Legora builds an agentic operating system for legal work on OpenAI models

OpenAI published a customer story on October 28 about Legora, the Stockholm-based legal AI company founded in 2023, describing how it uses OpenAI models to build what chief executive and co-founder Max Junestrand calls “the agentic operating system for legal work”.

According to OpenAI, it has been a key model provider for Legora since the GPT-3.5 era, powering core workflows from agentic legal work to large-scale document review. Legora rebuilds on new model capabilities as they ship, which ties its product roadmap to model releases rather than a fixed feature cycle.

That pattern also shows up inside Legora. The company runs ChatGPT Work, Codex and other tools across its own teams; chief technology officer Jacob Lauritzen says engineers can prototype 20 solutions at once and pick the best within days, and that coding agents can search millions of lines of logs in parallel, sometimes going from bug report to working fix in about five minutes.

The legal team is the internal test case. General counsel Emma Kadri Bergström says her team analysed the NDAs it had historically signed, built a playbook and now lets AI take a first pass at incoming agreements, flagging terms outside acceptable parameters. NDA review fell from roughly 20% of the team's capacity to under 1%.

That matters because Legora grew from three offices to 18, completed three financing rounds and made five acquisitions in the past year, and its lawyers had to keep pace. AI also let the team take on work that might previously have been outsourced or narrowed, such as starting contract review and surfacing information across an M&A diligence set for lawyers to interpret.

For customers, the pitch is breadth as much as speed. Alex Fortescue-Webb, global head of legal engineering, says time and cost limits traditionally forced deep review of only the most valuable contracts, while Legora's Tabular Review workflow can analyse thousands of contracts across multiple questions in minutes, bringing risks outside the manually reviewed subset into view.

OpenAI cites a Fortune 100 customer that cut contract turnaround by 90%, saved more than $20 million on external litigation and diligence, and returned 9.5 hours per lawyer each week. Legora's own NDA reduction is put at more than 95%, and legal teams can also reach its capabilities through the Legora plugin for ChatGPT.

The figures come from a vendor and customer narrative and have not been independently audited, but they frame where legal AI is competing: not single-turn answers, but auditable workflows that chain models, memory, tool use and review steps. Lauritzen expects longer-running agents that can work through millions of pages of diligence overnight, plus real-time collaboration in which a lawyer and the model revise a contract back and forth; he says model improvements force him to reconsider the roadmap roughly every two weeks.