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.
Companies spend huge sums on call centres, and customers hate being stuck in menus.
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.