Legora raised $550 million in October at an $8.8 billion valuation after passing $100 million in ARR in April, less than 18 months after launch. The Stockholm company now serves more than 1,800 customers with agents that draft Word documents and review thousands of contracts at once. The case matters for business because it shows where vertical AI agents already earn money and displace hourly billing.
How Legora scaled legal agents to 1,800 clients
The company began in a Stockholm basement as Leya with a simple combination of ChatGPT, Swedish law and search. CTO Jacob Lauritzen, a Y Combinator alumnus, first helped founder Max Junestrand with the application before joining as head of engineering. It raised $550 million in March at a $5.55 billion valuation, then repeated the amount in October at $8.8 billion. Its latest ARR figure is reported above $200 million, and the Wexler acquisition was its fifth this year.
The platform is built around projects that hold files, an agent, Tabular Review and lists. In Tabular Review each row is a document and each column is a question, so 10,000 employment contracts can be checked for termination periods, IP assignment and severance terms. The grid supports up to one million cells but starts with 10 rows to refine prompts before a full run. Lists work as a Notion-style database with triggers, while intake runs as a Slack bot that turns requests into tickets.
Evaluation is shaped by confidentiality because almost everything uploaded belongs to firms or their clients and cannot be read by staff. Three engineers and three lawyers build RL environments that simulate customer work, including a 10,000-file data room for due diligence with red-flag reports. Results are plotted on a heat map by practice area, where a new model often improves some tasks and regresses on others. In production, failures are grouped into 10 to 20 known modes and checked with low-cost LLM monitors behind feature flags.
What this means for corporate AI adoption
For companies that buy legal services, routine reviews in the US are already moving from hourly billing to flat project fees, with some firms doing the work for free to retain clients. Value shifts toward M&A, litigation and direct client time, which changes how legal budgets are negotiated. Small legal teams gain the ability to run large portfolio reviews internally, while large firms can process far larger document volumes without adding headcount.
The limits concern model choice, infrastructure and hallucinations. Legora runs entirely on Azure because law firms accepted Microsoft through 365 and rejected other clouds at the start. The company long resisted a model picker, arguing that lawyers should not decide between quality, cost, latency, governance and regional availability themselves. Every agent draft carries citations that highlight a bounding box in the source PDF, with post-training of open-weight models used for narrow tasks such as legal research.
The marker to watch is how deployment splits between direct use and embedded access over the next two years. Lauritzen expects large enterprises to reach Legora through tools such as Claude Cowork, while law firms continue working inside Legora itself. If forward deployed engineers rotate through clients and feed integrations back into the core product, the platform model holds. Continued demand on $400 million M&A deals would confirm that specialized agents retain their place alongside foundation models.
