Summary
In this Y Combinator Startup School talk, Max Junestrand, CEO and co-founder of Legora, recounts how a small team of three Swedish engineers built what has become one of the leading legal AI platforms in the world. Legora — described as an agentic operating system for lawyers — now counts over 3% of the world’s lawyers as active users and grew from $1 million to $100 million in ARR between October 2024 and mid-2026, scaling the team from three people to over 750 employees globally.
Junestrand traces the company’s origins to a pre-GPT startup called Judelica, founded by a lawyer, physicist, engineer, and psychologist who noticed law students were spending internships summarizing court cases using early Google BERT models. When GPT-3.5 arrived, Junestrand dropped out of college to join the effort full-time. The team deliberately avoided fine-tuning proprietary legal models — a prevailing assumption at the time — betting instead that foundation models would improve rapidly enough to handle legal complexity natively.
The talk offers candid lessons on selling AI into conservative professional markets, the role of domain expertise (or lack thereof) in founding vertical AI companies, and why speed of learning matters more than prior industry knowledge. A Q&A with YC partner Gustav Armstrong covers how founders should think about building on models that are “not quite good enough yet” — a question directly relevant to anyone targeting sectors like mechanical design, medicine, or finance where foundation model capabilities are still maturing.
📺 Source: Y Combinator · Published August 25, 2026
🏷️ Format: Keynote Launch
