Legora, the Stockholm legal AI startup, says its own lawyers now spend under 1% of their time on NDA review, down from about 20%. The figure comes from the company's General Counsel, Emma Kadri Bergström, in a customer story OpenAI published on October 28. It is a claim from the company and its model supplier, not an independent measurement.
What Legora says it built
Founded in 2023, Legora calls its product "the agentic operating system for legal work," according to CEO and co-founder Max Junestrand. His stated ambition is that every piece of legal work in the world should run on the platform. OpenAI models power core workflows, from agentic legal work to large-scale document review, and the company says GPT-3.5 was the moment many of its early legal ideas became possible.
The pitch starts with a familiar scene: a junior lawyer told to review an impossible pile of contracts overnight, who can only cover a fraction. Legora says it can supply that kind of review "on tap".
The company eats its own cooking
Kadri Bergström says her team studied the NDAs it had signed in the past and built a playbook. Legora now takes a first pass at incoming agreements and flags terms outside acceptable limits. She also says agents can start reviewing contracts in M&A diligence, with lawyers checking and interpreting the output.
On the engineering side, CTO Jacob Lauritzen says coding agents let teams try many approaches at once: "We can explore 20 solutions at the same time and, within days, pick the one we know will be the best". Agents can also comb millions of lines of logs, sometimes going from bug report to working fix in around five minutes, the company says. Legora has also widened its internal use of ChatGPT Work and Codex.
What we can't tell from the post
The story is published by OpenAI about a customer, so it reads as a showcase. It gives no customer numbers, no error rates and no detail on how often lawyers overrule the AI's first pass. Kadri Bergström's line that AI "doesn't just make you faster. It also increases quality" is an assertion, not a tested result.
The headline figure is also loose. The post frames it as the share of the legal team's time spent on NDA review, but it does not say how that was measured, over what period, or whether the time lawyers spend checking the AI's first pass is counted. Under 1% of time could describe a real shift in workload, or it could leave out the human review that still has to happen.
The post also offers no accuracy figures for Legora's tools, and we did not find any in it. That silence is not proof of a problem, but it means the quality claim rests on the company's word alone.
The growth figures are self-reported too: three offices to 18, three financing rounds and five acquisitions in the past year. The post does not give funding amounts or revenue.
Next to watch: whether Legora or its law-firm customers publish accuracy data. If an AI does the first pass on your contracts, how much checking would you want a human to do?
Sources
Reported by the WattsUpNext desk from the sources linked below. Spot an error? Tell us at corrections@wattsupnext.com.
The WattsUpNext Brief
Get stories like this in your inbox.
One email each weekday, only the topics you choose. Real news, sourced, no fluff. Unsubscribe in one click.
