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.
The early years were the unglamorous part: selling software to a cautious industry that guards its claims data and does not like experiments.
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.