capital and Nirvana Capital. The company became Immutable in 2019 and built Immutable X with StarkWare. Then the money got enormous. In March 2022 a US$200 million round led by Temasek, with Tencent and Animoca Brands, valued the company at US$3.5 billion. The same year brought the fall. The IMX token had traded near US$11 in November 2021 and slid to just above US$1 by July 2022. At least 20 people were laid off, including core members of the Gods Unchained team. In December 2023 an Adults Only rating from the ESRB briefly removed the game from the Epic Games Store. Immutable has kept going. Guild of Guardians launched in May 2024, and in August 2026 the company released Immutable Audience, a Steam promotion tool that does not use blockchain. The lesson is not about crypto. Two brothers who nearly became lawyers took a childhood obsession seriously, built something small that worked, and let that first week of proof pay for everything after. ## Go1 (Australia) URL: https://startup.talk/australia/go1/ Founders: Vu Tran, Andrew Barnes, Chris Eigeland, Chris Hood Founded: 2015 (some outlets say 2014); Based in: Australia (Brisbane, Queensland); Sector: Edtech / corporate online learning content The email arrived in Andrew Barnes's own inbox at around 2am. He had sent it to himself, a reminder to apply to Y Combinator, and he had misjudged the time zones. The deadline was only hours away. The application was finished at the last minute, with about sixty seconds spare, and Vu Tran, filling in the form beside him, let his bacon go cold. That panicked night is the part of the Go1 story founders rarely hear. The company is usually told as a unicorn tale. It started with two schoolboys at Cannon Hill Anglican College in Brisbane who built software together and earned, by Barnes's own telling, about a dollar an hour. They could have made fifteen stacking shelves. Tran's start was harder than his friend's. His parents are Vietnamese refugees. His father left school after grade three. Tran became a medical doctor and, even now, still sees patients as a GP on Thursdays, a quiet reminder of where the safety net sat. He also carried a credit card at 22.5 percent interest for six years, finally cleared in early 2020, long before the money came. Barnes went a different way. He studied economics and won a Rhodes scholarship to Oxford for a master's in educational technology. Chris Eigeland, a year below them at school, sought the pair out, wanting to ride on their coattails, as he put it. He studied law and, after the 2011 earthquake, lived in Haiti and started a charity that ships learning hardware. Chris Hood, who met Barnes at a tech conference, ran the software side. He often had to build features the founders had already sold. That, he said, is how they learned what customers wanted. The company took shape in Logan, a lower-income pocket of southern Brisbane, with a state grant of A$178,000 in 2016. Then came the 2am email, and the Y Combinator place. The idea was plain enough. Companies were buying training from dozens of publishers, so Go1 collected the content into one subscription. About a year in, Barnes set the goal of a billion learners. Money followed in steps: A$4 million in January 2017, a A$10 million Series A, then a A$30 million round seven months later. In mid-2021 SoftBank led a round that made Go1 Brisbane's first unicorn, and a 2022 raise put the valuation above US$2 billion. In 2023 it bought the German reading app Blinkist. Barnes later stepped back from the co-CEO job, leaving Eigeland in charge. What stays with you is how little of this was planned. The founders were a doctor, a Rhodes scholar, a human rights lawyer and an engineer who built things before they were sold. The pivotal moment was a rushed form filled in against the clock. Not every window is announced in advance. Sometimes a founder only learns it exists when someone is already racing to close it. ## Judo Bank (Australia) URL: https://startup.talk/australia/judo-bank/ Founders: Joseph Healy, David Hornery Founded: 2016 (founded 16 Oct 2016 as Judo Capital; earlier planning from about 2015); Based in: Australia (Melbourne, Victoria); Sector: Fintech / SME banking Joseph Healy had just walked away from the biggest business banking division in Australia, and he did something unusual with the time. He took a year off at the University of Nottingham Ningbo China. It was there, away from the boardrooms of Melbourne, that he began working on what became Judo Bank. He had spent his career inside the machine. He started at Lloyds in London, moved through Citibank and CIBC World Markets, joined ANZ in 1998, then NAB in late 2006. From about 2008 to 2014 he ran NAB's business banking, the largest such unit in the country. He knew the system from inside, and he left it. The question he kept coming back to, as a podcast write-up about him puts it, was why start a new bank rather than reform an old one from within. His answer was in the numbers of the market. Australia's small and medium businesses relied on four big banks, and the banks were moving them onto formula-driven lending. In Britain, Aldermore, Shawbrook and OakNorth had built challenger banks around the same gap. Healy wanted one for Australia. David Hornery, a fellow NAB and ANZ veteran, joined him in 2015. The name they chose came from a 1999 Harvard Business Review article about 'judo strategy', the idea that a smaller player can win by using a larger one's weight against it. The next step was money, and a lot of it. Judo was formally set up in October 2016, seeded by a consortium of family offices led by Geoff Lord. In 2018 it raised more than A$140 million and arranged a A$350 million debt facility. A banking licence came from APRA in April 2019, and the company dropped 'Capital' from its name. Two months later came a A$400 million Series B. Then the pandemic arrived. In April 2020 the federal government committed A$500 million to Judo through two funding schemes, and in May Judo raised A$230 million at a valuation above A$1 billion. Prime Minister Scott Morrison later held it up as a fintech success story. It listed on the ASX in 2021 under the ticker JDO. For the year to June 2025 Judo reported statutory profit of A$86.4 million, up about 24 percent, on a lending book of A$12.5 billion. It guided to underlying profit before tax of A$180 million to A$190 million for FY2026. Healy has announced he will step down, with Chris Bayliss taking over. The lesson is in the order of events. The company did not begin with a pitch deck. It began with a year of quiet thinking by a man who had run the incumbent's business and decided that insiders see the gap better than anyone. Often the best start-up idea is a frustration a founder has carried for a decade. ## Envato (Australia) URL: https://startup.talk/australia/envato/ Founders: Collis Ta'eed, Cyan Ta'eed, Jun Rung Founded: 2006; Based in: Australia (Melbourne, Victoria); Sector: Creative marketplace / digital assets The site was supposed to take three weeks to build. It took six months. That is how Envato began. In 2006 Collis and Cyan Ta'eed, with their friend Jun Rung, set out to create a marketplace where designers and developers could sell what they had made, and they underestimated the job by a factor of eight. Along the way they worked their own freelance jobs, out of a garage, to pay for the work. There was no investor. There never would be. Cyan had grown up in Sydney. After university she started a freelance graphic design business, and she knew first hand how hard it was for a lone designer to sell a template twice. The marketplace sold WordPress themes, stock photos, fonts and graphic templates, and creators earned a cut each time someone bought them. It was an early version of an idea that later became everyday. The money story is where Envato is unusual. The Ta'eeds never took outside funding, so every decision was theirs, and every mistake. After launch, they spent about eighteen months travelling, determined to prove that a business could run from anywhere with Wi-Fi. When they returned they set up a Melbourne office. The team grew quickly and they started a family there. Freedom did not make it easy. A stretch of consolidation followed, as the company cut back to fewer products when demand shifted. Cyan left in 2018 to start Hey Tiger, an ethical chocolate brand. In 2020 Collis stepped down as CEO, handing over to Hichame Assi. The couple moved to Darwin to focus on philanthropy, including work with First Nations communities, and Cyan chairs the Bright Moon Trust, which invests in First Nations entrepreneurs. Then came the exit. In May 2024 Shutterstock agreed to buy Envato for US$245 million in cash, around A$375 million. The Ta'eeds owned about 68 percent. Collis's brother held roughly 16 percent, with Jun Rung holding part of the remainder. Shutterstock said the deal would add around 650,000 subscribers to its own base of about 1.15 million. The quiet lesson is in that six months. If they had known the build would take half a year, they might never have started. Underestimating the work is a mistake founders rarely forgive themselves for, but the six months did not stop them. They paid for their ignorance with freelance work, and the result was a company they owned almost outright. ## Deputy (Australia) URL: https://startup.talk/australia/deputy/ Founders: Ashik Ahmed, Steve Shelley Founded: 2008; Based in: Australia (Sydney, New South Wales); Sector: Workforce management / HR software for shift workers When the lockdowns began in 2020, Deputy's customers disappeared. Cafes, restaurants and shops were the heart of its business, and the pandemic halved a user base of two million. The company laid off 30 percent of its roughly 300 staff and had to fight to stay alive. That low point is easy to forget now. For most of its history Deputy had looked like a quiet success. It had been founded in 2008 by Ashik Ahmed and Steve Shelley. Ahmed had been the chief technology officer of an aviation services business run by Shelley, and the two moved from planes to rosters, building software that would let a manager schedule a shift from a phone. For about eight years, Deputy was bootstrapped, a stubborn choice in a start-up scene that raised money early. The first outside cheque came in January 2017, a Series A led by Boston's OpenView of around US$25 million. Late 2018 brought the Series B: A$111 million, which valued the company at A$423 million and was one of the largest for an Australian software company. Then came the pandemic. The company that had grown by selling to hourly workers lost them overnight. Survival meant cutting a third of the staff and rebuilding for a different kind of recovery. It worked, slowly. Revenue roughly trebled in three years, and annual recurring revenue passed A$100 million by 2023. The statutory loss of A$24.9 million turned into a A$8.5 million profit in FY23. In February 2023 Ahmed stepped down after six years as CEO, handing over to Silvija Martincevic, based in the US. In 2024 a A$37 million investment from US staffing firm Express Employment Professionals put its reported valuation above A$1 billion, although the company was reluctant to discuss it publicly. A spokesperson said backing from a customer or partner was strong validation. What is the lesson? Deputy was built to serve the workers who were hit first and hardest, and it survived the worst year of its life by shrinking first. Founders who build for a narrow market should know that the market can vanish, and that a reputation for frugality, earned in eight bootstrapped years, is what lets a company make the hard cuts without breaking. ## Harrison.ai (Australia) URL: https://startup.talk/australia/harrison-ai/ Founders: Dr Aengus Tran, Dimitry Tran Founded: 2018; Based in: Australia (Sydney, New South Wales); Sector: Healthtech / AI medical imaging Dimitry Tran wanted to be a doctor. Medicine in Australia was costly for international students, he has said, so he studied accounting and finance instead. He graduated top of his class. He was twenty-two when he and his younger brother Aengus moved from Ho Chi Minh City to Sydney. Aengus was fourteen. Their father was a teacher who had owned one of the first computers in Vietnam, and that machine, they have said, sparked their interest in technology. The brothers both wanted to study medicine. Only one could afford it. Dimitry's university chancellor introduced him to Paul Ramsay, the founder of Ramsay Health Care, and he joined as a strategy consultant, then rose to head of innovation. And with his income, he paid for Aengus's medical degree. Aengus studied in Sydney, with a view at one point to specialising in cardiology. While still a student he built an AI tool to help choose which IVF embryos to implant. The work showed the brothers something bigger. If AI could do the sorting work of a specialist, it might help a health system short on specialists. In 2018 they founded Harrison.ai in Sydney. They named it after Dr Tinsley Harrison, the author of Harrison's Principles of Internal Medicine, whose view that caring for another person is the greatest honour and responsibility Dimitry says drives the company. Aengus became CEO. Dimitry took the business side. The plan worked through partnerships. In 2021 they formed annalise.ai with I-MED Radiology, and later franklin.ai with Sonic Healthcare for pathology. In November 2021 they raised A$129 million, the largest Series B for an Australian start-up at the time. In February 2025 they raised US$112 million, about A$179 million, with the National Reconstruction Fund taking a stake. Its chest X-ray product is cleared for up to 124 findings and its CT brain product for up to 130, with 12 FDA clearances and clearance in 40 countries. Dimitry moved to the US to open the American market. There is a loop worth noticing here. The brother who could not study medicine built a company that serves doctors, and the one who could study it built the tool. The lesson for founders is not that sacrifice pays. It is that a family's early trade-offs can shape which problem its members end up solving. ## Eucalyptus (Australia) URL: https://startup.talk/australia/eucalyptus/ Founders: Tim Doyle, Benny Kleist, Alexey Mitko, Charlie Gearside Founded: 2019; Based in: Australia (Sydney); Sector: Digital health / consumer telehealth When the Hims & Hers offer came, the price was more than three times what Eucalyptus had been worth just three years earlier. Tim Doyle, the outspoken chief executive who had started the company with three friends, was about to take home a reported US$160 million and a new job as the American buyer's head of international. The pitch he gave for the company at its start was oddly humble. It was, he has said, a bet that generalist talent from Australia could build a globally competitive consumer healthcare business. Not a medical breakthrough. Not a patented drug. A bet on people who could learn fast. Four founders began it in 2019: Doyle, Benny Kleist, Alexey Mitko and Charlie Gearside. The public record says little about their childhoods or schools, and the story is better told by what they built than by what they were. The idea was simple to say and hard to do. Treat conditions people are shy about, such as hair loss, erectile dysfunction, weight and menopause, through online clinics where a doctor consults, prescribes and a pharmacy ships. Instead of one brand, they built several. Pilot for men's health. Juniper for women's weight loss and menopause. Kin for fertility. Software for prescription skincare. Each brand spoke in the voice of its customer. Money followed fast. An A$8 million Series A in May 2020. A$30 million that July of 2021. A$60 million in January 2022 led by Bond, the fund of Mary Meeker. Local backers Blackbird and AirTree stayed in, and so did Woolworths' venture arm. Then came the hard part. In 2023 an A$50 million raise valued the company at around A$520 million, and the same year Australian press raised questions about the online prescription model itself. Telehealth that sells medicines sits in an uncomfortable spot between healthcare and retail, and regulators and doctors watch it closely. The company also was not profitable, with a A$15.2 million after-tax loss in the 2024 financial year. What changed the maths was growth in the United States and Britain. By 2025 the business was running at more than US$450 million a year in revenue, with triple-digit growth in some quarters. In February 2026 Hims & Hers agreed to pay US$1.15 billion, with about US$240 million at closing and the rest deferred over 18 months plus earnouts tied to targets through early 2029. For a founder reading this, the useful point is the shape of the bet. Eucalyptus did not win by inventing a new medicine. It won by building a repeatable machine for launching brands, and by hiring people who could learn a new category in months. Doyle's own description of the company, generalists with ambition, is the closest thing to a founding philosophy the record gives us. ## Redbubble (Australia) URL: https://startup.talk/australia/redbubble/ Founders: Martin Hosking, Peter Styles, Paul Vanzella Founded: 2006; Based in: Australia (Melbourne); Sector: Online marketplace / print-on-demand For weeks at a time, nothing moved. No new artists. No new buyers. No new revenue. Martin Hosking, running a start-up that was meant to be a global marketplace, has said he asked himself why he was doing it at all. "Failure is so close and that makes you very anxious," he told SmartCompany years later. He admitted he thought about quitting several times. It is an odd confession from a man whose early life looked like a ladder. Hosking took an honours degree in history and economics and was recruited straight out of university into the Department of Foreign Affairs and Trade. He spent seven years there, serving in embassies in Egypt and Syria. He has described the work as heavy with responsibility and the training as excellent. What almost nobody connects to Redbubble is why he left. It was not ambition alone. His wife, a veterinarian, would have struggled to continue her profession if the couple kept moving from posting to posting. So he stopped. He later did an MBA at the University of Melbourne, spent two years at McKinsey, then moved into the internet, saying he wanted to be part of something big. He became a senior vice president at LookSmart, a company started by friends, and later chaired the software firm Aconex. Redbubble began in 2006 with Hosking, Peter Styles and Paul Vanzella and about A$2 million of investor money. The first idea was modest: a place to show and sell the founders' own artwork. It quickly became bigger, a platform where artists upload designs and outside printers make and ship the products. That is where the long wait began. Artists would not come without buyers, buyers would not come without artists, and the print partners needed volume before they would take part. Hosking has said the first three to four years were very hard, and real growth did not come until about 2010. What kept the team going, he said, was a clear purpose of serving artists, the "psychological prop that we needed." In hindsight he would have studied buyers earlier, because the company leaned too far toward the artist side. In 2011 he left Aconex to run Redbubble full time. In May 2016 it listed on the ASX, raising about A$40 million at a value near A$288 million. The price of that success was a messy second act: he stepped down as chief executive in 2018, a successor was ousted in 2020, and lawsuits and a brand rebrand to Articore followed. Revenue in the 2023 financial year was A$290.7 million. The lesson is in the choice at the start. Hosking gave up a diplomatic career for his wife's work, and then spent four years in the quiet valley between idea and traction. Patience was the product. ## Mr Yum (Australia) URL: https://startup.talk/australia/mr-yum/ Founders: Kim Teo, Kerry Osborn, Adrian Osman, Andrei Miulescu Founded: 2018; Based in: Australia (Melbourne); Sector: Hospitality tech / QR ordering and payments Kim Teo ate out nearly every meal. That small habit became a company. She would scroll through restaurant Instagram pages to see what the dishes actually looked like, then be left with a couple of questions to ask staff. Restaurants, she noticed, avoided putting photos on menus because it looked tacky. The quiet part of her story is that Mr Yum was not her first try. She had started Simply Borrowed, a bridesmaid dress hire platform. She had built Neighbour Flavour, an app to buy home-cooked meals from people nearby, and she stopped working on it in 2016. She co-founded a start-up incubator, Pitchblak, led a network for women founders in Melbourne and mentored at the Melbourne Accelerator Program. Years of learning, in other words, with no hit. Mr Yum came in November with co-founders Kerry Osborn and Adrian Osman. They decided on a plan and launched a basic version within weeks: dish photos, plain explanations of unfamiliar ingredients, filters for diets. It ran in a web browser rather than an app, because, Teo said, people hate downloading apps. Her advice to founders was blunt: do not build the Rolls-Royce product. A pilot at three venues drew up to a quarter of customers within days. The site had about 20,000 monthly users and 100% month-on-month growth. They had asked investors for A$700,000. They closed A$1.5 million, double the target, from angels, the hospitality group Australian Venue Co and venue owners. Then the pandemic turned a nice-to-have into a necessity. Venues needed contactless ordering, and QR codes were everywhere. In 2021 Mr Yum raised A$11 million, led by TEN13 with AirTree. By 2022 it had gathered more than A$100 million in venture capital. The price came next. With growth money gone cold, Mr Yum cut about 17 percent of its staff, the harshest cuts among the Australian rivals at that point. In September 2023 it did something many said would fail: it merged, all stock, with rival me&u, with Teo as chief executive. The combined group covered more than 6,000 venues across Australia, New Zealand, Britain and the United States, and processed more than A$2 billion a year. By 2024 Teo was talking about getting through the messy first year and coming out profitable, and about an almost even split of old Mr Yum and me&u staff. What the record gives a first-time founder is the pattern of Teo's earlier attempts. Two small ventures that did not scale were not wasted. They were rehearsals for the third one, which made a real hospitality customer's problem the starting point. ## Tyro Payments (Australia) URL: https://startup.talk/australia/tyro-payments/ Founders: Peter Haig, Andrew Rothwell, Paul Wood Founded: 2003 (incorporated 3 Feb 2003 as MoneySwitch Ltd); Based in: Australia (Sydney); Sector: Fintech / merchant payments and banking Before it had a bank licence, a listing or a rival bidding for it, Tyro was a company with a different name and a narrow ambition. On 3 February 2003 three men, Peter Haig, Andrew Rothwell and Paul Wood, set up MoneySwitch Ltd. Its pitch was plain: Australia's card payment system had been controlled by the big banks for years, and nobody had built a new way in since 1996. There is little public record of who these three were before this. Their schools, hometowns and early jobs have not been told in the interviews that reached this review, and the story is better followed through what the company did than through any invented portrait of them. What they did was difficult and unglamorous. Becoming a new EFTPOS entrant meant plugging into a closed system run by banks that were also their potential competitors. The company became the second in Australia, after GE Capital, to get a Specialist Credit Card Institution licence. By 2012 it was moving more than A$3.5 billion a year in card transactions. Then came the move that set it apart. In 2015 it became an authorised deposit-taking institution, the first Australian technology company to get a banking licence. That allowed Tyro to hold merchant money and lend against it, rather than merely processing payments. It also brought the pressure of a public company. In December 2019 Tyro listed on the ASX. The range was A$2.50 to A$2.75, and the final price was A$2.75, giving a market value of A$1.37 billion. It opened at A$3.40. Within weeks a short-seller, Viceroy Research, published a report alleging many terminals had been taken offline, and Tyro asked for a trading halt the same day. A connectivity failure in January 2021 hit merchants too. Chief executives came and went: Gerd Schenkel resigned in 2017, Robbie Cooke in 2022. In September 2022 Potentia made an unsolicited bid, Westpac later joined, and a later offer of A$1.60 a share, about A$875 million, was also rejected. Potentia left in May 2023 and the shares fell nearly 20 percent. For the 2025 financial year Tyro reported revenue of A$487.3 million, and in December 2025 agreed to buy the Melbourne fintech Thriday to move into small-business financial software. The lesson for founders is in the first line of the story. Tyro was born as MoneySwitch, a name about moving money. It took twelve years of licences and regulation to become a bank. A hard, boring infrastructure problem can be the moat. ## Campaign Monitor (Australia) URL: https://startup.talk/australia/campaign-monitor/ Founders: David (Dave) Greiner, Ben Richardson Founded: 2004; Based in: Australia (Sydney); Sector: Email marketing software (SaaS) The idea for Campaign Monitor came from a client who kept asking for something the two friends could not find. Dave Greiner and Ben Richardson ran an IT firm, Switch IT, and their customers wanted bulk email newsletters. Nothing on the market suited. So they paused the consulting and built their own. The quieter truth is how long they had known each other. The pair have been friends since they were toddlers. In 1999, while still IT students in Wollongong, they started Switch IT together. Five years in a business before Campaign Monitor, learning how to deal with real clients and real invoices. What they built, they built slowly. For about six months they worked on it a few hours a day, with a version-one rule: only features customers must have, a deadline, and a promise to use the product themselves. One account places the first build in a garage in the Sutherland Shire. The date is 2004. The pricing was their sharpest decision. They skipped the monthly fee that email firms used and charged per recipient. Anyone could send a campaign to as many as five subscribers for free, so a web designer could test it on a real client before spending a cent. They aimed it at web designers, the people who actually built clients' sites and newsletters. From there it grew by word of mouth. Their own description, years later, was 15,000 clients, 1,000 new users a month and 90 million emails delivered. They answered every support question themselves for a long time. And they paid for everything from revenue: the company was profitable from the first day, and Greiner said they were proud of the bootstrapping heritage. It was a decade before outside money arrived. In 2014 Insight Venture Partners put in US$250 million, around A$266 million, the first funding the company had ever taken. It was one of the largest early private tech investments in Australian history, and it changed the nature of the business from two friends to a global software company. What is plain is the shape of the choice. They built something small, charged fairly, kept the books clean, and took no one's money for ten years. For a founder tempted to raise before there is a product, this is the counter-example: patience, free tiers for the customer and profit for the founders can fund a very large exit. ## Brighte (Australia) URL: https://startup.talk/australia/brighte/ Founders: Katherine McConnell Founded: 2015; Based in: Australia (Sydney); Sector: Fintech / clean-energy and home-upgrade finance In 2015 the McConnell family had a battery humming in the house, one of the first in New South Wales. Katherine McConnell's children, then small, learned that a battery runs down overnight if you use too much. It was a lesson in limits, taught in her own home. That year she was 39, and a senior manager at Macquarie Bank, running energy leasing. It was a safe, senior job. She walked out of it to start Brighte on her own. To understand why, go back to where she began. McConnell was born in Sydney and grew up in the outer western suburbs. Her father had come to Australia as a refugee in his teens, with a Russian and Ukrainian family background behind him. Money, in that house, was never a given. Years later she would say a line that sounds like a family creed: energy equipment should be accessible to everyone. At Macquarie she had worked on a way to make solar financing reach more people. Inside a big bank it stayed a product. Outside, she believed, it could be a company. What almost nobody noticed at the time was how personal the bet was. Her own family had put solar on the roof in 2010 and a battery in 2015, so she knew what a household feels when it signs up for this technology, and what it fears. The early money was the hard part. A lender needs two kinds of cash: equity to build the business and debt to lend out. Her seed round closed in 2016. In 2017 Atlassian co-founder Mike Cannon-Brookes' fund, Grok Ventures, led a A$4m round, and Grok has since backed every raise. In 2018 came an A$18.5m Series B and a A$20m debt line from NAB. She has talked about the waiting involved, the long months before an investor says yes. Then the market shifted under her. Solar loans were not enough, so around 2020 and 2021 Brighte widened into home improvement, backed by a A$100m round in which Grok put in A$78.8m. By December 2024 the company had raised about A$150m in equity and almost A$1bn in debt, and financed roughly 180,000 solar installations. She is still the only founder, still CEO, and now running neighbourhood electrification pilots in Wollongong. The lesson she lives is plain. She did not invent a technology. She noticed a household's problem in her own kitchen, then stayed long enough to be believed. ## Instaclustr (Australia) URL: https://startup.talk/australia/instaclustr/ Founders: Ben Bromhead, Adam Zegelin, Peter Lilley, Doug Stuart Founded: 2013; Based in: Australia (Canberra; US presence in California); Sector: Managed open-source data infrastructure (Cassandra, Kafka, PostgreSQL) Before Instaclustr there was relational.io, and almost nobody remembers it. Ben Bromhead and Adam Zegelin were students at the Australian National University in Canberra, friends who liked building things together. With two older colleagues, Peter Lilley and Doug Stuart, they started a product called relational.io. It was the sort of early venture many founders quietly bury. Bromhead, though, mentions it on his own company profile, a small clue that the company everyone knows came out of an earlier attempt, not a flash of genius. The real turn came in a plain engineering moment. Zegelin has described how the two were working on a proof-of-concept that needed a database which could scale, stay available, and handle flexible data shapes. Their research led them to Apache Cassandra. It was powerful and also famously hard to run. When they looked for someone who would run it for them, there was nobody. The gap was the business. So in 2013 they set up Instaclustr in a shared innovation space on the ANU campus, with the university's venture arm, ANU Connect Ventures, behind it from the first day. Canberra is not Silicon Valley, and they did not pretend otherwise. At one point 20 of 25 staff worked from the capital. The team was bootstrapped at the start, with Lilley and Stuart in the early cheques-and-spreadsheets role, while the two engineers built the platform. Money arrived in steps. A A$2m seed in 2014. A US$7m Series A in 2016. A US$15m round in 2018 as customers wanted Kafka, Spark, Redis and Elasticsearch managed too. Total funding stayed modest, about US$22m, which forced the company to earn its way. The pressure was always the giants. Amazon and others offered their own managed databases. Instaclustr's answer, as it said in 2022, was to stay purely open source rather than lock customers in. In 2020 Lilley moved up from COO to CEO. In April 2022 NetApp announced it would buy the company; the price was never published, though local reports put it above A$500m. That same year Bromhead and Zegelin received the ACT Chief Minister's Pearcey Entrepreneur Award. For first-time founders the quiet lesson sits in the order of events. The winning idea did not arrive first. It arrived because a failed or forgotten project forced two friends to solve a problem for themselves, and then notice others had it too. ## Zeller (Australia) URL: https://startup.talk/australia/zeller/ Founders: Ben Pfisterer, Dominic Yap Founded: 2020; Based in: Australia (Melbourne); Sector: Fintech / business banking and payments Zeller asked investors for A$6.3m before it had anything to show them. No product, no customers, no launch date. There was a plan, and two people who had done something similar once before. Those two were Ben Pfisterer and Dominic Yap, who had worked together at Square. In 2014 Pfisterer had launched Square in Australia and ran it as country manager until the end of 2019. Yap joined him very early, in strategy and growth, and stayed until 2020. Between them, they had watched thousands of small merchants struggle with the same thing: the banks. What the public rarely heard is how little they could offer investors beyond their reputation. Square Peg Capital led the seed, and the firm's co-founder Paul Bassat joined the board. Addition, the fund started by Lee Fixel, put money in before launch too. It was a bet on two operators, not a deck. They assembled a technical team, splitting the heavy engineering between two CTOs, David Conn and Alfred Wong. The target was ambitious. Zeller wanted to replace the merchant account, the terminal, the business transaction account and the card with one product, built for small businesses that felt ignored by the big four banks. The product went public on 4 May 2021 from Melbourne. Growth was quick enough that Spark Capital led a round valuing it at about A$400m. Then, in March 2022, Headline and Hostplus led a A$100m Series B. The price of speed is exposure. Taking on four entrenched banks means incumbents can cut fees overnight, and a business account carries trust that a startup must earn one merchant at a time. Pfisterer has talked about that fight as taking on the big four. Public detail on the founders' childhoods is scarce, and the story rests on a simpler human fact. They built Zeller by going back to work they already knew in their bones, and asking investors to fund their judgement before a single customer existed. For founders, the takeaway is sharp: experience that was once someone else's job can become your seed round. ## Pet Circle (Australia) URL: https://startup.talk/australia/pet-circle/ Founders: Mike Frizell, James Edwards Founded: 2011; Based in: Australia (Sydney); Sector: E-commerce / pet supplies and pet insurance At first, the stock seemed to be vanishing. Mike Frizell and James Edwards, working in a rough warehouse with worn desks and bags of dog food, worried they were losing inventory. The culprit, they later learned, was a ferret. It is a funny detail, but it shows what the early days were: two people, a shed, and a lot of animals. Frizell was a former investment banker and management consultant. In January 2011 he began working on an idea at Vibewire, a co-working space in Ultimo, Sydney. His motivation was domestic. Buying pet food felt to him like the milk deliveries his family had in the 1990s, something that should just arrive. Next to him sat Edwards, who held a PhD in applied mathematics and a computer science degree and had been building warehousing software. They merged their efforts. The first stock lived in a 16 square metre shed a block away. The first real warehouse, a Kennards Storage unit, needed a whole weekend of scrubbing before they could move in. Edwards made an unusual choice. Most startups build the shopfront first. He built the back end, an in-house system called MinPin, to forecast orders and run warehouse flow, because the real enemy was delivery. Pet food bags in 2012 were mostly paper, weighed about 20kg and burst easily. Solve that, he reasoned, and the rest follows. AirTree founder Craig Blair came in as adviser and Chair, and AirTree backed the company. Then came the name. The business launched as Paws for Life. By 2014 the range had grown beyond cats and dogs, and on radio the name was a mess: paws could sound like pause, for like four. They renamed it Pet Circle. The change took about a month and traffic dipped briefly. Frizell said afterwards that customers knew who they were. Growth afterwards was steep. Turnover passed A$70m by mid-2017, with more than 130 staff. In December 2021 a A$125m round valued the company above A$1bn, and in 2023 it added pet insurance. Frizell has said he learned to treat data as one input, not the final word, and to stay close to customers by being one himself. The warehouse has moved four times since the shed. The instinct to fix the heavy, unglamorous problem first has not. ## Koala (Australia) URL: https://startup.talk/australia/koala/ Founders: Dany Milham, Mitch Taylor Founded: 2015; Based in: Australia (Sydney; founded in Byron Bay); Sector: Direct-to-consumer furniture and mattresses In 2019, Dany Milham walked away from the company he had built. His father had been diagnosed with stage-four cancer, and Milham went home to Byron Bay to look after him for a year. The company was Koala, the mattress brand he had started four years earlier. Milham was born in Byron Bay, swapped surf life for computers, and, by one account, founded his first startup around 13. Mitch Taylor, his co-founder, had once coached him in under-14s rugby. The two worked out the Koala idea while surfing off a quiet beach south of Yamba. The model was simple and slightly cheeky: compress a foam mattress, put it in a box, sell it online, and make the marketing impossible to ignore. Early marketing showed off how little the bed moved when one sleeper shifted. Backed in part by Australian cricket captain Steve Smith, the business turned over more than A$1m in under three months in 2016. It added Japan in 2017 and later sofas and furniture. Then came the unglamorous part. In 2020 some manufacturing moved to China, which drew public criticism. An IPO planned for 2022 was abandoned. Milham, having left, started Milkrun, a ten-minute grocery delivery business backed by Tiger Global, AirTree and Atlassian's co-founders. Milkrun collapsed. What the headlines did not say was what came next. After Milkrun and the loss of his father, Milham came back to Koala in 2024 as chief executive, taking over a company that had lost a co-CEO the year before. He has said Koala has always been his life's work. The second act worked. Koala expanded into the US and UK, turned profitable, and on 31 March 2026 listed on the ASX at A$3.40 a share. The float raised A$68.1m, of which A$20m was new capital and the rest came from existing holders selling down. Taylor sold about A$8.3m. Milham sold none. The company forecast revenue near A$332m for FY26. Asked about the listing, Milham said they had not done it with that goal in mind. For founders, the telling choice is the one that looks like a loss at the time. He left for a reason money could not fix, and the company he came back to was still his. ## Who Gives A Crap (Australia) URL: https://startup.talk/australia/who-gives-a-crap/ Founders: Simon Griffiths, Jehan Ratnatunga, Danny Alexander Founded: 2012 (crowdfunded 2012); Based in: Australia (Melbourne); Sector: Consumer goods / social enterprise (toilet paper) For 50 hours in 2012, Simon Griffiths sat on a toilet in a draughty warehouse and asked strangers on the internet to pay for toilet paper that did not exist yet. He has said he would never do it again, but that it was the right thing to do. His legs paid for it. He ended up with nerve damage that lingered for about three years, after a scare over deep vein thrombosis was ruled out. The stunt was the end of a long run of near-misses. Griffiths grew up in Western Australia, looking after neighbours' cats for pocket money and making sandboards to sell to friends. After school he took a gap year, taught himself to code and paid his way through the University of Melbourne with freelance web work. He lived at Trinity College on a scholarship, treasurer of the music society, the sort of student who signed up for everything. Then came the trying. A short stretch in management consulting left him unfulfilled. In 2007 he co-founded Ripple, an online fundraising platform, with his university friend Jehan Ratnatunga and Matthew Tilleard. It had early success, but it could not sustain itself. Next he opened the Shebeen, a bar in Melbourne's CBD that handed over every dollar of profit to charity. It closed too. He later described it as a three-year education in business, a costly one. What almost nobody knew when the toilet went on camera was how many times he had already failed at doing good for a living. The idea itself arrived in a bathroom, staring at a six-pack of toilet paper, with the knowledge that about two billion people have no toilet at all. He and Ratnatunga teamed up with Danny Alexander, who had a product development degree and had worked on toilet-building programs in Africa. The campaign asked for A$50,000. It raised A$93,000. A first run that could easily have been their last sold out within days of launching. A retailer deal stalled, which pushed the founders to sell straight to households online, and that decision became the company's shape: recycled paper at the door, wrapped in loud paper, with half of profits going to sanitation. The pandemic delivered the biggest test. In 2020, when shelves emptied, demand jumped by 1,100 per cent. By then the company had given about A$8.3 million to partner organisations, and A$5.85 million of that in a single year. Griffiths has visited WaterAid projects in Timor-Leste to see where the money lands. The lesson is in the order of events. The famous stunt looks like a gimmick, but it came after a failed platform and a failed bar. Griffiths had spent years finding out what people would actually pay for. A toilet roll, it turned out, is easier to sell than an abstraction like impact, and the failures taught him to sell the roll. ## Cluey Learning (Australia) URL: https://startup.talk/australia/cluey-learning/ Founders: Mark Rohald, Robert Gavshon Founded: 2017 (incorporated 19 July 2017; launched July 2018); Based in: Australia (Sydney); Sector: Edtech / online tutoring On 9 December 2020 a young Sydney tutoring company rang the bell on the ASX. Its shares opened at A$1.27, six per cent above the A$1.20 it had asked for, and the raise brought in A$30 million. For Mark Rohald, the chief executive, it was a quick arrival. Cluey Learning had been incorporated on 19 July 2017, with Rohald appointed chief executive on 1 July and to the board the day the company came into being. The tutoring service itself launched in July 2018. Barely two years later it was a listed business, with a market value of about A$143.5 million at listing. The pitch was plain. Parents wanted one-to-one help for a child who was falling behind or aiming higher, and they wanted it without a drive across town. Cluey recruited tutors, trained them, and ran live sessions online. By the time of the listing it had around 620 tutors and had delivered more than 192,000 sessions, 52,700 of them in the most recent quarter. Then the pandemic sent every classroom online, and the company's model looked prescient rather than novel. The founders kept real skin in the game. A company history deck showed the founders held about 30 per cent, with chairman Robert Gavshon holding roughly 8 per cent. A listing like that is a promise to stay. Growth meant buying as well as building. Cluey added Code Camp, which teaches children to program, as a second business. It also set out to buy Art of Smart, a larger tutoring group, but that deal fell over when the conditions in the share purchase agreement were not met. The harder years came after the first day. Pandemic demand eased and the share price drifted. In December 2025 the company asked shareholders for fresh money through an entitlement offer, with the founders in line to receive new shares. In June 2026 it proposed leaving the ASX altogether, explaining that trading volumes were low and a few investors held most of the stock, so the market for its shares was illiquid. Shareholders approved the move on 13 August 2026, and Cluey left the ASX on 6 October 2026. The arc is a useful one for a founder weighing a listing. Going public gave Cluey capital and a profile inside three years. It did not give it liquidity, and a small company with concentrated owners can find the public market is a cost rather than a benefit. The business of putting a good tutor in front of a child stayed the same throughout. ## Mable (Australia) URL: https://startup.talk/australia/mable/ Founders: Peter Scutt, Tony Charara Founded: 2013 (operating since 2014); Based in: Australia (Sydney); Sector: Care marketplace / aged care and disability (NDIS) Peter Scutt's parents wanted to stay in their own home. That was all, and it turned out to be very hard to arrange. Scutt went looking for the right home care and found a system built around agencies and phone calls. He has described the experience through the company he later founded: he struggled to find the right solution, and the struggle stayed with him. He was not a care worker or a policy expert. He was a son with a problem, and a friend and colleague, Tony Charara, who thought the problem could be fixed with software. The two started Better Caring in 2013, an online marketplace where older Australians and people with disability could find independent support workers directly. The platform was operating by 2014. Later they renamed it Mable, a name chosen to suggest freedom of choice and control, the thing Scutt's parents had wanted. The timing mattered more than anyone could have planned. As Australia rolled out the NDIS and shifted aged care towards consumer-directed funding, people were handed budgets and told to choose. Many had no idea where to begin. Mable offered a list of workers with reviews, a way to book them, and lower prices because it skipped agency overheads. The company says it has delivered A$180 million in savings to customers since opening. The money followed. In 2019 Ellerston Capital's JAADE Fund led a A$15 million round. In September 2021 General Atlantic led a A$100 million raise, and the Paralympic champion Dylan Alcott was among the backers. The company bought two NDIS businesses, Leapin and HomeMade, and by its tenth anniversary in October 2024 had made more than 200,000 connections and counted nearly 250 employees. Then came scrutiny. The consumer regulator, the ACCC, found Mable had used unfair contract terms between November 2023 and August 2024, including a minimum A$5,000 penalty fee and automatic approval of service logs. Mable admitted the terms were unfair, changed its website and terms, and offered a court-enforceable undertaking. Reports in 2025 said the business, not yet profitable, was being shopped to buyers at a price near A$1 billion. The lesson for a first-time founder is how plain the origin was. Mable began with a family inconvenience, not a market map. A marketplace is a promise of fairness to both sides, and the regulator's finding showed how quickly that promise can be tested once a platform gets big enough to write the rules. ## Airtasker (Australia) URL: https://startup.talk/australia/airtasker/ Founders: Tim Fung, Jonathan Lui Founded: 2012 (idea 2011); Based in: Australia (Sydney); Sector: Online services marketplace The orders were silly. Somebody in Sydney wanted fried chicken at 2am, and the people who drove it across town were not couriers but the founders of a start-up, doing the job themselves because no one else was yet on the platform. Tim Fung grew up in St Ives Chase, a quiet northern Sydney suburb. He went to university in the city and then into Macquarie Bank, arriving in the years he later described as an insane run of success, roughly 2004 to 2008. For a young graduate it looked like a staircase going only up. Then, in 2009, the staircase ended. Fung was made redundant. What almost nobody connects to the Airtasker story is where he landed. He took a job at a modelling agency, Chic Management, and from the back of that office he helped launch a mobile phone company called amaysim. The banker who lost his seat ended up building a telco out of a modelling business. It was there that he met Jonathan Lui, and it was there that he met the investors who would later write the first cheques for Airtasker. The idea arrived in 2011, as the two moved apartments. Half of the misery of a move was small jobs somebody else could do. Fung has said he knew it was a winner when he pitched it to a housemate who disagreed with him about most things. The company launched in 2012 with A$1.5 million. A second raise of A$2M in 2013 valued it at A$10M. It was not smooth. Fung has said they tried a lot of things and failed at a lot of them, and that it took years to get going. The founders ran errands themselves. At one point pamphlets they had been paid to distribute were reportedly thrown away, leaving them to face unhappy customers. Rivals TaskBox and Occasional Butler were bought rather than beaten. Money came, and went faster. A$6.5M in 2015, A$22M from Seven West Media in 2016, A$33M in 2017 to try the United Kingdom. Fung has recalled a stretch when the company burned about A$2 million a month. In 2017 Freelancer accused it over an ad campaign, in 2018 came a data breach, and by December the tax office was auditing its R&D claims. The company listed on the ASX under the code ART. In its 2026 financial year, group gross marketplace value reached a record of about A$240 million, revenue was about A$57.8 million, and the loss at the operating line had narrowed to about A$21 million. Management is now aiming for positive free cash flow in the 2027 year. The redundancy that looked like an ending is the quiet hinge of the whole story. A man pushed out of a bank's boom had nothing to defend, took a job nobody would have predicted, and sat close enough to the right people when an idea walked in. First-time founders often wait for a perfect launch. Fung's version is less tidy: lose the plan, stay near good people, and be willing to deliver the chicken yourself. ## Octopus Deploy (Australia) URL: https://startup.talk/australia/octopus-deploy/ Founders: Paul Stovell, Sonia Stovell Founded: 2011 (started as a side project; profitable 2012); Based in: Australia (Brisbane); Sector: Developer tools / continuous deployment software Paul Stovell almost sold Octopus Deploy for A$100,000. The offer came when the product had no revenue to speak of. He has said he thought it was a great deal, and he would have taken it. It was his wife, Sonia, who stopped him. She reminded him that he had always wanted to lead a software company, and she predicted he would regret a sale. She was right. She is now the company's chief financial officer. The code had begun as a side project in 2011. Stovell was a developer who had noticed something odd: plenty of teams could set up a build server that tested their code automatically, but far fewer could push that code out to the live servers without a manual scramble. Octopus aimed at that one gap. It was not trying to be a whole platform. It worked. By 2012 the business was profitable and covered both founders' salaries, which gave the Stovells something rare for a young software company: freedom. They did not need anyone's money, so they did not take it. Stovell said later he had been turning away venture capital firms for years, and he never wrote a pitch deck. What the outside world noticed over the following decade was the customer list. Roughly 25,000 organisations came to use Octopus, among them Microsoft, NASA, Xero, Disney and Stack Overflow. More than half its customers were in North America, even though the company ran from Brisbane. The change came in April 2021. Insight Partners approached Stovell, and he agreed to sell a minority stake, the first outside capital the company had accepted, for US$172.5 million. He did not need the cash to survive. His reasoning was about speed: he saw that Insight could help sell to large enterprises, especially in the US, faster than he could alone. Headcount, which had been around 80 the year before, was heading for more than 180 and then past 300. Later, the company bought the Kubernetes delivery firm Codefresh. The lesson sits in the order. For ten years Stovell had room to say no to money, to buyers, and to growth for its own sake. He nearly gave that away for A$100,000. The cofounder who said wait protected the one asset that made every later negotiation easy: the choice to walk away. ## Nuix (Australia) URL: https://startup.talk/australia/nuix/ Founders: Tony Castagna, David Sitsky Founded: 2000 (company's own account); Australian entity incorporated 2005; Based in: Australia (Sydney); Sector: Data analytics / investigation software The company says it was born in the year 2000. The paperwork tells a quieter story. Nuix Australia Pty Ltd, a Victorian court filing shows, was incorporated on 15 November 2005, and the company's own marketing has long pointed to an earlier start. Two dates, one company, and a lot of years in which very little was said in public. What is documented is the pair at the beginning: Tony Castagna and David Sitsky, two men building software in Sydney to make large piles of unstructured data searchable. Their own histories before that are thinly recorded. What the record does show is how small the thing was. In 2006 Nuix had two developers. Ten years later it had more than 400 people. The product was unglamorous and that was its strength. Investigators drowning in email archives, scanned files and seized drives needed to find one thing in millions. Nuix patented several features of its engine in 2011. Then came the two moments that gave it a public face. In 2013 it gave free licences to the International Consortium of Investigative Journalists for the Offshore Leaks investigation. In 2016 its software helped turn 2.6 terabytes of scanned Panama Papers documents into something reporters could search. That was the turn. A tool for forensic labs was suddenly the machine behind the biggest leaks in a generation. The listing in December 2020 should have been the crowning day. The IPO valued the company at about A$1.8 billion, with Macquarie Group holding roughly 30 percent. Instead, within months the story soured. Questions came over revenue forecasts in the prospectus, and ASIC investigated before saying in February 2022 it would take no further action. In May 2021 the company cut ties with Castagna after the Australian Federal Police began looking at alleged breaches of the Corporations Act. In June 2021 federal agents raided the office in an insider trading investigation, and the chief financial officer left by mutual agreement. Leadership kept turning over. Rod Vawdrey became chief executive in 2017 and later resigned; Jeffrey Bleich became chairman in November 2020; Jonathan Rubinsztein took over as group chief executive in October 2021. What stays with you is the distance between the product's reputation and the company's. The software earned trust by helping reporters and investigators uncover other people's secrets. The company lost trust partly through its own opacity, a founding date that depends on who you ask and a prospectus that drew regulators. For a founder, the lesson is plain. Technology can carry you a long way, but once you sell shares to the public, the numbers you promise and the people you keep around you become the product too. ## Freelancer.com (Australia) URL: https://startup.talk/australia/freelancer-com/ Founders: Matt Barrie Founded: 2009; Based in: Australia (Sydney); Sector: Online freelance marketplace Before the marketplace with millions of users, there was a craft-supply shop run from home. Matt Barrie, a Stanford-trained engineer, reportedly ran a small online dropshipping business selling craft goods, the kind of side project that teaches you what a customer actually wants and how small the margin on a parcel can be. It is an odd detour for a man who would end up running one of the largest freelance platforms in the world. Barrie was born in Adelaide in 1973. He took a master's degree in electrical engineering at Stanford in 1998, then worked in information security, reportedly at a firm called Securify. Back in Sydney, he began teaching computer and network security at the University of Sydney in 2001, and technology venture creation from 2010. He is still an adjunct associate professor there. That teaching life matters to the story: he was a founder who also lectured on how founders fail. In 2009 he launched Freelancer.com. The idea was plain. An employer posts a job. Freelancers bid. The site takes a fee, 10 percent, reduced for paid members, with a five dollar minimum. Later it added contests, where designers and developers compete for prize money. What made it work was scale on both sides. Cheap bids drew employers; steady work drew freelancers; and each pulled the other in. The company listed on the Australian Securities Exchange in 2013 under the ticker FLN, and then used the market's money to buy its way into new corners. Escrow.com came in 2015. Prolancer and Nubelo followed in 2016, Channel 40 in 2018, and the freight marketplace Loadshift in 2021. The growth had a cost. In December 2015, Australia's Information Commissioner ruled that Freelancer had breached the Privacy Act by exposing the personal information of a former user who had posted critical comments about the company. It ordered A$20,000 in damages, a written apology and staff training. The company said it intended to appeal. A separate privacy complaint was reported in 2018. For a marketplace built on public reputation, a ruling over how it treated a critic was an uncomfortable mirror. Where it stands is harder to summarise, since the figures I could verify are older than a founder would like. What is clear is that a marketplace which started as one engineer's bet that talent and work could meet online is still run by him sixteen years on. The lesson is in the small business before the big one. Barrie learned the mechanics of online commerce on a shoestring. Founders who start with a humble shop often understand the unit economics better than those who start with a vision. ## Spaceship (Australia) URL: https://startup.talk/australia/spaceship/ Founders: Andrew Sellen, Paul Bennetts, Kaushik Sen, Dave Kuhn Founded: 2016; Based in: Australia (Sydney); Sector: Fintech / superannuation and investing It took about three days. That is how long Andrew Sellen says it took to raise A$1.6 million for a startup that had barely begun. For anyone who has spent months chasing cheques, the number sounds like a typo. The idea came over beers at the end of 2015. Sellen and his three co-founders, Paul Bennetts, Kaushik Sen and Dave Kuhn, all came out of financial services, and all had a similar itch. They were frustrated trying to launch fresh ideas inside large institutions. Sellen's particular complaint was about default super options: heavily weighted to Australian shares, particularly the big banks and miners. He called that risky. So they built Spaceship, a mobile-only super fund that put money into large technology companies and startups, with the biggest weighting in names like Apple. Members could see their holdings down to the stock level. An AI-driven app would track where their money went, and a single tap would merge scattered super accounts. The backers were a small who's who of Australian tech. The round was the first startup investment made through Mike Cannon-Brookes' Grok Ventures. Also in were Simon Clausen, an early Salesforce investor, Aconex founder Leigh Jasper, Upguard's Mike Baukes and Alan Sharp-Paul, and Vinomofo's Andre Eikmeier. Cannon-Brookes said he planned to move his own super into the fund when it launched. That single line did more for credibility than a brochure. But retirement money is not a social app. In 2018 the regulator fined the company over its super product statements, a reminder that a friendly interface does not exempt a fund from the rulebook that governs the old giants. The ending came in September 2024, when eToro acquired it. The four who began over beers had built something a global platform wanted, in under a decade. The lesson for future founders is in the seed round. It closed fast because the founders were credible insiders who named a specific villain, the default fund, and had angels who understood fintech. Speed in fundraising is rarely luck; it is usually the compound interest of a reputation and a sharp argument. ## Basiq (Australia) URL: https://startup.talk/australia/basiq/ Founders: Damir Cuca Founded: 2016; Based in: Australia (Sydney); Sector: Fintech / open banking API Two of Australia's biggest banks, rivals in almost every other way, wrote a cheque to the same small startup. That oddity is the clearest fact in Basiq's story, and it says a lot about what the company was selling. Basiq was founded in 2016 by Damir Cuca, who still leads it as chief executive. Publicly available accounts say little about his early life, schooling or the years before the company. What they do show is a founder who spotted a plumbing problem. Australia was heading toward open banking, a regime in which customers could allow other companies to see their bank data. Someone had to build the connections. Basiq set out to be that someone, described at the time as the country's first open banking API platform. The early money came from the very institutions that stood to be disrupted. In June 2017, NAB Ventures and Westpac's Reinventure put money into a seed round whose size was never disclosed. By October 2019 a Series A brought those two back, alongside Salesforce Ventures, one of the first investments from Salesforce's new Australia Trailblazer Fund. Visa also made an undisclosed investment. The customers were banks, lenders and fintechs, none of whom had to be persuaded that data access was a problem. What a startup in that position needs is credibility and patience, since regulatory timelines move slowly and revenue depends on them. Neither the sums raised nor the revenue were ever made public, so it is not possible to say how long the wait took. The ending is still unwritten. What remains is a clear shape. A founder chose infrastructure over an app, picked the least glamorous layer of a coming regulatory change, and got incumbents to fund it. Backing from NAB, Westpac, Salesforce and Visa meant potential customers were also investors. For first-time founders, the point is simple. If your product depends on a rule change, find the people the change affects most and bring them in early. Sometimes your first investors are also your first customers. ## Aconex (Australia) URL: https://startup.talk/australia/aconex/ Founders: Leigh Jasper, Rob Phillpot Founded: 2000; Based in: Australia (Melbourne); Sector: Construction project collaboration software The name is a mash-up. Aconex stands for Australian Construction Exchange, a label that sounds humble for a company that ended up managing projects worth hundreds of billions of dollars. Leigh Jasper and Robert Phillpot founded it in 2000 on a plain observation: building sites generate mountains of paper, and when something goes wrong, nobody can prove who said what. Accounts of Jasper's childhood and schooling are not well documented, so what follows is the business. The product put drawings, instructions and approvals into one shared online record. For the people paying for a tower or a railway, that record was insurance against disputes, which is exactly what makes construction so expensive. Growth was slow and then global. In 2006 the company won an Emerging Exporter award. In September 2008, in the thick of the financial crisis, the US firm Francisco Partners put in A$107.5 million of private equity, an enormous vote of confidence at a frightening time to bet on property. The company listed on the ASX on 9 December 2014, by which point its projects reportedly carried an aggregate value above A$800 billion across more than 40 offices. Then it bought its way into new territory: the Incite Keystone platform from CIMIC in August 2015, and Germany's Conject for A$96 million in March 2016. Revenue for the 2016 financial year was A$123.4 million. The ending came from California. In December 2017 Oracle agreed to pay A$7.80 per share in cash, valuing the business at about US$1.19 billion. Shareholders approved in March 2018, and the Supreme Court of Victoria signed off the next day. Jasper did not disappear. In 2016, he appeared on the list of angels who backed a young Sydney super fund, Spaceship, alongside Mike Cannon-Brookes. The founder of a company built on trust between strangers was now lending his name to someone else's. The quiet lesson here is about the choice of problem. Aconex did not promise to change construction. It offered to settle arguments in it. Founders in old, slow industries can take that: sell the cure for the expensive pain, not the vision of a better world. ## Kogan.com (Australia) URL: https://startup.talk/australia/kogan-com/ Founders: Ruslan Kogan Founded: 2006 (launched 2006-07; Kogan Technologies); Based in: Australia (Melbourne); Sector: E-commerce / online retail At six years old, Ruslan Kogan and his cousin sat tugging at their own baby teeth. The plan was simple: the Tooth Fairy paid coins, so more teeth meant more money. He has told Forbes Australia the instinct to turn something small into cash never really left him. He was born in Babruysk in Soviet Belarus and arrived in Melbourne around 1989 with his parents and sister. The family landed in the housing commission flats of Elsternwick. His parents had no business background, and both worked three or four jobs after migrating. "I saw that my parents would just pursue opportunities and take risks and work really hard," he has said. So the boy did the same. Near the golf club, he collected lost balls, cleaned them and sold them back to golfers. At Melbourne High he built websites for money. He studied IT at Monash, then took a consulting job at Accenture, where, by his own account, he did not enjoy corporate life. At night he sold imported T-shirts and wallets on eBay. What almost nobody knew was how little was behind the next step. In 2006 he listed plasma and LCD televisions made in Chinese factories on eBay as pre-orders, with auctions starting at one cent. Customers paid before the goods existed. He put the first shipment on credit cards. People told him it would fail, and no investor wanted to know. He quit Accenture soon after the first shipment of mostly plasma screens was fulfilled. The pre-order trick was the engine. Money from one batch bought the next, and the company never raised a dollar from outside investors until its ASX listing in July 2016, about ten years in. By then customers had heard of the brash founder who dared JB Hi-Fi's boss to a A$1 million bet, sold free HDMI cables to mock retailers' margins, and put a parody product up against the internet filter. The bravado had a price. Regulators ordered it to change price-comparison ads in 2009, a court fined it A$350,000 in 2020 over false discounts, and a Mighty Ape purchase turned painful, with a write-down that pushed the group into a loss. In 2026 an ABC investigation questioned pricing on its exclusive brands, which Kogan denies. Yet the business is bigger than the garage it began in. In the year to June 2026 the group moved about A$1.04 billion in gross sales and made A$41.8 million in adjusted EBITDA, though the shares fell sharply on the day. The lesson sits in those baby teeth. Kogan never waited to be funded or approved. He found the cheapest way to start, let customers pay first, and kept control until he had real numbers to show. ## Vinomofo (Australia) URL: https://startup.talk/australia/vinomofo/ Founders: Justin Dry, Andre Eikmeier Founded: 2011 (Qwoff wine review site from c.2006); Based in: Australia (Adelaide / Melbourne); Sector: Online wine retail / e-commerce It was just after midnight on 1 July 2013, and Justin Dry watched the sales screen. The first night back under the founders' control brought in A$230 at 12.02am. Then nothing much. To pay that week's wages, the team still had to sell another A$20,000. Dry has said as much in an interview with Startup Daily. It was the low point of a year that had started like a win. The story begins earlier with Qwoff, a community wine review site. Andre Eikmeier, Dry's brother-in-law, had the idea around 2006. Dry came back from overseas excited about Facebook, and together they built the site. People loved it. Nobody paid for it. For years they tried to make money and lost some of their own, which is where, Dry has said, they learned the most. They became known as the social media wine guys. Then they saw Groupon. They moved all their resources to daily wine deals, and in roughly three months were hunting for capital. What almost nobody knew was how close they came to a different life. They were nearly done with a major media company when Catch of the Day called, about a week before signing. The founders liked its culture and track record. They needed customers more than cash, they said, and the deal closed in four to six weeks, with Catch taking 70 per cent. It did not fit. A year on, Dry made what he called a really awkward call to Catch of the Day's co-founder to ask for the business back. On 1 July 2013 they had it again, and a quiet first night of A$230. The wages held. By 2016 Blue Sky Venture Capital put in A$25 million, then the largest single round for an Australian tech start-up with no overseas investor. Dry and Eikmeier kept their majority. The cost showed up later. Eikmeier stepped back from daily work. In 2021 Dry stepped aside as CEO as the company prepared to list, and then pulled the float when the pandemic sales boost looked temporary. Profits slid to breakeven and then one loss-making year. Dry returned in late FY2023, and Forbes Australia reports turnover near A$90 million and a goal of A$100 million in FY27. For founders: the first night after you take your company back may be the smallest number you ever see. Keep the lights on, and the audience you built for free will carry you. ## Menulog (Australia) URL: https://startup.talk/australia/menulog/ Founders: Dan Katz, Leon Kamenev, Kevin Sherman Founded: 2006; Based in: Australia (Sydney); Sector: Online food delivery marketplace On 26 November 2025, an app that millions of Australians had used for takeaway stopped working. Menulog, which began in Sydney in 2006, closed its Australian operations. Its customers, restaurants and couriers were handed to Uber Eats. The end is the part most people remember. The beginning is quieter. Dan Katz, Leon Kamenev and Kevin Sherman started the business in Sydney, and little about their early lives is on the public record. That matters, because in this story the company is the character that shows its hand. The early years were methodical. Menulog opened Brisbane, Canberra and Melbourne in 2007, then Adelaide and Perth in 2008. An iPhone app followed in 2009 and an Android app in 2011. In 2012 it crossed the Tasman to New Zealand. By July 2014 it had handled over 22 million meal orders. Then came the deal that made it famous. In early 2015 it agreed to merge with rival EatNow, with Menulog holders taking 70 per cent. Months later, Britain's Just Eat bought the merged group, paying with new shares valued at about A$855 million. For a company founded less than a decade earlier in Sydney, it was one of the larger exits in Australian internet history. Selling to a global parent meant the founders' story ended and the platform's story passed to others. In 2016 Menulog lost a Victorian tribunal case brought by a pizza shop, Pizza Fellas, over use of its brand name in Google Ads, a small reminder that marketplaces fight over their restaurants' names. In 2016 it reached ten million orders in a year. Then the pandemic pushed demand up, and the company halved its pickup commission. But the market had changed. Just Eat merged with Takeaway.com in 2020. In May 2024 Menulog left New Zealand. Eighteen months later the Australian business closed, with the parent saying it wanted to focus on other markets. In the end Menulog shows two things about platforms. A marketplace built on local restaurants can be copied by anyone with money and couriers, and a founder who sells at the peak gives up the vote on what comes next. It also shows that a 2006 idea, executed city by city, could be worth hundreds of millions of dollars before the market got crowded. ## Humanforce (Australia) URL: https://startup.talk/australia/humanforce/ Founders: Bruce Mackenzie Founded: 2002; Based in: Australia (North Sydney); Sector: Workforce management / HR software (SaaS) Before Bruce Mackenzie flew into a roadshow of eleven venture capitalists, he rehearsed on people who he did not think would invest. It was practice, he has said, for a pitch he had avoided for seventeen years. Humanforce began in Sydney in 2002. Little public detail exists about where Mackenzie grew up or studied, so the story starts with the work. He built software that helped employers roster staff, track hours and feed payroll. No outside money came in. He has told SmartCompany that the worry never left: could the company cover payroll every fortnight, and what was coming over the horizon? "We never knew what was coming over the horizon," he said. There was little room to fail. The company grew just fast enough, he says, "to keep our heads above water." It went through three major growth phases, each of which drained cash. The switch to a subscription model came about 14 years in, a long wait by software standards, and the business later turned toward gig and flexible workers. What almost nobody knew was how that independence looked from outside. When venture capitalists finally met him, many asked the same blunt question: why had he not raised money thirteen or fourteen years earlier? Mackenzie's answer is honest, even a little rueful. In hindsight, he said, raising earlier would have been smarter. In January 2019 Accel-KKR put A$22.5 million into the company, which by then had 114 staff and had grown about 40 per cent a year for five years. The money was earmarked for North America. A second investment followed in May 2022, and then came acquisitions, including intelliHR in 2023, reportedly for more than A$77 million, and the Australian operations of Wagestream. The company says it now serves about 2,000 customers and more than half a million users, with offices in Australia, New Zealand and the UK. In January 2026 it named Cameron Partridge chief growth officer. The lesson is double-edged. Going without outside money gave Mackenzie control and a business that paid its own way. It also meant years of fortnight-to-fortnight stress, and he himself says he would do it differently. Do not wear self-reliance as a badge if a good partner could have taken the weight. ## Lendi (Australia) URL: https://startup.talk/australia/lendi/ Founders: David Hyman, Sebastian Watkins, Martin Lam, Mark Kalajzich Founded: 2013; Based in: Australia (Sydney); Sector: Fintech / mortgage broking The first money David Hyman and Sebastian Watkins ever put into a business was borrowed from their parents. Hyman has described the pair, who met at their first job out of university, as people who never really wanted to work for someone else. Their first business together ended in an exit with LivingSocial, as a 2023 Fintech Chatter podcast episode recounted. That meant that when they turned to home loans in 2013, they already knew what it felt like to sell a company. With Martin Lam and Mark Kalajzich they started Lendi. The target was enormous and unglamorous: the mortgage market, dominated by four banks. Lendi set out to make borrowing feel like an app, while keeping a human broker in the loop. The podcast notes describe the early years as bootstrapping, raising capital and what they call blitz scaling. What almost nobody outside the industry saw was who ended up on the share register. Banks that Lendi was meant to challenge, including CBA, ANZ and Macquarie, became shareholders. The turn came in 2021. Lendi merged with Aussie Home Loans, the broker franchise owned by CBA, and brought in Domain Home Loans, too. The ACCC approved it in March and the deal closed in early May. Hyman became group CEO, with Watkins as chief operating officer. The combined business included more than 1,000 Aussie brokers, and in the first full year it recorded A$33.6 billion of settlements. There is a price to being swallowed by the system you set out to beat. The start-up that sold itself as an alternative now sits inside the network it once chased, and its founders' identity as outsiders has shifted. By the time of its tenth anniversary in 2023, Lendi described itself as Australia's fifth-largest mortgage distributor, handling around A$30 billion of home loans a year. For first-time founders, the interesting detail is the very first cheque. Two friends who borrowed from their parents ended up running a platform that banks wanted to own part of. Small family money, spent well, can buy a very long runway. ## Propeller Aero (Australia) URL: https://startup.talk/australia/propeller-aero/ Founders: Francis Vierboom, Rory San Miguel Founded: 2014; Based in: Australia (Sydney); Sector: Drone surveying / construction and mining 3D mapping software Francis Vierboom likes to describe the box that made his company credible in plain terms: a pizza box with a solar panel on it. The joke carries a small history. Before Propeller, Vierboom had been part of Flirtey, a drone delivery startup. Delivery was the idea that drew the headlines, and he knew what that was worth and what it was not. A drone that drops a parcel makes the news. It does not necessarily make a business. So when he and Rory San Miguel started Propeller in Sydney in 2014, they did the unglamorous thing. They built drone solutions for paying customers first, watched what those customers kept asking for, and slowly discovered that the real prize was not the flying. It was the map. Mines and building sites wanted to know, to the centimetre, how much earth had moved. Drones could answer that, but only if the data could be trusted. That was the gap the pizza box filled. AeroPoint, a ground target with its own positioning, gave the surveys something solid to hang on. More than a thousand units had sold by 2017. The early years were not comfortable. Silicon Valley investors told them the drone bubble would burst, and Vierboom has said he disagreed because he saw high-value commercial uses. The founders started inside the Cicada Innovations incubator and the Startmate accelerator, leaning on mentors to avoid the common mistakes. Their own lesson, which Vierboom repeated to SmartCompany, was to avoid building too much product before chasing customers. The money followed that discipline. A seed round of about A$4 million led by Blackbird Ventures arrived in 2016, and the team rebuilt the product the same year. A Denver office opened in 2017. In 2018 came a US$10 million Series A, again led by Blackbird, with Sequoia China joining. By then Propeller had paying customers in 20 countries, mostly in construction and mining across Australia, the UK and the US, and it had outgrown its incubator desks for a Surry Hills office with a team of 25. Vierboom was candid about the temptation that follows a raise. Hiring too fast feels like progress, he warned. A Series B of US$18 million, co-led by Blackbird and Costanoa Ventures in 2020, tested that discipline again. The founders' path is a useful inversion of the usual drone story. A startup chasing attention for drone delivery was followed by a company that earned money by measuring dirt. For a first-time founder the insight is quiet: the stunt that gets you noticed is rarely the product that gets you paid. Vierboom seems to have learned that early, and he built the second company around it. ## Tanda (Australia) URL: https://startup.talk/australia/tanda/ Founders: Jake Phillpot, Josh Cameron, Alex Ghiculescu, Tasmin Trezise Founded: 2012; Based in: Australia (Brisbane); Sector: Workforce management / rostering / payroll software (HR tech) The idea for Tanda did not come from a boardroom. It came from a student guild office, and from the slow misery of working out what people should be paid. Three of the four founders, Jake Phillpot, Josh Cameron and Alex Ghiculescu, were executive officers of the QUT Student Guild in Brisbane. Running a guild means employing staff, and employing staff means timesheets. According to early coverage, the founders complained constantly about how long it took to process them by hand. Complaining is cheap. They decided to build the fix. Tasmin Trezise completed the founding four, all of them recent QUT alumni. Tanda dates itself to 2012. The first thing they built was not a grand platform. It was software that calculated Modern Award conditions, the thicket of Australian pay rules that decides what a casual worker earns on a Sunday or after a late shift. It is the sort of problem that looks boring until you get it wrong and underpay someone. Outside validation came early. In 2014 the team won the qutbluebox Innovation Challenge and its A$60,000 first prize, money they said they planned to use to take the product global. Then the company did something unfashionable. It did not chase venture capital. Tanda added rostering, payroll and HR products, hired compliance experts to track Fair Work rules, and grew on its own earnings. It picked up large customers along the way; Domino's chose the Brisbane start-up to overhaul its workforce management. The company describes itself as founder-led and independent. For fourteen years that stayed true. In September 2026 it was reported that Thoma Bravo, the US software investor, had led a growth equity round, Tanda's first outside money, with Phillpot still CEO. About 8,000 businesses in hospitality, retail, healthcare and quick-service restaurants now use the product. What lingers is how ordinary the origin is. Nobody in that guild office was looking for a billion-dollar market. They were irritated by a spreadsheet. The lesson for a first-time founder is to treat your own recurring annoyance as a product spec, and to be patient enough to let the compliance detail that frustrates everyone else become your moat. ## Power Ledger (Australia) URL: https://startup.talk/australia/power-ledger/ Founders: Jemma Green, David Martin, John Bulich Founded: 2016 (idea conceived 2015); Based in: Australia (Perth); Sector: Blockchain peer-to-peer energy trading / climate tech Jemma Green was writing a doctoral thesis about households that make their own power when she noticed who was left out. Rooftop solar was spreading across Perth, but not to people who lived in apartments. She was born in Perth and raised in the eastern hills. After a degree in finance she started her career in banking at J.P. Morgan in London, and spent years on environmental and social issues inside a bank. Then she came home and went back to study, taking on a PhD in sustainable energy. The question she kept returning to was simple. If someone has surplus solar, why can't they sell it to the neighbour next door? In 2015 she met David Martin at an industry gathering. Martin had around twenty years in utilities and knew how slowly the grid moves. Between the banker-turned-researcher and the utilities man came the idea for a blockchain ledger that could record each tiny electricity transaction and settle it in real time. They founded Power Ledger in Perth in 2016, with John Bulich as a third co-founder. The money arrived in an unusual way. In 2017 the company ran an initial coin offering. The pre-sale raised about A$17 million and sold out within 72 hours on the Ethereum network. Martin had predicted the final raise would land between A$20 and A$30 million. It closed near A$34 million, one of the largest successful ICOs by an Australian startup. Green did not pretend the instrument was safe. She acknowledged that ICOs were highly speculative investments, and said the company set itself strict self-imposed rules around the sale. That candour matters: tokens, not equity, funded the build. Recognition followed. In 2018 Power Ledger won Richard Branson's Extreme Tech Challenge, and Green was named EY Australian Fintech Entrepreneur of the Year. Origin Energy trialled the peer-to-peer platform, and the company set about proving the technology in real energy markets. The price of that fast start is a long tail of expectation. A token boom raises money quickly and then asks you to justify it for years, across crypto cycles and slow-moving regulators. Green's origin story is a quieter lesson: a researcher's irritation with a gap in the system, paired with a partner who knew how the system worked, can be enough to start something. The ICO was the megaphone. The thesis question was the engine. ## Prospa (Australia) URL: https://startup.talk/australia/prospa/ Founders: Greg Moshal, Beau Bertoli Founded: 2012; Based in: Australia (Sydney); Sector: Fintech / small business lending Prospa's own first loan was small: A$20,000, written in 2012 for a business owner whom the banks had passed over. The two men behind it knew exactly what that owner felt like. Greg Moshal began in a family business before running his own, and the struggle to get finance was familiar to him. Beau Bertoli had spent five years at FlexiGroup, the listed financial services company, and by his own count Prospa was the fourth business he had started. They were introduced by a mutual friend and bonded, as the story goes, over the pitfalls of small business and money. Moshal holds a Bachelor of Commerce in accounting from Monash University. Together they pitched an online lender for small firms, with seed backing from Entree Capital. The idea was straightforward. Banks asked for property security and weeks of paperwork. Prospa asked for a few data points and decided quickly, using a credit engine that now weighs more than 450 signals. Loans ran from A$5,000 to A$1 million. It grew into Australia's largest fintech lender, and AirTree put in A$25 million in 2017 at a valuation of A$235 million. Then came the stumble. The company's float was delayed days before opening after the corporate regulator raised concerns about the fairness of its contracts. The delay lasted about twelve months. It re-emerged and listed on the ASX on 11 June 2019, selling 29 million shares at A$3.78 to raise A$109.6 million. The shares rose 18% on debut, and the company was valued at about A$610 million. The price of public life came quickly. The pandemic hit small business borrowers hard, and by 2024 the shares were trading around A$0.45, far below the IPO price. A Salter Brothers Tech Fund-led consortium took the company private that year. Prospa has kept going. In 2024 it bought the Zip Business loan book of about 370 small businesses, and in March 2026 it partnered with Qantas Business Rewards. The insight for a founder is unsentimental. A float can validate a business and still not protect it from a bad cycle, and a company built to lend to the smallest, most fragile firms carries their risk in its own numbers. ## Ansarada (Australia) URL: https://startup.talk/australia/ansarada/ Founders: Sam Riley, Andrew Slavin, Rachel Riley, Daphne Chang Founded: 2005; Based in: Australia (Sydney); Sector: Virtual data rooms / M&A deal-management software (SaaS) Sam Riley once told investors the company name was a puzzle made from the names of the people who started it. Ansarada was founded in Sydney in 2005. Riley, the chief executive, started it with three others: Andrew Slavin, who became chief technology officer, Rachel Riley, who later ran the governance, risk and ESG business, and Daphne Chang. The name, as Riley explained it on an earnings call in 2023, was built from the founders' names. It was an unusually personal choice for a company that would spend nearly two decades in the cold business of mergers and acquisitions. That business was virtual data rooms, the secure online spaces where buyers read a seller's confidential documents. The big names, Datasite and Intralinks, were American. Ansarada was a small Sydney outfit that competed on guidance rather than storage, helping deal teams see what they were missing. It grew out of a serviced office, then another, as work poured in. For years the founders and some employees owned most of the company. It later listed on the ASX, with the code AND. The founders had more than a data room in mind. Alongside it Ansarada built governance, ESG and board-management tools, products that were early and, as the company later admitted, cash-flow negative. Then in February 2024 a rival came to the door. Datasite offered A$2.50 per share in cash, a 19% premium to the last close, by scheme of arrangement. Reported deal values ranged from about A$212 million to A$263 million depending on how they were counted. Datasite wanted the data room and deal products, not the early-stage ESG, GRC and Board businesses. So Riley took them. He agreed to carve those units out for himself, a quiet detail that tells you what he cared about. He described the sale as the culmination of almost eighteen years of work to improve how deals get done. The human detail is that a founder sold the thing he was best known for and kept the part still unproven. For a first-time founder, it is a reminder that an exit is not always an ending. Sometimes the buyer wants the finished product, and the founder keeps the question that was never answered. ## Bigtincan (Australia) URL: https://startup.talk/australia/bigtincan/ Founders: David Keane, Geoff Cohen Founded: 2011; Based in: Australia (Sydney; later Waltham, Massachusetts, US operations); Sector: Sales enablement software By the time David Keane started Bigtincan, he was no beginner. He had already built two companies of his own, and neither of them was a software darling. His early career was in networking. He began at Novell, then ran Asian marketing for the Boston networking firm Wellfleet Communications. Later he founded Quadtel Asia in Singapore and Veritel Wireless in Australia. These were the years of boxes, cables and carriers, not apps. That history matters, because it explains the bet he made around 2011. Salespeople were carrying phones and tablets into meetings, and the documents they needed were stuck on laptops and file servers. Bigtincan started in Sydney as a way to put the right content in a rep's hand, on the device they already had. A second name appears beside Keane's in some accounts, Geoff Cohen, as co-founder. The company went public on the ASX in the mid-2010s, and it grew the way many Australian software firms do: by buying. It picked up smaller tools for coaching, content and analytics, and bundled them into one sales enablement platform sold to large customers in the United States. The centre of gravity shifted from Sydney to Massachusetts. Then came the harder years. Sales enablement became a crowded category with well-funded rivals such as Seismic and Highspot. Bigtincan's operating revenue and other income slipped from about A$123 million to about A$117 million in the 2024 financial year, and the share price sat far below where investors had hoped. In December 2024 the American private equity firm Vector Capital offered A$0.22 a share in cash, roughly A$183 million for the whole company. Keane, as co-founder and CEO, was expected to stay on. The deal closed in April 2025, and the stock left the ASX. Within months Vector had also bought Showpad, and the two businesses were combined under the Showpad name, so the Bigtincan brand itself faded. For a founder, the lesson is in the order of things. Keane's earlier ventures taught him how to sell to big buyers and how to run across continents. The mobile-first idea gave him an opening, but categories fill up fast, and a first-mover lead is only worth what you do with it before the big players arrive. Sometimes the best exit available is the one that keeps the product alive under someone else's roof. ## Hipages (Australia) URL: https://startup.talk/australia/hipages/ Founders: Roby Sharon-Zipser, David Vitek Founded: 2004; Based in: Australia (Sydney); Sector: Online marketplace for tradies and home services The company that now sends hundreds of thousands of job requests to Australian tradies began in a garage, as a plain directory of trade businesses. Two friends started it in Sydney in 2004: David Vitek and Roby Sharon-Zipser. One report describes them as childhood friends. Nothing about the first idea was grand. It was a listing book, moved online, for plumbers, sparkies and builders. That was also the problem. A directory makes money when someone pays to be listed, and a tradie who is listed but never called has little reason to renew. For years the founders ran a business that looked like the internet but behaved like the yellow pages. The turn came in 2011. They pivoted into an online marketplace. Instead of waiting to be browsed, the platform took the job a homeowner wanted done, matched it with tradies, and charged the tradie for the work sent their way. Homeowners got quotes without ringing round. Tradies got customers without door-knocking. Revenue became something a tradie wanted to keep paying for. Money followed in stages. Before listing, the company raised about A$50 million from venture firms including KTM Capital, Right Click Capital and Kresta Capital. Then in November 2020, in the thick of a pandemic renovation boom, hipages listed on the ASX, raising A$100.4 million at A$2.45 a share and opening at a valuation around A$318 million. Public life has been less smooth than the pitch. In the 2025 financial year revenue grew 10 percent to A$83.1 million and EBITDA before significant items rose 20 percent to A$19.6 million, yet net profit fell by about a third, to roughly A$2.4 million, as the company moved its platform and kept investing. Sharon-Zipser still leads the business as CEO and co-founder. The takeaway for a founder is that the first version of an idea is often the wrong business model for the right customer. Hipages had the customer from day one, the tradie. It took seven years and a pivot to find what that customer would actually pay for. ## Sendle (Australia) URL: https://startup.talk/australia/sendle/ Founders: James Moody, Sean Geoghegan, Craig Davis Founded: 2014; Based in: Australia (Sydney); Sector: Parcel delivery / logistics for small business Within a year of launching, Sendle was being challenged by the company it set out to beat. In 2015 Australia Post filed a trademark dispute against the startup's slogan, 'Post without the office'. A national monopoly against a team of founders in Sydney: James Moody, Sean Geoghegan and Craig Davis. The ruling did not come until 2017, and it went Sendle's way. The pitch behind the slogan was simple. Small online sellers hated the post office queue and the price gap between their parcels and a big retailer's. Sendle offered pickup from the door, flat prices and, from 2015, a claim to be Australia's first carbon neutral delivery service. By 2016 it had passed 100 million kilometres and won a Telstra Business Award for new business in New South Wales. Then it widened the net. A DHL eCommerce partnership in 2017 opened international sending, and plug-ins for eBay, Shopify, WooCommerce and Xero put it where sellers already worked. It entered the United States in 2019 and Canada in 2022. Across rounds from A$5 million in 2016 to A$45 million in 2021, and a further A$16 million in 2024, it raised about A$100 million. The ending was fast. In 2025 Sendle merged into a new group, FAST Group, with the US firms ACI Logistix and FastMile. In December 2025 an investor fund suspended redemptions, citing significant deficiencies in ACI Logistix's financial statements. Federation Asset Management, Sendle's key investor, said the partner was not current on its financial obligations at the time of the merger. In January 2026 Sendle halted bookings in Australia and Canada. In February it went into creditors' voluntary liquidation. Moody did not respond to press requests for comment. In July 2026 McKenna Worldwide bought the brand, trademark and domain, though not the company or its debts, and the website came back in October. The lesson is uncomfortable. Sendle won the fight it started with a monopoly and survived years of competition. What it could not control was the quality of a merger partner's books. For a founder, the company you merge with becomes your risk, and due diligence is worth more than the headline growth story. ## Bugcrowd (Australia) URL: https://startup.talk/australia/bugcrowd/ Founders: Casey Ellis, Serg Belokamen, Chris Raethke Founded: 2012 (Ellis has said 2011); Based in: United States (San Francisco, California); founded in Sydney, Australia; Sector: Crowdsourced cybersecurity / bug bounty Casey Ellis sold his time by the day. That was the whole business he worked in, and it bothered him. Ellis grew up hacking stuff as a kid, as he has said in a podcast interview. He turned that into a career as a penetration tester and security researcher, then into solutions architecture, sales and security leadership. Along the way he worked with Serg Belokamen at a small consulting firm that sold security testing, one day at a time, at around A$2,000 a day. The discomfort was the model. A client paid for a day whether or not anything serious was found. Meanwhile Facebook and Google had shown a different approach: pay outsiders only when they report real vulnerabilities. Ellis has described the company's original idea as a napkin sketch on a flight. He and Belokamen joined the Sydney accelerator Startmate and launched Bugcrowd in 2012. The core idea was to pay for results, not hours. There was a second one. Testers could build a public reputation through the bugs they fixed, which helped them win more work. Then came the move. In 2013 Ellis went to Silicon Valley, because the buyers were there, and he has said the company would not have grown as quickly had it stayed in Sydney. Belokamen stayed in Australia and left. Chris Raethke, whose own Startmate company had failed, joined as a co-founder. The money followed: a seed round of US$1.6 million led by Rally Ventures in 2013, then US$6 million, US$15 million, US$26 million and US$30 million rounds through 2020, about US$78.7 million in all. Western Union signed on in 2015, and in 2020 National Australia Bank launched a bug bounty with Bugcrowd's help. In 2024 the company bought Informer and in 2025 Mayhem Security. What Ellis built was less a tool than a marketplace of people. For a founder, the lesson is to look at the invoice you hate sending. If your customers pay you for hours but really want results, the business model itself may be the product. ## Finder (Australia) URL: https://startup.talk/australia/finder/ Founders: Fred Schebesta, Frank Restuccia, Jeremy Cabral Founded: 2006; Based in: Australia (Sydney); Sector: Comparison and consumer finance media Fred Schebesta has described turning up to client meetings in shoes worn almost through. He was in his twenties and already running a business. At 22, while finishing a Bachelor of Finance at Macquarie University, he and school friend Frank Restuccia had started a digital agency called Freestyle Media in 2002. There was a version of success in that: Schebesta was named Australian Young Direct Marketer of the Year in 2007, and the agency was bought by the ASX-listed Q Ltd. The two stayed on to run it until late 2009. But money was tight. He has said they lived in a share house with five other people for about two and a half years, and that cash flow was one of their biggest challenges. They chased overdue invoices even when only a few cents were owed. He describes himself as someone who hates being ripped off. That instinct became the business. In 2006 the two launched finder.com.au, and it began as a family of sites with names like creditcardfinder.com.au and lifeinsurancefinder.com.au. The idea was a frugal one: help people compare products and avoid overpaying. The money came from referral fees when readers took up a card, loan or policy. The risky move was merging the many domains into a single brand. Schebesta has said it could have made all of the sites vanish from Google. The team pulled it off almost seamlessly. In 2015 they repeated the exercise in the United States, which meant starting again from the blueprint. For 15 years, Finder took no outside money. When it finally did, in December 2021, it raised A$30 million from Future Now Capital at a pre-money valuation of A$650 million, with revenue around A$96 million and a net profit of A$1.7 million for FY21. Third co-founder Jeremy Cabral later stepped down after 16 years to pursue AI. The human lesson is cheap and unglamorous. The share house and the worn shoes were not a phase before the business. They were a habit of mind that the business was built to sell: respect for other people's money starts with your own. ## Gojek (Indonesia) URL: https://startup.talk/indonesia/gojek/ Founders: Nadiem Makarim, Michaelangelo Moran, Kevin Aluwi Founded: 2010; Based in: Indonesia (Jakarta); Sector: Ride-hailing / super app (transport, delivery, payments) In 2015 the Transport Ministry banned motorbike taxi apps. By the end of the same day the ban was gone, overturned after President Joko Widodo criticised it. It was the first of many fights Gojek would have with the state, and the cheapest one it ever won. Nadiem Makarim was not supposed to be a motorbike man. He was born in Singapore in 1984 to Indonesian parents, and his mother's father, Hamid Algadri, had been a fighter in the independence struggle. He went to the United World College of Southeast Asia in Singapore, then to Brown University for international relations, then Harvard Business School. He spent about three years at McKinsey. Everything pointed to a boardroom. What bent the path was the ojek. Back in Jakarta he rode motorbike taxis constantly, and he saw the waste on both sides: drivers waiting for hours with no fare, passengers walking the pavement hunting for one. Before an app existed, he started with a telephone. In 2010 Gojek opened as a call centre with about 20 drivers. A friend from high school, Michaelangelo Moran, designed the first logo. For years it stayed small and unglamorous. What almost nobody remembers is how thin that first version was: a switchboard, a handful of riders and a promise. Money came late. A first round arrived from NSI Ventures at the end of 2014, then Sequoia India and Northstar in early 2015, just as the app finally launched in January. Nearly 30 million downloads followed within two years. Then the growth got expensive. Gojek added food, payments and dozens of lifestyle services, and by December 2019 it quietly shut most of the lifestyle ones because growth had stalled. In June 2020 it laid off 430 people. Drivers protested over tariffs and incentives again and again, in 2015, 2018, 2020, 2021 and 2024. Nadiem himself left in 2019 to join the cabinet as education minister. In May 2021 Gojek merged with Tokopedia to form GoTo, with 2020 gross transaction volume above US$22 billion. Years later his own story turned: he was named a suspect in a ministry procurement case in 2025 and, per reports, convicted in June 2026, which he has denied and said he would appeal. For a founder, the lesson sits in the 20 drivers. A multi-billion-dollar company did not begin with software. It began with a phone, a few riders and a plain observation about wasted time. The tech came after the pain was proven. ## Tokopedia (Indonesia) URL: https://startup.talk/indonesia/tokopedia/ Founders: William Tanuwijaya, Leontinus Alpha Edison Founded: 2009; Based in: Indonesia (Jakarta); Sector: E-commerce marketplace For years, William Tanuwijaya spent twelve hours a day in an internet cafe. He was not there to play. It was a part-time job that paid for his degree at BINUS University, and it lasted until he graduated in 2003. He came from Pematangsiantar, a town in North Sumatra, born in 1981 into a middle-income family that supported his schooling. After graduating he moved through ordinary jobs: software developer at Telkomsigma and Sqiva Sistem, game developer at Bolehnet, then IT and business development manager at Indocom Mediatama in 2006. A decent career, steady, safe. In 2007 he started building something else. The itch was simple. A person in a small town could not reach the products that a person in Jakarta took for granted, books for instance. Technology, he thought, could close that gap. He asked his friend Leontinus Alpha Edison to join him. The name was a mash-up of toko, Indonesian for shop, and encyclopedia: a store for everything. What almost nobody mentions is what else was happening that year. Around the same time his father was diagnosed with cancer. The company was being born under the shadow of family illness, which says something about how much a salaried job was worth against a dream still two years from launch. Tokopedia launched on 6 February 2009 and opened to the public on 17 August. The first money was small: Rp2.5 billion from PT Indonusa Dwitama. East Ventures, CyberAgent and SoftBank Ventures Korea followed. In 2014 Sequoia and SoftBank put in US$100 million. Then Alibaba led US$1.1 billion in 2017 and another US$1.1 billion in 2018, valuing it around US$7 billion. Growth came with scars. A data breach in 2020 reportedly exposed 15 million user records. The US trade office listed it as a notorious market over counterfeit goods. In 2021 it merged with Gojek into GoTo. In December 2023 TikTok agreed to buy 75% for about US$1.5 billion, and in 2024 around 70 staff were laid off. The pattern for founders: a cafe seat, a small-town frustration, and one friend willing to join. Scale came later. The founding impulse was personal and specific, and it survived a family crisis. ## Traveloka (Indonesia) URL: https://startup.talk/indonesia/traveloka/ Founders: Ferry Unardi, Derianto Kusuma, Albert Founded: 2012; Based in: Indonesia (Jakarta; group HQ also Singapore); Sector: Online travel and lifestyle Ferry Unardi was a Microsoft engineer in Seattle, working on real-time media for Lync, when he did something odd for a junior programmer: he booked a flight to China, on a whim, just to see what the market had to offer. He was born in Padang in 1988 and grew up there. High school brought programming competitions, then a move to Purdue University in Indiana, where he studied computer science and mathematics, graduating in 2008. At Purdue he met a classmate, Albert, who would later join him in building a company. Microsoft hired him straight after. It was the sort of job families brag about. But the China trip was a seed. He studied Mandarin and visited large Chinese e-commerce companies to watch how they worked. He resigned from Microsoft in 2011. What nobody expected was the thread that tied it together: a plain travel hassle. Visiting relatives in Indonesia was hard to organise from abroad, and he saw that the logistics could be simplified. Traveloka launched in September 2012 in Jakarta as a metasearch site, with Ferry as CEO and Derianto Kusuma as CTO. In 2013 it became a full agency with several payment options, which mattered in a country where credit cards were rare. East Ventures backed the seed, then Global Founders Capital. Hotels and apps followed in 2014, and by that year-end it had ten million monthly visitors. In July 2017 a US$350 million round from GFC, GIC, Qatar Investment Authority and Expedia made it a unicorn. Then 2018 saw Kusuma resign as CTO. In 2020 COVID crushed flight bookings to near zero. It laid off 100 people and refunded about 150,000 tickets, roughly US$100 million. By the end of 2020 the travel business was profitable again. IPO plans with a SPAC were dropped in 2021, and by January 2022 it was valued at about US$3 billion. The quiet lesson: curiosity before certainty. A engineer with a safe job bought a ticket to look at a market he did not yet understand, and that trip set up everything after. ## Bukalapak (Indonesia) URL: https://startup.talk/indonesia/bukalapak/ Founders: Achmad Zaky, Nugroho Herucahyono, Fajrin Rasyid Founded: 2010; Based in: Indonesia (Jakarta); Sector: E-commerce / small-business digital services The first big job Achmad Zaky did as an ITB student paid IDR 1.5 million. It was quick-count software for a television station watching a national election. Another early venture, a noodle-making business, failed. He was born in 1986 in Sragen, Central Java, and started learning computers in 1997 from books and a machine his uncle gave him. At State High School 1 in Solo he won the National Science Olympiad in computer science. Then came the Bandung Institute of Technology, informatics engineering. He graduated and set up a consulting company called Suitmedia. Bukalapak, which means open a market stall, began in 2010 with Nugroho Herucahyono. The premise was to digitise small-business transactions. Early traction came from an unlikely corner: fixie bike sellers, which briefly made it known as a bike marketplace. Fajrin Rasyid joined to run the money. What almost nobody mentions is who made the early cheque. In June 2011 Takeshi Ebihara, a Japanese venture capitalist who had mentored the founders, invested Rp2 billion. The mentor came first, the money second. By 2013 daily transactions averaged around Rp500 million with more than 80,000 sellers. In 2014 Emtek joined the Series A, other backers included 500 Startups and GREE Ventures, and in November 2020 Microsoft added US$100 million. Then the company changed shape. Zaky resigned in January 2020, Nugroho stepped down as CTO, and Rasyid left for Telkom Indonesia in June. The IPO in 2021 was the biggest on the Indonesian exchange at the time, up to US$1.5 billion. About six months later the stock had lost nearly 66%, CEO Rachmat Kaimuddin resigned four months after listing and hundreds of staff were laid off. 2021 revenue was Rp1.9 trillion. The takeaway for a young founder: noodles failed and a bike marketplace worked, and a mentor stood behind both. Early failures are cheap tuition, and an early believer often matters more than an early cheque. ## Ruangguru (Indonesia) URL: https://startup.talk/indonesia/ruangguru/ Founders: Adamas Belva Syah Devara, Muhammad Iman Usman Founded: 2014; Based in: Indonesia (Jakarta); Sector: Education technology In November 2022, Belva Devara told the public he was sorry. Ruangguru, the education startup he had co-founded, was laying off hundreds of people. He admitted he had failed to anticipate the economic downturn, and promised affected staff their severance and entitlements. It was a hard turn for a company that had just posted its first profit. In 2021 it reported about US$3.7 million in profit on revenue near US$102.6 million, against a loss of about US$1.2 million the year before. That year Fast Company ranked it number 25 among the world's most innovative firms. Eleven months on, the mood was different. The company started in April 2014, when Belva and Muhammad Iman Usman set out to help students find online tutors. By August 2014, more than 1,000 teachers had registered. East Ventures was an early backer. A beta Android app launched in February 2016, with a full release in April. The model widened: school curriculum, exam prep, a revenue take of around 20% per transaction, Kompas reported. UOB Venture Management joined a Series B in July 2017. In December 2019 General Atlantic led a US$150 million Series C, and in April 2021 Tiger Global put in US$55 million. It took the brand abroad as StartDee in Thailand and Kiến Guru in Vietnam, and in 2022 bought Kalananti, a coding provider, and Schoters, a study-abroad platform. What almost nobody remembered afterwards was the other side of fast growth. In March and April 2021 outsourced workers alleged they had been paid less than the local minimum wage, a claim Belva answered on Twitter. Then came the layoffs. The founder's apology is the human moment. A company can be profitable in one year and cutting jobs in the next, and the founder is the one who has to say so in public. For a first-time founder, the lesson is blunt: growth capital is a loan against optimism, and you answer for it to your people. ## OVO (Indonesia) URL: https://startup.talk/indonesia/ovo/ Founders: Lippo Group (corporate founder); Karaniya Dharmasaputra (president director from 2019) Founded: 2017 (wallet launched Sept 2017; started as Lippo's mall loyalty app); Based in: Indonesia (Jakarta); Sector: Fintech / e-wallet In October 2017, a rule written a year earlier by Bank Indonesia caught up with Grab. Under the central bank's e-money licensing regime, e-commerce firms without a licence had to stop letting users top up balances. Grab, with millions of riders, suddenly had no wallet. OVO had one. Its e-money licence arrived in August 2017, and by September the wallet was live. It had begun life inside the Lippo Group, the conglomerate of the Riady family, as a way to reward shoppers across Lippo's malls. Nobody had planned for it to become the payment rail of a ride-hailing giant. The pairing made a quiet kind of sense. The Riady family was already an investor in Grab through Venturra. So the app that had been a mall loyalty tool was relaunched as 'GrabPay powered by OVO'. By 20 December 2017 it was the fourth most popular app on Google Play in Indonesia. That is the part of the story that is easy to miss. OVO did not win a customer fight; it won a licensing one. Competitors such as GoPay and Telkomsel's T-Cash had head starts, but OVO had paperwork, a retail empire for merchants, and a partner that needed it. When it first launched, it leaned hard on the Lippo name, and a survey found 63% of its users recognised it as a Lippo product. Money followed the usage. In December 2017, Tokyo Century was reported to have put in US$116 million for 20%. Grab invested in 2018. Tokopedia joined in March 2019, when CB Insights estimated OVO's value at US$2.9 billion. That October it was named Indonesia's fifth unicorn and its first in fintech. It bought a peer-to-peer lender, Taralite, and added investment and insurance products through partners such as Prudential and Manulife. Then the owners changed. In 2021 Grab lifted its stake to 79.5%, and in October bought out Tokopedia and part of Lippo's shares to reach about 90%. The company that started as a Lippo mall app was now mainly a Grab asset. Leadership shifted too, with Karaniya Dharmasaputra named president director in 2019 after Andrian Suherman. OVO is also a story with few visible heroes. There is no garage, no single founder's late night. It is a corporate decision, a licence, and a partner in a hurry. For a first-time founder the lesson is unglamorous and useful. Regulation can be a moat. When a rule forces a giant to look for a partner, the startup with the licence in hand is the one that picks up the phone. OVO's breakthrough was not an invention. It was being ready when the door opened. ## Xendit (Indonesia) URL: https://startup.talk/indonesia/xendit/ Founders: Moses Lo, Tessa Wijaya Founded: 2015; Based in: Indonesia (Jakarta); Sector: Fintech / payments infrastructure Xendit did not start as a payments infrastructure company. It started as a Bitcoin remittance idea, and it was one of the first Southeast Asian startups to be accepted into Y Combinator in 2015. Inside YC the team changed course. Bitcoin was not the product that would work in Indonesia, so they turned to something familiar from the US: a peer-to-peer payments app in the spirit of Venmo. Friends would send money to friends, simply and for free. It stalled, and the reason is the interesting part. The app could take money in, but there was no way to send money out at scale, around the clock, to any bank account in Indonesia. Banks closed their rails at night and on weekends. A consumer app cannot promise instant payment on top of plumbing that sleeps. So the founders, Moses Lo and Tessa Wijaya, did the unglamorous thing. Rather than keep tuning an app, they built the pipes themselves, negotiating directly with banks. In 2016, after a mini hackathon, they released APIs that let other businesses use that infrastructure. The customer changed from a person sending money to a developer who needed to collect and disburse it. That shift made everything else possible. One product, the virtual account, gave each payer a unique identifier, which let merchants reconcile who had paid what without a finance clerk matching transfers by hand. For online shops, marketplaces and, later, other startups, this was the thing they had been missing. The money arrived once the pattern was clear. Kleiner Perkins led the Series A. Accel led a US$64.6 million Series B. Tiger Global led a US$150 million Series C around September 2021 at a billion-dollar valuation, and in May 2022 Coatue and Insight Partners co-led a US$300 million Series D, taking total equity raised to about US$538 million. Xendit opened in the Philippines in 2020, its second market, and projected US$15 billion in payment volume for 2022, up from US$6.5 billion the year before. This is a company that was willing to abandon two ideas in its first year. First the Bitcoin remittance, then the consumer app. Neither failure was fatal because the learning was cheap and early. The lesson for a first-time founder is that the real product is often the problem the first product ran into. Xendit's customers were not the people it first pictured. They were the ones who hit the same wall. ## Kopi Kenangan (Indonesia) URL: https://startup.talk/indonesia/kopi-kenangan/ Founders: Edward Tirtanata, James Prananto, Cynthia Chaerunnisa Founded: 2017; Based in: Indonesia (Jakarta); Sector: Food and beverage / coffee retail On its first day, the first Kopi Kenangan outlet in Kuningan, Jakarta, sold 700 cups. It was a big number for a shop that nearly did not happen. Raising the Rp150 million for it had been hard, and it had taken six months of planning to open the door. The founder, Edward Tirtanata, had not begun in coffee. He started in his father's coal business, and it was hard going and unprofitable. He moved to consumer goods instead, launching a tea brand called Lewis & Carroll. That strategy was flawed too. So here is the part nobody sees from the queue outside a Kopi Kenangan today: the company is the third attempt. Coal, then tea, then coffee. Each exit taught something about how a brand is made rather than how a commodity is traded. The idea that stuck was plain. He wanted Starbucks-quality coffee at a lower price. Together with James Prananto and Cynthia Chaerunnisa he opened in 2017, and he gave the signature drink a name that people remembered. Kopi Kenangan Mantan is palm sugar milk coffee made with fresh milk, not condensed milk, and its name literally means 'coffee of ex-lover memories'. The idea was that customers would remember the drink the way they remember past loves. Growth was fast. More than 230 outlets opened within two years. Investors followed: Barclays took a stake in 2018, KKR in 2019, and in May 2020 a further round was led by Barclays, with Warburg Pincus, CVC and others taking part. By the end of 2022 the chain was opening about a store a day. In April 2022 it set a MURI record, opening 26 outlets in 13 cities in a single week. The brand then crossed borders under the name Kenangan Coffee, to Malaysia in 2022, Singapore in 2023, the Philippines in 2024, Sydney in 2025 and Taipei in 2026. It had more than 900 outlets in 2024 and cleared the 1,000-store milestone in 2025. By some accounts it uses about 1,500 tonnes of Indonesian beans a year. Through the coal years and the failed tea brand, he tried twice, did not quit, and the third idea carried the weight of the first two. For a first-time founder the takeaway is that the product that works is often a pivot from the one that did not, and the name on the cup can be as important as the beans inside it. ## Halodoc (Indonesia) URL: https://startup.talk/indonesia/halodoc/ Founders: Jonathan Sudharta Founded: 2016; Based in: Indonesia (Jakarta); Sector: Healthtech For years, doctors in Jakarta met a junior drug salesman called Budi Jonathan. He often waited past midnight to pitch medicines to overworked physicians. What they did not know was that Budi was Jonathan Sudharta, the son of the man who owned the company. Sudharta had grown up with the label that fathers dread. As a teenager he was a fighter and an average student. At 13 he and friends were badly beaten by students from another school. His father's friends, as Bloomberg later reported, worried he would one day ruin the family business, a pharmaceutical trading house his father started in 1975 that became Mensa Group. He went to Hale School in Perth, then studied commerce at Curtin University. He played bass in a band and produced films and concerts for Indonesian diaspora communities. Then he went home and joined Mensa as a trainee. The moment he points to as a turning point came on his first day, at a port. He handed out small bonuses, 1,000 rupiah each, about ten US cents. He wanted to give an elderly worker more, but a manager stopped him. The man's gratitude for so little stayed with him. 'It was a big slap in my face,' he told Bloomberg. 'That changed my life.' He hid his surname and worked his way up, spending about 13 years in the company. That is the quiet reveal in his story: the pseudonym. He chose to earn the doctors' trust as a stranger rather than as the boss's son, and the contacts he built, with doctors, hospitals, pharmacies and the health regulator, became the raw material for Halodoc. In 2016, at 34, he co-founded it. Nadiem Makarim of Gojek encouraged him to start his own company, and advised him to go where people feel the most pain. Sudharta had watched patients wait hours for a short consultation, then wait again for medicine. Halodoc offered online consultations, medicine delivery, home lab tests and hospital bookings. Gojek integrated it into its app in 2017. Mensa did not fund it, though Halodoc rented offices in a Mensa building. By July 2019 it claimed about 40 million users, 22,000 licensed doctors and 1,000 pharmacies, and the Gates Foundation's investment arm had joined an extended Series B, its first equity investment in an online health platform. In 2023, Astra Digital led a US$100 million Series D. The lesson is a small one with a long tail. The years of being a nobody in a doctor's waiting room were not a detour. They were the market research, and the only kind that cannot be bought. ## eFishery (Indonesia) URL: https://startup.talk/indonesia/efishery/ Founders: Gibran Huzaifah Amsi El Farizy, Chrisna Aditya, Muhammad Ihsan Akhirulsyah Founded: 2013 (8 October 2013); Based in: Indonesia (Bandung); Sector: Aquaculture technology / IoT On 31 July 2025, Indonesian police detained Gibran Huzaifah, the former chief executive of eFishery, along with two former executives. The man once named to Forbes 30 Under 30 Asia was now at the centre of a financial reporting scandal. Police said the first portion they could prove, in a separate matter, was Rp15 billion. How a catfish farmer's idea reached a US$1 billion valuation, and then collapsed, is what the public record shows. It is a story with an unfinished ending, and it is worth being careful with the facts. Gibran studied at the Faculty of Life Sciences and Technology at Institut Teknologi Bandung, class of 2007. As a student he developed a strong interest in fish farming. After graduating he ran a business called AvanyaG Corp, then, from 2011 to 2013, a catfish farming venture called Dorri Foods Indonesia. That is where he learned the industry from the pond up. On 8 October 2013, with Muhammad Ihsan Akhirulsyah and Chrisna Aditya, he co-founded the company that became eFishery. Its first product was an internet-connected feeding machine that dispenses feed to fish in medium and large ponds. In 2015 it received a Rp1.5 billion grant from Bank Mandiri to develop it. The company added software for monitoring shrimp and fish farms. Recognition followed. Forbes 30 Under 30 Asia in 2017, a stage at TED, programmes at Endeavor and Unreasonable, and executive courses at Harvard in 2019 and Stanford in 2023. In January 2022 came a Series C backed by SoftBank, Temasek and Sequoia India, and in 2023 a US$200 million Series D at a US$1 billion valuation. Then came the reversal. In late 2024 an audit alleged fraud going back to 2018. It reportedly found two sets of financial statements, one for auditors, shareholders and banks, and one showing the real condition. A reported profit of Rp261 billion masked, it said, an actual loss of Rp578 billion, and the company allegedly claimed 400,000 feeders when about 24,000 existed. Gibran and Chrisna Aditya were removed from their roles, and the company laid off about 100 staff in January 2025 and about 300 in February. These are allegations, and the case is ongoing; Gibran has been detained but the details of the police case have not all been made public. What the record supports is narrow but stark. A founder with real expertise in the thing he sold, a farm feeder that actually existed, stood on numbers that were reportedly not true. For a founder, the lesson is plain: the audit eventually arrives, and the thing it measures is not the pitch but the pond. ## Ajaib (Indonesia) URL: https://startup.talk/indonesia/ajaib/ Founders: Anderson Sumarli, Yada Piyajomkwan Founded: 2019; Based in: Indonesia (Jakarta); Sector: Fintech: retail stock, mutual fund and crypto investing app He was a first-time founder when he decided to ask other people for their money, and the people he asked were not impressed. Anderson Sumarli has since let slip what that felt like: investors looked at the age on his face and saw inexperience. Some said no. A few said no more than once. He had a CV that should have settled the question. A Cornell degree in applied economics and management, finished with top honours. Two years at IBM from 2014, then two years in the grind of management consulting at Boston Consulting Group. Yada Piyajomkwan, a Thai co-founder, joined him as chief operating officer. None of it helped much in the first meetings. In a country where fewer than one in a hundred people owned a share, a young man in a T-shirt promising to bring millions of strangers into the stock market sounded like a pitch deck, not a plan. Ajaib launched in 2019 with a plain idea. Stock trading in Indonesia felt like a bank counter: forms, minimums, jargon. The two founders built an app that felt closer to a messaging tool, aimed at first-time investors in their twenties. Mutual funds, stocks and, later, crypto, all in one place and all under licence from the financial regulator OJK. That licence detail matters more than it sounds. Plenty of apps were chasing the same crowd. Ajaib's founders worked through the paperwork of an actual brokerage rather than renting someone else's rails, and then packed the product with education so a nervous beginner would not freeze on the first screen. What nobody on the other side of the table expected was how fast the audience would arrive. The pandemic sent millions of young Indonesians home with time and a phone. In 2020 came a Series A, and in 2021 Ribbit Capital, a firm that knows consumer finance, backed a US$65m extension. By October 2021 DST Global led a US$153m Series B that pushed Ajaib past a US$1bn valuation, making it the seventh Indonesian unicorn and, by some counts, the fastest to get there at about two and a half years. The price of a rocket start is that the ground is always moving. Retail trading cooled after the boom. Rules changed. Competitors with deeper pockets and older licences fought for the same first-time investor. The company has said it reached some two million customers within roughly three years. The lesson in Sumarli's story is smaller than the headlines. The people who said no in 2019 were not wrong to want proof. He gave them proof by shipping, not by arguing. A founder who is underestimated has one cheap advantage: nobody is watching closely, so the work gets to speak first. ## Stockbit (Indonesia) URL: https://startup.talk/indonesia/stockbit/ Founders: Wellson Lo, Johny Susanto, Sigit Kouwagam Founded: 2012; Based in: Indonesia (Jakarta); Sector: Fintech: social investing community and stock trading app (also owns Bibit mutual-fund app) The idea did not start in Jakarta. It started in Melbourne, where two students, Wellson Lo and Johny Susanto, met and got to talking about the stock market back home. What they noticed was lopsided. Professional investors in Indonesia had terminals, analysts and a stream of private information. Ordinary people had rumour and WhatsApp groups. In 2013, only about 21.8 per cent of Indonesians were financially literate by the national measure. The gap between the two sides was the product. They built a social analytics site first, a place where beginners could see news, data and what other investors were saying, all in one feed. It was not a broker, not an app store hit, not anything an investor could easily value. For roughly three years it ran on the founders' own money. The seed cheque from Ideosource only arrived in 2015, and it was announced as the money that let the team finally hire. Sigit Kouwagam joined as the third co-founder and became chief executive, with Lo as chief product officer. Their pitch was always education. Kouwagam has put it plainly: there is still a huge gap in how Indonesians understand money, and information should be open to everyone, not a few. The community grew in that direction, one stock thread at a time. The turn was not a launch day. It was a slow accumulation. By 2019 the forum was big enough to attract East Ventures into a Series A, and in the same year Bibit, a mutual-fund app with a robo-advisor aimed at first-time investors, joined the group. Suddenly the company had two doors: one for people who wanted to chat about stocks, another for people who wanted to buy a fund for as little as the cost of a lunch. The pandemic then did what no marketing budget could. By mid-2021, more than 500,000 investors had joined Stockbit's social network, and with Bibit the group served over a million retail investors in more than 500 cities. Stockbit's users had more than doubled in a year, and Bibit's had grown about fivefold. That year, the company raised US$95m. The cost of success is regulation. A forum can say anything; a broker cannot. The team moved into securities in late 2021 and had to turn a community culture into a licensed, supervised business. If there is a lesson for a first-time founder, it is the order of things. Stockbit did not buy its audience and then look for something to sell. It spent years being useful for free, while paying for it from its own pockets, and let the money show up after the trust. ## Pluang (Indonesia) URL: https://startup.talk/indonesia/pluang/ Founders: Claudia Kolonas, Richard Chua Founded: 2019; Based in: Indonesia (Jakarta); Sector: Fintech: investing app (gold, US stocks, crypto, local equities) Before Pluang had a logo, it had a classroom. Claudia Kolonas and Richard Chua met at Harvard Business School as section mates, and the company's story starts there. Kolonas did not arrive at that classroom as a novice. For more than a decade she had worked inside Indonesian financial services at Celebes Capital, a venture fund that invested in banks and other financial institutions. She had seen from the inside who got served by Indonesian finance and who did not. The rich had brokers. The rest had a savings book and gold jewellery. That last detail became the product. Pluang launched in 2019 with gold, the asset that Indonesians already treat as savings. The idea was to let anyone buy a sliver of it from a phone. Not a complex derivative, not a stock tip, just a familiar thing in a new, smaller package. The first money was modest. A Series A of about US$3m in 2019 funded a small team and a licence-hungry start. Then the pandemic pushed Indonesia's retail investing into a boom, and the funding followed the users. In March 2021, a US$20m pre-Series B came from Openspace Ventures with Gojek's venture arm joining in. By September, Square Peg had led a US$35m Series B. In January 2022 Accel added US$55m, and the Series B reached US$110m. Axie Infinity co-founder Jeffrey Zirlin was listed among the investors, an odd detail for a gold app that had since added crypto. The broadening was the real strategy and the real risk. Gold bought the trust, but US stocks and crypto drew heavier regulation and sharper cycles. When crypto fell from its 2021 peak, apps that leaned on it felt the cold. Pluang's later US$10m round, led by MUFG Innovation Partners, came with a move into local Indonesian equities, a more conventional and safer base. The shape of the bet is simple. Start where people already are, with something they already hold, and expand only once trust is earned. For a first-time founder the quiet lesson is Kolonas's own: a decade of watching how money works for customers who rarely get a seat at the table is worth more than a clever app. She built the app for the customer she had spent ten years watching. ## Amartha (Indonesia) URL: https://startup.talk/indonesia/amartha/ Founders: Andi Taufan Garuda Putra Founded: 2010; Based in: Indonesia (Jakarta); Sector: Fintech: peer-to-peer microlending to rural women entrepreneurs His family thought he had lost his mind. Andi Taufan Garuda Putra was 22, an ITB graduate with a job at IBM, one of the names Indonesian parents love to say at weddings. Then in 2009 he resigned, and the new office was a village in Ciseeng, in Bogor, with rice fields on one side and a market on the other. To his family, it looked like a fall in status. He had been working at IBM for two years. He later described the job as a way to learn management before running something of his own. What he did not say was that the job felt empty; one profile quotes him saying it lacked meaning and the will to live. A second thing pulled harder. Around Ciseeng, loan sharks, the rentenir, charged farmers and small traders interest that kept them poor. Banks did not go there. With about Rp10m of his own savings, he started Amartha in 2010 as a Grameen-style microfinance institution. His first loans went to just five borrowers, none above Rp1m. Women formed the groups, because the idea, borrowed from Bangladesh, India and Latin America, was that shared responsibility did the work collateral could not. As he put it later, shared responsibility is the core value of Amartha. It worked, slowly. By 2014 he had about 5,000 customers, five branches and 40 staff. But growth came with a price. His first co-founder left in 2013 because the pace was too slow. From 2014 to 2016, the model hit a wall: there was demand for roughly Rp500m in loans and he could fund only Rp200m. Borrowers drifted away. The turn was a decision to take the lenders online. In April 2016 Amartha relaunched as a peer-to-peer fintech, letting ordinary people fund loans to village women. At the launch event, fewer than ten people showed up. Three months later, Rp10bn had been lent. Then the numbers outran the doubts. About Rp700bn disbursed in 2018, a Series B led by LINE Ventures in late 2019, and in 2023 a US$28m round led by Women's World Banking. A 2023 profile counted about Rp1.7tn lent to more than 360,000 women, with roughly 2,300 field officers walking the villages. There were rough patches too. In 2020, while serving as a presidential special staffer, he drew criticism for letters to district heads that appeared to promote his own company. What the story teaches is the order of things. Taufan did not begin with an app. He began with five borrowers and a family that disapproved. The technology came six years after trust. ## Mekari (Indonesia) URL: https://startup.talk/indonesia/mekari/ Founders: Suwandi Soh, Daniel Witono, Anthony Kosasih, Dirman Suharno Founded: 2015; Based in: Indonesia (Jakarta); Sector: B2B SaaS: accounting, payroll, tax and CRM software for SMEs Most small shops in Indonesia keep their books in a notebook and a drawer full of receipts. That mess of paper is where Mekari starts. The company began as Jurnal, a cloud accounting product run since 2015 by PT Jurnal Consulting Indonesia. Suwandi Soh is listed as the chief executive. With him are Daniel Witono on product, Anthony Kosasih on operations and Dirman Suharno, all named as co-founders in profiles of the group. Indonesia has tens of millions of small businesses, and a very small share of them used proper accounting software. Jurnal went after that gap with a plain offer: your books, online, simple enough for an owner who is not an accountant. East Ventures backed the company, and Mandiri Capital joined among its investors. By 2020 a business paper reported more than 10,000 users in several cities, and the company's own materials now speak of tens of thousands of firms. The turn came when Jurnal stopped being only an accounting tool. Around 2018 and 2019 it joined other products under the Mekari name. Talenta for payroll and HR, Klikpajak for tax filing, Qontak for customer conversations, Sign for digital signatures. A customer who started with the books could add salaries, taxes and sales without leaving the system. That bundle earned the group a major investor from abroad. In May 2022 Money Forward, the Japanese cloud accounting firm, led a US$50m Series E. The hard part is the unglamorous one. Software for small business is a long, slow sale to people who do not trust new tools. Integration of several products is also harder than naming them one brand. For a founder building for small businesses, Mekari's path suggests a simple rule: sell the boring problem first, then widen the offer once the customer already trusts you with their books. The most stubborn customers are the ones who stay. ## Sirclo (Indonesia) URL: https://startup.talk/indonesia/sirclo/ Founders: Brian Marshal, Leontius Adhika Pradhana, Andreas Thamrin Founded: 2010; Based in: Indonesia (Jakarta); Sector: E-commerce enabler / commerce SaaS and fulfilment Twice, Brian Marshal has said, his company came close to bankruptcy. Not in the scrappy early days, but after Sirclo had already raised tens of millions of dollars and was counted among Indonesia's better-funded e-commerce enablers. That is the part of the story investors rarely hear. The public version is tidy: a seed round from East Ventures in 2014, a Series A of roughly US$5 million in 2019, US$6 million more in 2020, then a US$36 million round in 2021 led by East Ventures and Saratoga, with Traveloka and Sinar Mas Land alongside. A name on every list of Indonesian startups to watch. The real shape was messier. Marshal has described a product that began as something very similar to Shopify, a tool to let any seller open an online store. It looked simple. Then the team learned what Indonesian sellers actually struggled with. Not the storefront, but everything behind it: getting parcels moved, answering customers, keeping stock straight. A website was the easy part. So Sirclo drifted toward the physical world. Warehousing, fulfilment, brand services, the unglamorous machinery that makes an online order arrive. It is slow, capital-hungry work, and it does not scale like software. Then the money tightened. Funding slowed after the pandemic boom, and a company built to grow on the next round suddenly had to live on its own cash. Marshal has said it came within reach of bankruptcy twice. What almost nobody saw from outside was what he chose to do. He did not chase another pivot. He stopped the experimental software projects, downsized a couple of teams, shut down a few initiatives, and went back to the services that had always earned money. It was a retreat, and he has been open about that. But it worked. In a 2026 conversation, Marshal laid out a plan built around consistent annual profit first, then financial checks before any possible IPO, and a logistics network built with local partners rather than warehouses in every city. The listing is an option he is keeping open, not a promise. There is something worth noticing in the order of that plan. For a decade, the startup playbook said raise, grow, raise again. Sirclo's near misses forced the opposite lesson: earn the right to raise. For a first-time founder, the takeaway is uncomfortable but useful. The business you start with is often not the one that pays you. Sirclo found its footing only after it stopped admiring its own ambition and went back to the oldest thing it did well. ## Flip (Indonesia) URL: https://startup.talk/indonesia/flip/ Founders: Rafi Putra Arriyan, Luqman Sungkar, Ginanjar Ibnu Solikhin Founded: 2015; Based in: Indonesia (Jakarta); Sector: Fintech / consumer payments and money transfer The product that would become one of Indonesia's best-funded consumer fintechs began with a student's irritation at paying to move his own money. Rafi Putra Arriyan was studying at the University of Indonesia when he ran into a small, constant tax on student life. Sending money from one bank to another cost a fee, and sometimes took longer than it should. It was not a crisis. It was the kind of annoyance everyone shrugs off. He did not. In 2015, with fellow students Luqman Sungkar and Ginanjar Ibnu Solikhin, he started Flip. Three classmates, no obvious pedigree in finance, taking on a problem the banks had been comfortable with for years. A year later the company had a licence from Bank Indonesia, a step that separates a campus idea from a real payments business. The pitch was disarmingly simple: transfer to any bank without the fee. Students and small online sellers, the people who most felt the cost of every small transfer, were the natural first users. For them, saving a few thousand rupiah each time was not a gimmick but a habit worth forming. For years the company grew quietly, away from the headlines that followed ride-hailing and e-commerce. Then investors noticed. In December 2021 Flip raised US$48 million in a Series B co-led by Sequoia Capital India, Insight Partners and Insignia Ventures Partners. It was Insight Partners' first investment in Indonesia. Six months later Tencent led a second close, with Block, the company formerly known as Square, joining, and the round passed US$100 million. By 2022 Flip reported more than seven million users. Public coverage says little about the founders' families or the early lean months, and we will not guess at them. What the record does show is a clear pattern: the founder personally felt the problem, and the first customers were people just like him. That is the useful lesson for a first-time founder. You do not need a grand theory of the market. A recurring, small pain that you feel yourself, repeated across millions of people, can be enough to build a company on. ## SiCepat (Indonesia) URL: https://startup.talk/indonesia/sicepat/ Founders: The Kim Hai, Rudy Darwin Swigo Founded: 2014; Based in: Indonesia (Jakarta); Sector: Logistics / e-commerce last-mile delivery SiCepat was not the first thing The Kim Hai tried. He set up the company in 2014 after several earlier business setbacks. What the record does show is the scale of the bet. Indonesian online shopping was taking off, and with it a quiet crisis: parcels. Sellers on marketplaces needed a courier that was quick, cheap and predictable, and the incumbents were stretched. SiCepat, whose name means roughly "the quick one," went after that gap. Delivery is brutal work. Margins are thin, trucks and riders cost money before they earn it, and every missed parcel becomes a customer complaint. A new courier cannot win by being clever. It has to be reliable, again and again, at huge volume. The money followed the volume. In 2019, SiCepat raised US$50 million in a Series A from Barito and Kejora Intervest. In 2021, it closed a US$170 million Series B, with Telkom's venture arm MDI Ventures among the backers. The company has reported moving around 1.2 million shipments a day, and has been building toward electric mobility as part of a broader logistics ecosystem. Its CEO has also argued publicly that foreign tech startups may find Indonesian express delivery harder than they expect, a reminder that this is a market where local networks, relationships and daily execution matter more than glossy technology. The company has leaned on a web of affiliates rather than owning everything. The Kim Hai is the name most consistently linked with the founding, and Rudy Darwin Swigo is named as a co-founder in at least one Indonesian article. The lesson sits in the pattern, not in a single scene. A founder who has already failed a few times knows how much of a business is plain operations. SiCepat chose a boring, essential, capital-intensive problem and kept at it. For a first-time founder, that is a quiet argument for picking the unglamorous job that everyone needs done well. ## Tiket.com (Indonesia) URL: https://startup.talk/indonesia/tiket-com/ Founders: Natali Ardianto, Wenas Agusetiawan, Dimas Surya Yaputra, Mikhael Gaery Undarsa Founded: 2011; Based in: Indonesia (Jakarta); Sector: Online travel / ticketing Investor after investor said no. Natali Ardianto has written about that stretch, when the team behind tiket.com went from meeting to meeting and kept getting the same answer. It was not his first rejection by the market. After graduating in 2008, Ardianto co-founded Urbanesia, an online lifestyle directory in Jakarta, with friends. He left it because he judged its future to be limited. He then launched Golfnesia, billed as Indonesia's first online golf course booking site. Both failed. Two failed ventures can harden a founder or end him. What happened next is the part few people know. Ardianto teamed up with colleagues Wenas Agusetiawan, Dimas Surya Yaputra and Mikhael Gaery Undarsa to rebuild as a new company, tiket.com, founded in August 2011. Some accounts speak of a larger group at the start, and Ardianto has said that after the investor rejections, only he and those three remained. Think about what that costs. A founding group that shrinks as the money refuses to come. Friends who decided it was not worth the risk. And a founder who already carried two failures, with nothing yet to prove it would be different. The turn was not a clever pivot. It was marketing. Ardianto later shifted strategy toward aggressive digital and conventional advertising, putting the brand in front of people who had never booked a ticket online. The result was striking: tiket.com's profit in 2013 was thirteen times the year before. The story did not end at that peak. In June 2017, Blibli, part of the Djarum Group, bought the company. During the pandemic, when travel collapsed, its CMO Gaery Undarsa said the firm chose not to lay off staff, unlike some competitors. The company reports around 1,200 employees today. For a first-time founder, the quiet lesson is that failure is cumulative data, not a verdict. Ardianto's earlier flops taught him what not to build, and the rejections taught him what customers would have to see before investors believed. ## Midtrans (Indonesia) URL: https://startup.talk/indonesia/midtrans/ Founders: Founded: 2012; Based in: Indonesia (Jakarta); Sector: Fintech / payment gateway In 2016 Midtrans put a number on its own progress: 1,000 businesses served, and 100 employees. For a company that had begun four years earlier under another name, Veritrans, it marked the moment a payment pipe became a real business. Payments in Indonesia then were not simple. Many people had no credit card. Online sellers wanted to be paid, but their customers wanted to use bank transfer, a virtual account or cash at a convenience store. The gap between those two wishes was where Veritrans started in 2012, with the idea, in its own words, that payments are the core of every business. The company grew by adding ways to pay. In 2013 it opened an office in Bandung and launched direct debit. In 2014 came virtual accounts, which let each customer get a unique account number to pay into. In 2015 it added over-the-counter payments. Each one removed a reason for a shopper to abandon a purchase. In 2016 it rebranded as Midtrans and expanded to Bali and Yogyakarta. The turn came in December 2017. Gojek bought three fintech companies at once: Kartuku, Midtrans and Mapan. Midtrans was described at the time as the biggest online payment gateway in the country. Each acquired company kept its own CEO, who also took a senior role in the Gojek group. The year ended with Midtrans launching Selly, an admin and logistics assistant app for small sellers. The deal changed what Midtrans was for. It helped power GoPay, which launched in 2018, and QRIS in 2020. Today it sits inside GoTo Financial alongside GoPay and Moka. The people who started it are largely unnamed in the company's own history, and their early lives remain private. What can be said is the shape of the idea. Midtrans did not invent a new way of paying. It listened to how Indonesians actually wanted to pay and made each of those ways work online. For a first-time founder, that is the lesson: adapt to the local habit instead of waiting for the customer to change. The founders who built the plumbing mostly stayed out of view, and the business grew all the same. ## Mamikos (Indonesia) URL: https://startup.talk/indonesia/mamikos/ Founders: Maria Regina Anggit Tut Pinilih, Bayu Syerli Rahmad, Gilbok Lee Founded: 2015; Based in: Indonesia (Yogyakarta); Sector: Proptech: boarding house (kos) marketplace and management Before Mamikos was a company, it was a side project that three coworkers worked on after their day jobs. Maria Regina Anggit Tut Pinilih was employed at an outsourcing software firm when she and two colleagues began building something for a problem she kept seeing in Yogyakarta: friends who could not find a room. Yogyakarta is a student city. Every year thousands arrive with a bag and an address on a scrap of paper, and the search for a kos, a boarding room, was done on foot. You walked the lanes, read the handwritten signs taped to gates, and knocked. Listings were scattered and disorganised. Anggit has described the gap as one that hit students, college students and young workers alike. The team was small. Bayu Syerli Rahmad handled operations and marketing, Gilbok Lee the technology, and Anggit led. They launched in November 2015. The name is a blend that stuck: Mami, as in a mother who looks after her tenants, plus kos. Building the app was the easy part. What almost nobody expected was that the hard customer was not the student but the landlord. Boarding owners in 2015 did not use apps, and many had no reason to trust three young people who said a platform would fill their empty rooms faster. So the team did what apps rarely do. They went to the owners one by one, house by house. Money was thin, which pushed them to improvise instead of buy attention. And Anggit has said she met scepticism as a woman founder in a field dominated by men. It slowly worked. Owners who listed saw rooms fill, and told other owners. In 2018 a small seed of roughly US$749k was reported, and in 2019 Alpha JWC Ventures was among the investors who backed the company to build technology and widen its reach beyond Java's student towns. The next move was bolder. Finding a room was only half of what a boarding owner needed. So Mamikos added tools to manage tenants and payments, and later began operating rooms itself under the Mamirooms brand. By early 2022 the company reported around 150,000 owners and service in about 300 cities, with six to eight million people searching each month. The price of growth in property is that the business gets heavier. A listing site is light; a room you manage has furniture, cleaners and a lease. Competitors with larger funding chased the same tenants. What Anggit's story gives a first-time founder is a quiet lesson. She did not quit and bet everything on day one. She started where she stood, inside a job, with two colleagues and a problem she could see in her own friends' faces, and she earned the trust of the people who held the keys, one door at a time. ## Sociolla (Indonesia) URL: https://startup.talk/indonesia/sociolla/ Founders: Christopher Madiam, Chrisanti Indiana, John Marco Rasjid Founded: 2015; Based in: Indonesia (Jakarta); Sector: Beauty e-commerce, retail and community (Sociolla, SOCO, Beauty Journal) Chrisanti Indiana was 22 when she flew home to Indonesia in 2014 after her studies in Sydney. She did not come back to take a job. She came back with a complaint. The complaint was simple, and she has put it in her own words: she found it very difficult to find original and safe products online. In Indonesia's beauty market, demand was loud and supply was murky. Plenty of sellers offered serums and lipsticks without approval from the food and drug agency BPOM, and nobody could tell the real from the fake until it touched their skin. She joined Christopher Madiam and John Rasjid, who had spotted the same hole. Madiam later said there was high demand for beauty products but not enough authentic supply. Their plan was dull to describe and hard to do: sell only genuine goods, buy directly from brands and authorised distributors, and let customers write reviews. The company started in a small shophouse. Sociolla went online in 2015, and East Ventures wrote the first cheque that year. What nobody tells you is that the first fight was with the brands, not the customers. Labels were wary of the internet because it was full of resellers who cut prices without permission. Indiana had to persuade them that Sociolla was a different kind of seller, one that would promote them. At the same time, shoppers kept drifting to cheaper marketplaces. Winning them back needed more than a catalogue, so the founders built a content arm and a community around reviews. In 2017 they launched SOCO.id and the SOCOBOX subscription, and in 2018 the SOCO Beauty app. Money followed. A US$12m round led by EV Growth arrived in May 2018, and Temasek and L Catterton backed a round of about US$60m in 2022. By 2020 the group estimated about 30 million users across its units. In 2025 it counted 100 stores for its tenth anniversary, and its CEO said 2024 revenue had grown around 50%. Indiana has been frank about being young. She says it gave her energy and a willingness to take risks, and also less experience. She leans on mentors, and warns that founders who do not learn fast get passed by bigger companies. The takeaway is in the order of her own sentence. Trust came first, scale second. A founder who begins with a problem she personally could not solve as a customer has already done her market research. ## Evermos (Indonesia) URL: https://startup.talk/indonesia/evermos/ Founders: Iqbal Muslimin, Ghufron Mustaqim, Arip Tirta, Ilham Taufiq Founded: 2018; Based in: Indonesia (Bandung); Sector: Social commerce for resellers, halal-focused Evermos began with a frustration. Ghufron Mustaqim has said that he and his co-founders were dissatisfied with how retail worked in Indonesia. Goods passed through layer after layer of distributors, and each layer added a margin. Online, the other problem was fake products. Prices went up and trust went down. Evermos was founded in November 2018 by four people: Mustaqim, Iqbal Muslimin, Ilham Taufiq and Arip Tirta. Their answer was social commerce. A reseller picks products from a catalogue and sells them inside her own circle of friends, family and community, mostly through chat apps. She never buys inventory. Evermos holds the stock, packs the order, ships it and handles customer questions. Brands pay a commission of around 30%, and most of it is shared with the reseller. What almost nobody saw coming was where it would work. Evermos did not start in Jakarta's malls. By 2021 about 100,000 active resellers were spread across more than 500 Tier 2 and Tier 3 cities, mostly in Java. Most of them were selling online for the first time, so the company ran training in the app and sometimes in person. The best resellers earned about US$200 a month, which in a smaller city is a real second income. The product mix was shaped by the audience. With nearly 90 per cent of Indonesians Muslim, Mustaqim argued, building for the Muslim market meant building for almost everyone. Fashion, halal health and beauty items, and food and drink from small Indonesian brands filled the catalogue, more than 500 brands by 2021. Investors saw it too. A Series A of about US$8.3m reportedly came in December 2019 with Jungle Ventures. In September 2021, a US$30m Series B led by UOB Venture Management's Asia Impact Investment Fund II followed, and a Series C of roughly US$39-40m was reported in 2023. Transaction volume rose more than 60 times in the two years before the Series B. The cost was in the unglamorous parts. Shipping out of small cities is expensive. Inventory has to be close to the buyer. The team spent its new money on warehouses closer to resellers, analytics and the recommendation tools that help a reseller choose what to push. The lesson sits in the friction. Evermos did not invent selling to friends. It removed the three things that stopped ordinary people doing it: stock, shipping and trust. If you want a market that has never bought from you, find the step they dread and take it off their hands. ## Sayurbox (Indonesia) URL: https://startup.talk/indonesia/sayurbox/ Founders: Amanda Susanti Cole, Metha Trisnawati, Rama Notowidigdo Founded: 2016; Based in: Indonesia (Jakarta); Sector: Agritech: online grocery with farm-direct supply chain Amanda Susanti Cole has a small, useful confession. Before Sayurbox she started several businesses, from school years through university, and none of them took off. She told an online discussion in April 2020 that she has always loved building new things and solving problems. The side projects were part-time and did not become anything. She treats that as part of the process: success, she has said, is a process. After university she began gardening, in her words just for fun. It did not stay a hobby. Working with growing things put her in the middle of a puzzle: farmers produced food, cities wanted it, and in between sat a long chain of middlemen. Most farmers struggled to reach buyers, and many earned little from what they grew. The company that came from this was Sayurbox, an online service that delivers vegetables and fruit from local farmers. It was the first business she ran full time. She was not alone. Metha Trisnawati and Rama Notowidigdo are listed as co-founders. And the first cheque came from a woman, something Amanda herself points to when she talks about women backing women. What almost nobody notices from outside is how hard the product is. Vegetables rot. Orders must arrive in a day, not a week. A grocery startup lives or dies by logistics. She has also said she works in industries dominated by men, and feels that women must work harder. The scale came fast. Forbes put her on its 30 Under 30 list in 2019. By 2020 the company was reported to work with about 10,000 farmers and about 80 staff. In 2021, a Series B of around US$15m led by Astra arrived, and in March 2022 a US$120m Series C led by Northstar and Alpha JWC Ventures, with IFC taking part. Then the bill. Pandemic demand faded, and in April 2023 layoffs were reported, with Amanda pointing to weaker online consumer growth after the pandemic. Her path leaves one practical note for a first-time founder. Several quiet failures were not wasted. They were practice at finding a problem worth full-time attention. The one that stuck started with dirt on her hands, not a slide deck. ## Kitabisa (Indonesia) URL: https://startup.talk/indonesia/kitabisa/ Founders: Muhammad Alfatih Timur, Vikra Ijas Founded: 2013; Based in: Indonesia (Jakarta); Sector: Donation-based crowdfunding (fundraising, zakat) platform At fifteen, Alfatih Timur sat in a classroom of students two years older than him. He had been placed in an accelerated programme, and the older kids had a word for him: cupu, roughly dorky. He kept his head down and kept studying. Born in Bukittinggi in December 1991, he went on to study economics at Universitas Indonesia. There he was a busy student: head of student affairs in the faculty council and in the university-wide student body, a regular in discussions and rallies. One lecturer changed his direction, Rhenald Kasali, who introduced him to social entrepreneurship. After graduating in 2013 he worked at Kasali's consultancy, Rumah Perubahan. It was there that he began to meet people at non-profits. They shared a problem. Many generous people wanted to give, but the people doing good work found it hard to reach them. Timur read about crowdfunding sites abroad and began building one. With Vikra Ijas, a university friend, he started small. One early campaign raised Rp160 million, about US$10,700, to stop a street children's school in Depok being evicted. It worked. Then came the stall. Timur took the site to campuses and the response was lukewarm. Crowdfunding was not widely understood in Indonesia, and asking strangers to hand money to strangers online sounds risky to many. What almost nobody knew was how he paid for the next step. He used what was left of his wedding money to improve the platform, according to a 2022 profile. Trust was the real product. Kitabisa built a verification routine he calls three plus one, including identity checks and document review, and let the public report suspicious campaigns. In 2015 500 Startups led a seed round, and later investors reportedly included IFC, Northstar and Endeavor Catalyst. The platform took a 5% fee on most campaigns and none on zakat or national disasters. When COVID-19 hit, donations jumped: 3 million individual gifts in March and April 2020, and the company said it channelled more than US$10 million. By January 2022 it was reported to have about 937,000 users and Rp376.95 billion collected. The cost of being a trust business is that a single bad campaign can unravel years of work. Timur's answer has been process, not slogans. His story hints at something a first-time founder can borrow. The boy who was the youngest in the room at fifteen learned to keep working while others talked, and years later he spent his wedding savings on the same habit. ## Zenius (Indonesia) URL: https://startup.talk/indonesia/zenius/ Founders: Sabda PS (Sabda Putra Subekti), Wisnu Subekti, Medy Suharta Founded: 2004 (legal entity 2007); Based in: Indonesia (Jakarta); Sector: Edtech: online video lessons and exam preparation In January 2024 Zenius sent students a short message. It apologised for the inconvenience and said it was temporarily halting operations, citing operational challenges. There was no date for a return. For a company that had spent twenty years telling teenagers that understanding beats memorising, it was a quiet ending. What many users did not know was how small it had started. Zenius did not begin as an app. In 2004 it was an offline exam-preparation operation, run by Sabda PS with Wisnu Subekti and Medy Suharta, coaching school leavers for the brutal state-university entrance tests. It only became a formal company in 2007. The first classroom came long before the first server. The founders' bet was that concepts, not tricks, would win. They filmed lessons and put them online while most tutoring in Indonesia still meant sitting in a rented hall. Over time the videos reached what the company later described as millions of students, and by December 2022 the platform reported more than 16 million users. Then the pandemic closed schools, and edtech became the fashionable place to put money. Investors including Northstar Group, Alpha JWC, Openspace Ventures and Beacon Venture Capital came in. In March 2022 a round led by Telkom's MDI Ventures took total funding to about US$40m, or Rp622bn. That same year Zenius bought the tutoring chain Primagama, tying its online lessons to a brand with physical branches and decades of recognition. By then the weather had turned. Edtech funding froze as pandemic habits reversed and students went back to classrooms. In May 2022 more than 200 employees were let go. A second round followed in August 2022 and a third in February 2023. Reports in 2023 also described a larger cut, though that figure comes from a single outlet and is not used here. The company that had grown by raising money now had to survive without it. In January 2024 the announcement came. Sabda said in a statement that the team had taken strategic steps to stop operations temporarily, but would not stop trying to realise the vision. The hard lesson is that a loyal audience is not the same as a business that pays for itself. Zenius had both a method and a following, and still depended on a funding climate it could not control. Twenty years of teaching, it turns out, did not guarantee a twenty-first. ## Alodokter (Indonesia) URL: https://startup.talk/indonesia/alodokter/ Founders: Nathanael Faibis, Suci Arumsari Founded: 2014; Based in: Indonesia (Jakarta); Sector: Healthtech: online doctor consultations, health content and appointment booking Most health apps begin with a doctor. Alodokter, in the way it tells its own story, began with a question: where does an ordinary Indonesian go to ask whether a symptom is serious? Nathanael Faibis and Suci Arumsari founded the company in 2014, with Faibis as chief executive and Arumsari as director. The starting problem was access. Indonesia is a country of islands, and clinics are clustered in big cities. A person with a worrying rash or a feverish child might wait hours at a clinic, or go without advice. Alodokter set out to put health information and doctors on a phone: articles written by medical staff, a place to ask questions, and later the ability to book appointments and consult a physician online. The early money was modest. A seed round in 2015 came from Fenox Venture Capital. In 2017 SoftBank Ventures Korea led a US$9m Series B with Golden Gate Ventures and Fenghe, a sign that regional investors thought Indonesian healthcare was ready for software. The larger step came in 2019, when a US$33m Series C led by insurer Sequis Life brought in Philips, Heritas Capital, Hera Capital and Dayli Partners. Having an insurer and a medical-device giant at the table said something about where the company wanted to go: toward becoming a platform connecting patients, doctors, hospitals and payers rather than a content site. In June 2021 MDI Ventures, the corporate arm of Telkom Indonesia, and Samsung Ventures added fresh capital. The pandemic put telemedicine into the national conversation, and also invited well-funded rivals such as Halodoc and Good Doctor. Alodokter had to defend a position it had built through content over years. There is a founder lesson here that is easy to miss. Faibis and Arumsari did not start with the glamorous part of healthcare, the hospital or the lab. They started with the unglamorous question millions of people were already typing into a search bar, and built outward from the habit that already existed. ## Fore Coffee (Indonesia) URL: https://startup.talk/indonesia/fore-coffee/ Founders: Robin Boe, Jhoni Kusno, Willson Cuaca, Vico Lomar Founded: 2018; Based in: Indonesia (Jakarta); Sector: Food and beverage: app-first coffee chain On 8 April 2025 Jakarta's stock index fell almost 8 percent in a single morning. Markets had reopened after the Eid holiday into the shock of American tariffs. Six days later a coffee chain was due to list. People inside Fore Coffee have said many doubted the IPO would work. The company went ahead anyway. The chain had begun upstairs. Robin Boe started Otten Coffee in 2012, selling beans and machines, with the first store in Medan, his hometown. East Ventures' Willson Cuaca funded it. Fore grew out of a business incubator on the second floor of Otten's Jakarta office. It opened in August 2018 as Otten's second store on Jalan Senopati, in South Jakarta. The idea was to treat the phone as the front door. Orders came through the app, with delivery and loyalty built in, rather than a queue at a counter. In February 2019 Fore raised US$8.5m from East Ventures, SMDV, Pavilion Capital, Agaeti Venture Capital, Insignia Ventures Partners and angels. Vico Lomar joined in 2020 as chief executive, the same year the pandemic emptied cafes. What nobody could have predicted was how that first foothold in delivery would look in hindsight. A chain built around the app survived lockdown better than one built around seating. It then grew outward. In November 2023 it opened a store at Bugis Junction in Singapore. By the end of 2024 reports counted 217 outlets. The IPO tested the numbers. On 14 April 2025 Fore listed on the Indonesia Stock Exchange. The offering drew 114,873 retail investors and was oversubscribed 200.63 times. Shares rose 34.04 percent to Rp252 on day one. In the first half of 2025 revenue grew 47 percent to Rp662bn, and the chain reached 261 outlets. A Katadata report puts it at 363 active outlets by the end of June 2026. The price of that scale is competition. Indonesian coffee is crowded with low-priced rivals and venture-backed chains, and every new outlet must earn its place. Fore's story has a lesson for founders about where good ideas come from. It was not a grand launch. It grew in a spare room above an existing business, with a founder's first store in his hometown behind it. Start small, upstairs, and see what the customer actually opens on their phone. ## Warung Pintar (Indonesia) URL: https://startup.talk/indonesia/warung-pintar/ Founders: Agung Bezharie Hadinegoro, Harya Putra Founded: 2017; Based in: Indonesia (Jakarta); Sector: Micro-retail tech: smart kiosk and supply chain for neighbourhood warungs Agung Bezharie Hadinegoro spent time as an investor before he became a shopkeeper's partner. He was an investment associate at East Ventures, the Jakarta firm that backs early-stage startups. At some point the person looking at business plans decided to build one himself. The company was announced in February 2018 as a project born inside East Ventures, with a US$4m seed round. SMDV, Digital Garage, East Ventures, Insignia Ventures Partners and Triputra Group put in money, along with unnamed angel investors. Harya Putra was named chief operating officer in the press statement. The problem it chose was everywhere. Indonesia has millions of warungs, the corner shops that sell cigarettes, instant noodles and phone credit. Owners buy stock from wholesalers on their own schedule, keep paper records and have little access to digital payments. Warung Pintar aimed to turn them into smart outlets: technology to run the shop, access to supplies and financial products in one place. The bet worked well enough to raise more money fast. A follow-on of roughly Rp57bn was reported in mid-2018. In January 2019 came a Series B of US$27.5m, with existing investors returning and the payments app OVO joining as a new backer. By August 2021 Bezharie was telling DealStreetAsia that the company was in talks about a Series C. Why does that matter? Because a startup that sells to shopkeepers lives or dies on their daily routine, and the founders had to earn a place in it. The tension was always the same. Warung owners could not be forced to adopt anything. Every smartphone, every fridge and every supplier relationship had to prove itself in a business that runs on cash and daily habit. Meanwhile a wave of better-funded B2B commerce startups was chasing the same shops. The company that began as an East Ventures project kept those ties through its funding rounds, with East Ventures among the backers named in the seed. There is a founder lesson in the order of events. Bezharie had seen hundreds of pitches from the investor's side of the table, then picked a market so ordinary that most people had stopped noticing it. The opportunity was not in a new behaviour but in an old one that millions repeated every morning. ## Payfazz (Indonesia) URL: https://startup.talk/indonesia/payfazz/ Founders: Hendra Kwik, Ricky Winata, Jefriyanto Founded: 2016; Based in: Indonesia (Jakarta); Sector: Fintech: agent-based payments, e-money and lending for the unbanked Hendra Kwik grew up in Jambi, a city on Sumatra far from the glass towers of Jakarta, and he has been blunt about it. "I come from a village," he told KrASIA in 2020. He meant it as an explanation, not an apology. He studied chemical engineering at the Bandung Institute of Technology, one of Indonesia's best-known universities. After graduating he took a job in Brazil with a global oil and energy company, a long way from both Jambi and Bandung. Later he visited Silicon Valley and MIT, and toured startup ecosystems in Latin America. When he came back to Indonesia he chose to join Kudo, a startup whose founder was an ITB senior, over other offers. At Kudo he handled business development and growth strategy, and that meant travelling to towns and villages across Java. What he heard there sounded familiar. Villagers wanted to know how to get loans for their small businesses and how to send money to their families. Their questions matched the ones he remembered from Jambi. In 2016 he launched Payfazz with Ricky Winata and Jefriyanto. The model was simple. Payfazz would recruit agents, often small shopkeepers, who could sell mobile credit, take bill payments and handle transfers for neighbours with no bank account. The company earned a small commission, and so did the agents. The obstacles were not glamorous. Financial literacy in rural areas was low. Fraudsters posed as Payfazz agents. When COVID-19 hit, the average transaction size dipped slightly. Even so, by August 2020 the company reported about 250,000 agents serving around 10 million users, handling more than Rp1tn a month in transactions. In early June 2020 it had closed a US$53m Series B led by B Capital Group and Insignia Ventures, bringing total funding to US$74.1m across four rounds. It also secured an e-money licence from Bank Indonesia. In 2022 it merged with Singapore's Xfers to form Fazz Financial Group, with Kwik as group chief executive. The founder's insight was to treat his own origin as a data point. Many founders build for users they imagine. Kwik built for users he remembered, and the questions they kept asking him were the business plan. ## Investree (Indonesia) URL: https://startup.talk/indonesia/investree/ Founders: Adrian Gunadi, Dr. Amiruddin, KC Lim Founded: 2015; Based in: Indonesia (Jakarta); Sector: Fintech, peer-to-peer lending (SME) The paperwork is dated 21 October 2024. On that day Indonesia's financial regulator, OJK, revoked the business licence of PT Investree Radhika Jaya, once one of the best-funded peer-to-peer lenders in the country. The stated reason was dry: the company had not met the Rp7.5 billion minimum equity requirement. It is a strange ending for a company that started with a promise of the opposite. In January 2017, Investree told the market it had disbursed Rp53.7 billion in loans with a 0 percent default rate. A few months later, on 31 May 2017, it was registered with OJK. Tech in Asia named it a favourite startup in 2016, and in 2017 it won an award for best P2P lending for SMEs. The founding team was Adrian Gunadi, Dr. Amiruddin and KC Lim, with early executives Andi Muhamad Andries on risk and operations and Dickie Widjaja as chief information officer. Public records say little about their childhoods or schools. What is documented is that they came from established national and international financial institutions, and that Gunadi became the face of the company. Investor magazine named him a Tokoh Finansial Indonesia in 2018. The money followed the story. Kejora led the early backing in 2016. In July 2018 a Series B led by SBI Holdings brought in six investors, among them Mandiri Capital Indonesia, Persada Capital, Endeavor Catalyst and 9F Fintech. Investree signed banks such as Danamon, Woori Saudara, Amar Bank and Bank Sumut as funding partners, and launched a sharia unit with a recommendation from the national sharia council. What the public record shows next is a sequence of official steps. OJK revoked the licence in October 2024. It ran a fit-and-proper reassessment of Gunadi, who failed, froze his bank accounts and began tracing assets. He was named a suspect for unlicensed fundraising under Article 46 of the Banking Law. A Red Notice request was filed in February 2025. Gunadi was later reported as CEO of a Qatar joint venture and then as repatriated to Indonesia and held by the police pending proceedings. Legal outcomes have not been reported here, and none are assumed. In March 2025 the shareholders dissolved the company by notarial deed, and OJK approved a three-person liquidation team. For a founder the lesson is narrow and practical. A licence is not a badge you win once. It is a standing condition, tied to capital you must keep in the bank, and it can be withdrawn on a single date. ## KoinWorks (Indonesia) URL: https://startup.talk/indonesia/koinworks/ Founders: Benedicto Haryono, Willy Arifin Founded: 2016; Based in: Indonesia (Jakarta); Sector: Fintech, SME neobank and lending In January 2022, KoinWorks announced the largest cheque in its history: US$108 million in a Series C round, led by MDI Ventures, the venture arm of Telkom Indonesia. The pitch was a bigger one than lending. It wanted to become a neobank for small and medium businesses. It had begun six years earlier, in 2016, when Benedicto Haryono and Willy Arifin founded the company. Haryono became CEO. Arifin took the role of executive chairman. That trail is long. In 2018 the company raised a Series A reported at Rp230 billion, led by Mandiri Capital Indonesia, with Gunung Sewu and Convergence Ventures joining. Quona Capital followed with an A+ extension. In 2019 came a Series B, reported at Rp170 billion, and a B2 of S$18.5 million from Saison Capital, the venture arm of Credit Saison. .. What the sequence shows is a company that kept adding new categories of investor: a state-bank venture arm, a Japanese consumer-credit group, a telecom giant. Each one validated a different part of the business, from marketplace lending to supply-chain finance and eventually banking-style services. For a founder, the takeaway is in the structure of the cap table. Raising from banks and telcos buys distribution and credibility, and it also ties your name to institutions whose own risks can become yours. Know who your investors are, because the public will. ## Qoala (Indonesia) URL: https://startup.talk/indonesia/qoala/ Founders: Harshet Lunani, Tommy Martin Founded: 2018; Based in: Indonesia (Jakarta); Sector: Insurtech On the day Qoala announced its US$65 million Series B in May 2022, the headline number was less interesting than the sentence beside it: the company said it had grown thirty times since its Series A, which closed in April 2020. That is two years, a pandemic, and a business that sold insurance, a product people usually avoid. Qoala was launched in 2018 by Harshet Lunani, who became CEO, and Tommy Martin, who became COO and later deputy CEO. What is documented is the model: car, bike, home and health insurance sold through partners and an omnichannel platform, not through a mountain of agents. The Series B was led by Eurazeo, with Flourish Ventures, KB Investment, MassMutual Ventures, MDI Ventures, SeedPlus and Sequoia Capital India all taking part. Then came a quieter move. In February 2021 the company bought FairDee in Thailand. Martin has said the acquisition tripled Qoala's business in that country after he joined. It was an early sign that the company meant to grow by buying distribution, not only by building it. In March 2023 Qoala completed a Series B+ of US$7.5 million, led by responsAbility, and said that gross written premiums had doubled since the Series B. The extension was small by comparison with the main round, and arrived in a year when many Asian startups could not raise at all. The fund-raising pattern is the point. Qoala took money from reinsurance-adjacent and insurer-backed investors, which is exactly who a young insurer needs on its side. For a founder in a regulated field, the lesson is that your investors are also your credibility with the regulator and the partners who will sell your product. What remains unknown is the human story: the moment Lunani and Martin first decided to build it. When that surfaces in an interview, it will be worth more than any funding figure. ## JULO (Indonesia) URL: https://startup.talk/indonesia/julo/ Founders: Adrianus Hitijahubessy, Victor Darmadi, Hans Sebastian Founded: ; Based in: Indonesia (Jakarta); Sector: Fintech, digital lending and credit scoring In July 2024, JULO told the press it had reached US$120 million in annual recurring revenue, riding a 50 percent growth in its loan portfolio. For a lender whose customers often had no bank file at all, it was a loud number. The company was started by three co-founders: Adrianus Hitijahubessy, who became CEO, Victor Darmadi and Hans Sebastian. Digital News Asia reported that Darmadi and Sebastian met while they were in Silicon Valley. The idea is easier to trace. JULO set out to bring AI-based credit scoring to the underbanked, people who have phones but not credit histories. Early backing came from Skystar Capital, which led the seed round, and East Ventures. In 2022 Credit Saison, the Japanese consumer-finance group, put in US$80 million in a Series B. The product has widened since. JULO raised its digital credit limit to Rp15 million, and launched an education-financing offer. In July 2026 it secured another US$5 million from existing investors, which tells us that its backers are still staying with it, though not at the scale of 2022. The lesson for a founder in credit is in the investor list. A strategic lender like Credit Saison brings not just money but underwriting experience and cheap funding lines. In lending, who finances your loan book matters as much as who finances your payroll. ## Dicoding (Indonesia) URL: https://startup.talk/indonesia/dicoding/ Founders: Narenda Wicaksono, Kevin Kurniawan, Akbar Hidayat Founded: 2015 (launched 5 Jan 2015; one thesis says 2014); Based in: Indonesia (Bandung); Sector: Edtech, developer community and training Narenda Wicaksono had a decision on his desk that most people would call a gift. His employer, a multinational, was offering him a Southeast Asia developer-relations role that, by his account, could have doubled or tripled his income. He had spent more than eight years in developer relations at Nokia and Microsoft. He also had a wife and two children. He said no, and went the other way. In a March 2015 interview with DailySocial, he admitted what frightened him: whether he could keep his children in school if the salary stopped. He settled it, he said, by deciding that only God can guarantee a person's life, so his job was to give his best. What pulled him was the developers he had met. Through his work he had mentored everyone from students to experienced engineers and seen talent in cities far from Jakarta. Distance from the capital and the language barrier held many of them back. "I couldn't leave my fellow Indonesian developers," he told the interviewer. Dicoding launched in Bandung on 5 January 2015 with two co-founders: Kevin Kurniawan, a former Nokia developer-experience colleague, and Akbar Hidayat, a freelance software developer. About 500 developers joined soon after. DailySocial, which published his interview, disclosed that it was an investor. The early years were built on events and challenges. In 2016 the government's creative-economy agency made it an event partner, and it became a Google Developers Authorized Training Partner. In 2017 learners could pay through Bukalapak and Tokopedia. In 2018 Google and Dicoding ran 1,000 Picodiploma scholarships. Between 2015 and 2018 it held 630 offline events. The people stories are the proof. Soesapto Joeni Hantoro, a Yogyakarta graduate, won competitions through Dicoding challenges and in 2018 was invited to LINE Developer Day in Tokyo. Junia Firdaus, a motorcycle-taxi driver, won a scholarship and finished an Android Developer Expert class in 86 days. By early 2019 about 140,000 developers from 454 cities had joined. That was the dream he had stated in 2015: "turn Indonesian developers into world-class developers." The quiet lesson is in the pay cut he accepted. He did not leave a job to chase an idea. He left to serve people he already knew, which is a much sturdier reason. ## Qiscus (Indonesia) URL: https://startup.talk/indonesia/qiscus/ Founders: Delta Purna Widyangga, Evan Purnama, Muhammad Md Rahim Founded: 2013; Based in: Indonesia (Yogyakarta/Jakarta); Sector: Conversational / omnichannel customer engagement (chat SDK and multichannel chat) In 2015 a co-founder walked out and Delta Purna Widyangga, then co-CTO, was handed the chief executive job of a two-year-old chat startup. He had not planned it that way. He had planned to run the enterprise side while Evan Purnama built the product. Delta's path to that seat was unusual for an Indonesian founder of that era. He studied electrical and electronic engineering at Nanyang Technological University in Singapore, graduating in 2009, and then took the safe route: software engineer at Accenture, then consultant at the SAP specialist ecenta AG. Neither job looked like a founder's launch pad. What almost nobody mentions is that he was already a founder by the time Qiscus began. In 2011 he co-founded a digital agency in Singapore called 1.618, and he stayed with it until 2017, which means that for several years he was running two companies at once, an agency that paid and a chat startup that did not yet. The agency was a quiet cushion under a risky bet. Qiscus began in 2013 with Evan Purnama, Muhammad Md Rahim and Amin Nordin. Money was thin. A seed of about US$100,000 arrived at the end of 2014, small even by Southeast Asian standards. Then Nordin left, and the product itself had to prove it was a business, not a feature. The team moved toward selling a chat SDK and later a multichannel customer inbox to companies, a far more patient and less glamorous market than consumer messaging. The fundraising story stayed modest. Reports said Qiscus was preparing a Series A, a Malaysian investor backed an earlier pre-Series A round, and later Init6 put in a strategic investment reported at US$2 million. A confirmed Series A never surfaced in what could be found. Instead the company said it had been profitable since 2019, and Delta said there were no plans for an urgent raise, though he would not rule one out. That is an unfashionable ending in a region that celebrates big rounds. It is also the point. Delta ran a profitable agency while the chat product searched for its market, and then he held the line on spending until Qiscus was earning its keep. For a first-time founder the lesson is practical rather than inspiring: a side income, a co-founder who stays, and a product that pays for itself can be worth more than a headline round. ## Kata.ai (Indonesia) URL: https://startup.talk/indonesia/kata-ai/ Founders: Irzan Raditya, Reynir Fauzan, Wahyu Wrehasnaya, Chris Franke Founded: 2015 (as YesBoss; rebranded Kata.ai); Based in: Indonesia (Jakarta); Sector: Conversational AI / chatbot platform By his own count, Irzan Raditya had failed at four startups before he built the one that lasted. He told KrASIA he treats each failure as a lesson: either you win or you learn. The earlier attempts started around 2012, when he was living in Germany and building companies as a software engineer. He came back to Indonesia and in 2015 launched YesBoss, a virtual assistant you could text, a concierge in your pocket. It was a charming idea and, as it turned out, a hard business. What the YesBoss story hides is the cost of the next step. A consumer assistant needs humans behind the screen, and the numbers did not hold. Irzan decided to leave consumers behind and sell to businesses. Pivoting meant changing the product and the people: some employees whose jobs no longer fit had to be laid off. He has described that period as part of learning to put people first, product second, then profit. The company he built with Wahyu Wrehasnaya and Reynir Fauzan was reborn as Kata.ai, a platform for building chatbots that understand Indonesian. Customers were large organisations, banks, telecoms, retailers, that needed to answer millions of messages. In 2017 it raised US$3.5 million in a Series A led by Trans-Pacific Technology Fund, with Red Sails Investment and the angel Eddy Chan taking part. The next years were quieter. Language models were still hard in Bahasa Indonesia, and AI talent was scarce. A Series B arrived in November 2020, led by TPTF and MDI Ventures with Buana Investama, though the amount was never disclosed. Kata.ai used it to push into commerce, healthcare and insurance and to launch a tool for small businesses to automate chat sales. Irzan has called himself a person in permanent beta, a phrase that sounds casual and describes the whole arc: five attempts, one pivot, layoffs, and a product still evolving. For a first-time founder the takeaway is plain. The company that lasted was not the first idea. It was the fifth, and it only appeared after he was willing to shrink the thing he loved. ## Nodeflux (Indonesia) URL: https://startup.talk/indonesia/nodeflux/ Founders: Meidy Fitranto, Faris Rahman Founded: 2016; Based in: Indonesia (Jakarta); Sector: Computer vision / vision AI (VisionAIre) Two friends who had met in junior high school ended up in the same lecture halls at Institut Teknologi Bandung, both studying industrial engineering. That small coincidence is where Nodeflux starts. Meidy Fitranto and Faris Rahman reconnected at ITB, bound by a shared love of IT. In 2016 they started a company, and its first idea was not the one that made it known. Nodeflux began as a big-data analytics platform. By late 2016 they had noticed something. Demand for computer vision was high and almost no Indonesian company was doing it. They pivoted, and the company became a vision AI business. The early work was unglamorous. Rahman, the CTO, later admitted that building the first systems took heavy automation and repeated modification, and that it was difficult at first. Fitranto pointed to the larger problem: top AI engineers in Indonesia are mostly from abroad or educated overseas, which he partly blamed on university curricula. Still, customers came. Police departments were early users. The platform, VisionAIre, read license plates, spotted illegal parking and counted vehicles from camera feeds, and Jakarta's government used it. In 2017 Qlue, a rival of sorts, put in seed money and became a shareholder, along with Adskom and Prasetia Dwidharma. East Ventures followed in 2018. Big events built credibility: the 2018 Asian Games and the IMF-World Bank annual meeting. By the time KrASIA interviewed them, Nodeflux connected to more than 3,000 cameras, ranked in the top 25 of the NIST face recognition test out of about 90 vendors, and was the only Indonesian company in NVIDIA's Metropolis partner program. Fitranto said the team was not rushing to close a round because cash flow was positive. The founder lesson is about timing and friendship. Two people who already trusted each other found a gap that nobody local was filling, and moved into it within months. ## Bridestory (Indonesia) URL: https://startup.talk/indonesia/bridestory/ Founders: Kevin Mintaraga, Doni Hanafi Founded: 2014; Based in: Indonesia (Jakarta); Sector: Wedding marketplace / online vendor directory Kevin Mintaraga set out to plan his own wedding and found it a hassle. That small irritation became Bridestory. He started the company in 2014 with Doni Hanafi, who later became chief operating officer. The idea was a place where brides could browse photographers, venues and dress makers and contact them directly. The seed money came from Sovereign's Capital, with Beenos Plaza, East Ventures and Fenox VC taking part, a respectable list for a wedding directory. What many people do not know is how quickly a big name appeared. In March 2015, within a year of launch, Rocket Internet led a Series A. The amount was not disclosed, though Mintaraga described it as a middle one-digit million dollar sum. Money from a Berlin-based builder of copies and scaled marketplaces pushed the company to expand, including a joint venture with MediaCorp in Singapore. Not everything from that period is clear. DealStreetAsia estimated a valuation of up to US$50 million. In 2018 the team launched Parentstory, a marketplace for children's activities aimed at young parents, with the same playbook in a new life stage. Late that year Mintaraga began talking with William Tanuwijaya, Tokopedia's co-founder and CEO, about working together. Those conversations led to an acquisition in June 2019. Assets, intellectual property and people moved over, while Bridestory and Parentstory kept running as independent brands. Mintaraga joined Tokopedia's management as a vice president. The takeaway for a first-time founder is that a personal frustration is enough of a starting point, if you meet investors early and keep your brand alive after you sell. ## Female Daily (Indonesia) URL: https://startup.talk/indonesia/female-daily/ Founders: Hanifa Ginting Ambadar Founded: 2009; Based in: Indonesia (Jakarta); Sector: Beauty media, community and e-commerce By late 2014 Female Daily Network said it drew several million visits a month and expected to pass 100 million pageviews that year. For a beauty site that started five years earlier, those numbers were hard to ignore. The network grew out of a community of women trading opinions on skincare, make-up and hair products in Indonesia. Reviews written by users, not by editors, became its main asset. Brands wanted access to that audience, and readers wanted honesty. In November 2014, e27 and DailySocial reported that Female Daily Network raised US$1 million in a Series A round led by Ideosource. The money was meant to widen its reach as the audience scaled. Online beauty then became a crowded field. Sociolla, a separate company, appeared in 2015 as a beauty e-commerce player and the term beauty tech in Indonesia is now closely tied to it. What a founder can take from this record is simple. Hanifa built a following first and raised money later, when the traffic was already there. A first round of US$1 million, about five years into the life of the company, is a reminder that community can precede capital. ## Moka (Indonesia) URL: https://startup.talk/indonesia/moka/ Founders: Haryanto Tanjo, Grady Laksmono Founded: 2014; Based in: Indonesia (Jakarta); Sector: Fintech / point-of-sale software for small businesses The company that would one day sell for roughly US$120 million was named after a drink. In 2014, two friends sat over coffee in Jakarta and talked about small businesses that still ran on paper notebooks. Both loved mocha. They called the idea Moka. Haryanto Tanjo was born in Medan in 1989. He left Indonesia for the University of California, Berkeley at 18 and studied industrial engineering and operations research. The jobs that followed read like a young consultant's checklist: a manufacturing engineer at Bayer, project work at Cisco, a spell at a small San Francisco consultancy, an MBA at UCLA, then McKinsey in Jakarta. He knew how to find inefficiency. He had spent years being paid to do it for other people. His friend Grady Laksmono, whom he knew from their college days in the United States, was the engineer. What they saw in Indonesia was a gap. Tens of millions of small merchants had no bank relationship worth speaking of and kept their accounts by hand. A point-of-sale app on a tablet could give a warung or cafe an instant record of what sold. Tanjo walked away from his job in 2014 and registered the company in August. Then came the part nobody puts in a pitch deck. Selling it was slow. Tanjo has said the biggest challenge was market education: merchants did not doubt they were busy, they doubted that software would help them. East Ventures wrote an early seed cheque, and the team spent its days in shops, showing owners the thing instead of describing it. What made it work was focus. Moka did not try to be a bank first. It became the register, and once the register worked, payments, loans and reports could sit on top of it. By 2018 Sequoia India and Southeast Asia led a US$24 million Series B, with SoftBank Ventures Korea and Singapore's EDBI alongside, taking total funding to about US$28 million. A year later Tanjo was on the Forbes 30 Under 30 Asia list. The ending was a handoff. Gojek, hunting for the merchants that fed its ecosystem, agreed to buy the company, and the deal closed in 2020 at a reported value near Rp2 trillion, a figure Gojek never confirmed. Moka's tools were folded into the GoBiz merchant app, and Tanjo moved into a commerce role at GoTo Financial. The lesson for a first-time founder is how small the first idea was. Two friends, one coffee, one observation about paper ledgers. The work was not the idea. It was months of standing in shops, explaining what a tablet could do, until enough owners believed it that a giant needed them. ## TaniHub (Indonesia) URL: https://startup.talk/indonesia/tanihub/ Founders: Ivan Arie Sustiawan, Pamitra Wineka Founded: 2016; Based in: Indonesia (Jakarta); Sector: Agritech / farm-to-market e-commerce and lending In November 2021, about 130 people who had lent a combined Rp14 billion through a farm-lending app opened their accounts and found no repayment. The app was TaniFund, a lending arm of TaniHub. The company said harvests had failed. It was a hard sentence to read for a business built on a hopeful idea. TaniHub began in 2016 as a way to link Indonesian farmers to buyers who wanted fresh produce, and for a few years it looked like proof that agriculture could be a tech business. Its co-founders included Ivan Arie Sustiawan and Pamitra Wineka. Behind them came the money. In 2019 state-linked venture arms MDI Ventures, from Telkom, and BRI Ventures made a joint US$25 million investment. In 2021 MDI Ventures led a US$65.5 million Series B, with BRI Ventures and Telkomsel Mitra Investasi taking part. Those were large numbers for farm-to-table in Indonesia, and the pitch was ambitious: buy from farmers, sell to restaurants and homes, and finance the farmers in between. The model was heavy. Cold storage, warehouses, logistics and credit all had to work at once, and food margins are thin. After the TaniFund problems, layoffs followed. TaniHub closed warehouses in Bandung and Bali, and stopped its household B2C business to concentrate on supplying businesses. Then the story moved from business trouble to a courtroom. In July 2025 South Jakarta prosecutors detained Sustiawan, former director Edison Tobing and the head of MDI Ventures, on allegations that investor money from the 2019 US$25 million round had been misused. In June 2026 a court sentenced executives from both venture arms in the case, and reporting says Sustiawan was found guilty of corruption and money laundering and given nine years. Wineka has not been implicated in the case in the reporting we have seen. What the case leaves behind is a warning that reaches past one company. When a startup is funded by state-owned money, every cheque has an audience that is not the founders. Rapid scale in a low-margin, weather-dependent business meant that a single bad harvest season could turn into a lending default, and the people hurt first were small lenders who had trusted an app. For a founder, the quiet lesson is about order. Credit, logistics and trading each deserve to be proven on their own before they are tied together, and investor funds deserve the discipline of a ledger that can survive being read aloud in court. ## Pahamify (Indonesia) URL: https://startup.talk/indonesia/pahamify/ Founders: Syarif Rousyan Fikri, Mohammad Ikhsan, Edria Albert Founded: 2018; Based in: Indonesia (Bandung); Sector: Edtech / online learning for high-school students At 15, most students are still deciding what to study. Syarif Rousyan Fikri had already started at the Bandung Institute of Technology, in electrical engineering. He was the kind of student who finishes early and then looks for the next thing to do. The next thing was a PhD at Nanyang Technological University in Singapore, where he reconnected with Mohammad Ikhsan. In 2016 the two started a side project that had little to do with their theses: a YouTube channel called Hujan Tanda Tanya, built around popular science. It won YouTube NextUp Indonesia in 2017. What surprised them was the comments. Students wrote that they understood basic science better from the channel than from their classes. Fikri has described seeing big value in that, in the mix of science, teaching and filmmaking. A channel was a hobby. A product built the same way could be a business. He took the next step the way few people do. Fikri left his doctoral program, a path that had held him since his teens, and founded Pahamify in 2018 with Ikhsan and Edria Albert, a game designer and ITB alumnus who ran a small studio in Bandung. The app combined animated lessons with game design so that studying felt closer to playing. The hard part was not the content. It was the customer. Parents were willing to spend heavily on offline tutoring but resisted paying for online lessons, and Fikri has named that as the main challenge. The pricing stayed freemium, with a live exam session for Rp20,000, a month for Rp40,000 and a year for Rp400,000. Then 2020 arrived. Schools closed, families went online, and Pahamify's user base grew fiftyfold that year while paid subscribers rose tenfold from 2019. It joined Y Combinator's Winter 2020 batch, and in November it raised a Series A from Shunwei Capital, Insignia Venture Partners and Lien Family Office, in an undisclosed amount. It still competes with larger names, Ruangguru and Zenius among them, and the founders have talked about adding younger students. The lesson is about listening to an audience before building for it. The channel was the research, and the comments were the proof, long before there was an app. ## Dekoruma (Indonesia) URL: https://startup.talk/indonesia/dekoruma/ Founders: Dimas Harry Priawan, Aruna Harsa Founded: 2015; Based in: Indonesia (Jakarta); Sector: Home and living e-commerce / interior design Before Dimas Harry Priawan joined Endeavor's elite network, a board member gave him an unglamorous piece of advice. During his 2021 selection panel, Raoul Oberman suggested he upgrade his camera and lighting. The pitch mattered, but so did how it looked on screen. It was a small, human note about a founder who had learned to work with less. Priawan started Dekoruma in 2015 with Aruna Harsa, an online marketplace for people squeezed into smaller urban homes. The idea was that decorating, buying furniture and hiring a designer should happen in one place. It was a good idea in a market where several well-funded rivals had the same one. For four years, from 2015 to 2019, Dekoruma had relatively little money compared with those rivals, and raising more was hard. Investors including BeeNext and Convergence Ventures backed an early round in 2016, and Blibli's investing arm and AddVentures came in for a Series B in 2018. But the competition spent faster. Priawan chose not to chase them. In his own words to Endeavor, the rule was to play your own game and not be distracted by what others were doing. That meant building things the marketplaces did not offer: custom-made furniture, a private label, and in 2019 a physical showroom, the Dekoruma Experience Center in Kebayoran Baru. A customer who touches a sofa buys differently from one who scrolls past it. The approach attracted bigger cheques. A pre-Series C round from InterVest followed in 2020, and in 2021 Dekoruma announced a US$15 million Series C, about Rp216.8 billion, saying it had reached positive EBITDA and planned an IPO. The IPO never came. Instead, in 2024 Blibli Group bought 99.83 percent of its Series C shares in a deal reported at Rp1.16 trillion. Priawan still mentors other founders through Endeavor's Scale Up programme, and the point he passes on is the one that kept his company alive in the lean years. A scarce budget can feel like a verdict, but it can also be a filter. Without the money to match rivals, Dekoruma had to pick a different fight, and the different fight turned out to be the one worth buying. ## Bank Jago (Indonesia) URL: https://startup.talk/indonesia/bank-jago/ Founders: Jerry Ng, Patrick Walujo Founded: 1992 (as Bank Artos in Bandung; relaunched as Jago in 2020); Based in: Indonesia (Jakarta); Sector: Digital banking In January 2016, a small Bandung bank sold new shares at Rp132 each and raised about Rp31.85 billion. It was a modest listing on the Indonesia Stock Exchange for a modest lender called Bank Artos, which had been founded in 1992. Nobody would have called it a technology company. The man who would transform it was born in Pontianak in 1965. Jerry Ng studied business at the University of Washington and built a career at Citibank Indonesia, BCA and Danamon before running Bank Tabungan Pensiunan Nasional, BTPN, as president director from 2008 to 2019. At BTPN he is credited as the initiator of Jenius, one of Indonesia's early digital banking services. It was a good vantage point for seeing what a bank could be when it stopped being a branch network. He left BTPN in 2019. Then came the move that few expected: with Patrick Walujo, he led a consortium that took control of Artos late that year, through his vehicle Metamorfosis Ekosistem Indonesia. They did not buy a brand. They bought a licence and a listing, and a clean place to start. In 2020 the bank became Bank Jago. In December, Gojek's payments arm took a stake of roughly 22 percent, and GIC invested in 2021. The idea was life-centric banking, built into apps people already used, including GoPay and the Bibit investment platform. The market loved it, then punished it. By late 2021 and early 2022 the shares traded around Rp18,000 to 19,500, more than 140 times the IPO price, with a market value above Rp240 trillion. By late 2022 they had fallen to about Rp3,300 to 3,600 as the digital bank frenzy cooled. The price cost investors dearly, but the bank kept building. Net profit in 2025 was Rp276 billion, up 115 percent, and customers reached 18.2 million; by July 2026 they were about 19.4 million, and the shares closed that week at Rp1,405. The quiet lesson is that Ng did not start a bank from nothing. He found a small, dormant licence and used twenty-five years of experience to give it a purpose. A founder's tools do not have to be new, and an overlooked asset can be worth more than a clean slate. ## DANA (Indonesia) URL: https://startup.talk/indonesia/dana/ Founders: Vincent Iswara Founded: 2017 (launched 11 November 2018); Based in: Indonesia (Jakarta); Sector: Fintech / digital wallet For about a year and a half, Vincent Iswara kept talking to Ant Financial about the digital wallet business he was already running in Indonesia. In the end the answer was a polite no. Ant's people decided his company was not the right vehicle for long-term growth. They did not say no to him. That detail is easy to miss in the DANA story, which is usually told as a big-money launch: an Emtek-backed wallet, Alipay-grade technology, a national licence. But the founder at the centre of it had first been turned down as a business and kept as a person. He went on to learn inside Ant's ecosystem in China, saw how a mobile wallet could change the way ordinary people pay, and came home to build something different. DANA was established in 2017. Iswara launched it on 11 November 2018, and the licence to operate as a fintech firm across Indonesia came on 5 December. The speed mattered. By the end of June 2019, DANA said it had 20 million users and 1.5 million transactions a day. Bills, game top-ups, barcode payments and online shopping were all folded into one app, working alongside banks such as Bank Mandiri, BCA and BRI. Then the hard part started. In an interview with Fintech News Singapore, Iswara described the first five to seven years as pretty much on the edge. Rivals handed out cashback of 50 to 100 percent, so users came for the discount and left when it ended. Regulation kept moving. Investors were not always sure. He has admitted that the team doubted, early on, whether they were even in the right industry. What almost nobody outside the company noticed was how many people stayed. Of the original 81 employees, 65 were still there after eight years, a retention rate most startups in a cashback war could not dream of. Iswara says the growth came from steady, small steps rather than one breakthrough, and that he still likes putting his hand to the dirty work with his team. The money followed the patience. In 2022 DANA raised a US$250 million round with Sinar Mas units and Lazada Group joining, and reports put its valuation above US$1 billion. Later 01Fintech invested too. The company now says it serves more than 200 million users. Iswara's own view of the market is unfashionable for a payments founder. He thinks Indonesia's problem is no longer access to financial apps but financial literacy, knowing what to do once the app is on the phone. There is a lesson in the 18 months of talks that ended in rejection. The business Ant turned down was not the end of Iswara's story. It was the apprenticeship for the one that followed. ## Ula (Indonesia) URL: https://startup.talk/indonesia/ula/ Founders: Nipun Mehra, Derry Sakti, Alan Wong, Riky Tenggara Founded: 2020; Based in: Indonesia (Jakarta); Sector: B2B e-commerce / retail supply chain for small shops (warung) Most people who sit on one side of the funding table stay there. Nipun Mehra did not. After years as an executive at Flipkart in India and then as a partner at Sequoia Capital India, he moved to Indonesia and started a company himself, one that would have to ask investors like his old colleagues for money. The idea was unglamorous on purpose. Indonesia has millions of small neighbourhood shops, warung, that buy stock from a long chain of middlemen and have little access to credit. Ula, founded in 2020, set out to give those retailers a single app for ordering goods at better prices, along with financial and supply-chain services built around the way they actually trade. It did not try to do this alone. Mehra's co-founders brought operating experience from the companies that had already tried to solve Indonesian commerce at scale. One came from Amazon. Derry Sakti had overseen P&G's operations in Indonesia, which meant he knew how packaged goods really reach a shop shelf. Riky Tenggara came from Lazada. Between them they covered the buying, the distribution and the technology. The money arrived quickly. A US$10.5 million seed round came in June 2020, followed by US$20 million of Series A in January 2021. In October 2021, a US$87 million Series B was co-led by Prosus Ventures, Tencent and B Capital, with Jeff Bezos's Bezos Expeditions, Northstar Group, AC Ventures and Citius taking part. A month later Tiger Global added US$23.1 million, bringing the round to US$110 million. Flipkart's Binny Bansal joined as an investor. With that capital Ula announced a paylater product aimed squarely at warung owners, letting them stock up now and pay later. It is a tempting product and a risky one. Lending to tiny shops with thin records is where many supply-chain startups have stumbled, and Ula's own public record on how its credit book performed afterwards is sparse. What the public record does show is the bet itself. Endeavor selected Mehra in October 2021, describing his goal as capturing a meaningful share of Indonesian retail, a market it put at more than US$300 billion. The lesson is hidden in the career move rather than in any number. An investor who has read thousands of pitch decks knows exactly how hard distribution is. Mehra chose to try it anyway, and chose partners who had already lived inside the problem. ## Blibli (Indonesia) URL: https://startup.talk/indonesia/blibli/ Founders: Kusumo Martanto, Martin Hartono, Lisa Widodo, Hendry, Lay Ridwan Gautama Founded: 2011; Based in: Indonesia (Jakarta); Sector: E-commerce / omnichannel retail The name was chosen to be said without thinking. Blibli is short for beli-beli, Indonesian for buy, buy. When it began in 2011 as a subsidiary of the Djarum group, the people behind it were asking Indonesians to do something that did not yet feel normal, which was to put a washing machine or a phone in an online basket and trust that it would arrive. The founders listed for the company include Kusumo Martanto, Martin Hartono, Lisa Widodo, Hendry and Lay Ridwan Gautama. Public records say little about their early lives, but they say plenty about what kind of company Blibli was meant to be. Djarum is a family conglomerate known for cigarettes and banking, and its investment arm, GDP Ventures, led by Martin Hartono, chose to treat e-commerce as a long game rather than a quick flip. That patience is the real story. Blibli spent its first decade in a market where Tokopedia, Shopee and Lazada burned cash on discounts. A backer that could outlast the burn bought it time. In 2017 Blibli bought tiket.com. In 2021 it took a 70.56 percent stake in PT Supra Boga Lestari, the operator behind Ranch Market and Farmers Market, which gave an online retailer a supermarket network and a click-and-collect option. Then came the public test. In late 2022, Blibli listed on the Indonesia Stock Exchange under the ticker BELI, raising Rp7.99 trillion, about US$509 million, the second-largest IPO of that year after GoTo. It became Indonesia's third unicorn on the way. In 2024 it added Dekoruma, a furniture and home decor platform. The cost of staying power still shows. In 2025, Blibli reported net revenue of roughly Rp22.36 trillion, up about 34 percent, while its operating loss narrowed to about Rp2.01 trillion from Rp2.36 trillion the year before. Analysts have questioned whether growth is coming at the expense of margins, and the company says profitability and margin improvement are priorities. Djarum's group still holds about 83.68 percent, so the public owns just a sliver of the story. There is a useful lesson here for founders who raise venture money in the usual way. Blibli's survival did not depend on a perfect product in year one. It depended on a patient owner, and on managers who used the time to buy the things e-commerce rivals had to build, such as stores, travel and home categories. Patience is a strategy only if the business uses it to get stronger. ## Bibit (Indonesia) URL: https://startup.talk/indonesia/bibit/ Founders: Sigit Kouwagam, Wellson Lo, Johny Susanto Founded: 2019; Based in: Indonesia (Jakarta); Sector: Fintech / robo-advisor mutual funds Sigit Kouwagam had already built one investing app before he built the one that made his name. He was a co-founder of Stockbit, the community and trading platform for Indonesian stock investors, alongside Wellson Lo and Johny Susanto. Their stated motive was simple. Too few Indonesians understood investing, and a social network of people sharing market views looked like a way in. But stocks are intimidating. Open any brokerage app and you face tickers, candlestick charts and the risk of losing money on a bad guess. In early 2019, the Stockbit team launched Bibit, a mutual fund app that did the opposite. Instead of asking users to pick shares, it asked about their risk appetite and suggested funds, the robo-advisor model. Reports differ on whether Stockbit launched Bibit or acquired the original app that year, but Kouwagam ended up as Bibit's chief executive, which tells you how the company saw itself. It was a product built by people who had already learned, through Stockbit, what first-time investors found confusing. Investors noticed fast. In January 2021 Sequoia Capital India led a roughly US$30 million round. By May 2021, about US$65 million, again led by Sequoia India, brought Prosus Ventures, Tencent, Harvard Management Company, AC Ventures and East Ventures to the table. In May 2022, Singapore's GIC led a round of more than US$80 million, with Prosus and existing backers joining. Kouwagam told TechCrunch that earlier funding had taken total backing to US$45 million before that round, and the new money was meant for products, technology and hiring. It is a tidy ladder, and it carries a risk that the founders have not publicly dwelt on. Robo-advisors earn on fees and assets under management. When markets fall, small savers tend to pause, and rivals such as Ajaib and Stockbit's own brokerage products compete for the same wallet. The company said its fresh capital would go partly to investor education. The lesson is in the sequence. Kouwagam did not start with a clean idea of what investing should look like. He started with a community, noticed where beginners stalled, and then built a second product just for them. Founders often look for a new market. Sometimes the market is the people already asking you the same question. ## Cakap (Indonesia) URL: https://startup.talk/indonesia/cakap/ Founders: Tomy Yunus Tjen, Yohan Limerta Founded: 2014 (as Squline; renamed Cakap in 2019); Based in: Indonesia (Jakarta); Sector: Edtech / online language learning In 2009, Tomy Yunus landed in Beijing and could not speak Mandarin. He was 23, and as he has told the Indonesian magazine SWA, that was the first time in his life he had really tried to learn a foreign language. It is an odd origin for a language-learning founder, and it is the reason the company exists. Yunus enrolled at the Beijing Chinese Language and Culture Center. In the same classrooms he met Yohan Limerta, a fellow Indonesian. Both were beginners, both struggling with tones and characters, and both knew how different the experience was in China, where the language was in the street, from Indonesia, where it was in a textbook. Yunus's background was in economics. He studied at Universitas Bina Nusantara in Jakarta and later took a master's in economics at Renmin University of China. When he looked back at Indonesia, he found something striking in a survey he conducted: only about 10 percent of Indonesians were fluent in a foreign language, usually English. The gap looked like a business. In 2014, the two founded Squline, a platform for learning Mandarin. The name was a problem. Squline meant little to most Indonesians, and by 2019 the team wanted a name that people could remember and that sounded like their country. They chose Cakap, an Indonesian word meaning able or skilled. The rename came with a wider scope. Cakap now offers courses for children and adults, including preparation for language proficiency certificates, with live online classes. Funding followed: US$3 million Series A+ at the end of 2020 led by Heritas Venture Fund, then a US$10 million Series B in December 2021 co-led by Centauri Fund, formed by MDI Ventures and KB Investment, and Heritas Capital. The company said it would spend on machine learning for personalised study. The founders have since been named to Fortune Indonesia's 40 Under 40 in 2024, and EY named both finalists for Entrepreneur Of The Year Indonesia in 2025. There is a small, human lesson in how this started. Yunus did not found Cakap because he was a gifted linguist. He built it because he had been a frightened beginner, in a foreign city, with a classmate who understood exactly how it felt.