Legora Says NDA Review Fell From 20% of Capacity to Under 1%
Legora says AI has cut the time its own legal team spends reviewing NDAs by more than 95%, from around 20% of the team's capacity to under 1%. The figures come from Legora.
Transcript
Legora says AI cut its own NDA review work by over ninety five percent. That number comes from Legora.
Legora's general counsel says those reviews once took twenty percent of her team's capacity. She says it is now under one percent.
Legora's CEO cites a Fortune one hundred customer with ninety percent faster contract turnaround, and nine and a half hours returned per lawyer weekly.
Legora says that customer also saved more than twenty million dollars on outside legal work. Its CTO says agents reach a working fix in about five minutes.
Legora reported these figures for OpenAI's own customer story page. No outside party checked them, and this is not a product launch.
Inboxsmith builds AI phone and email agents for small business. Please like and subscribe for more news.
Sources
Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:
- PROVENANCE OF THE WHOLE SCRIPT. Every sentence in this script traces to this single page, which is OpenAI's own customer story about Legora, a company that pays for the OpenAI API. The page carries OpenAI's customer story template fields, Company size Mid-market, Region Global, Industry Technology, Products API, Codex, ChatGPT, plus two Contact sales buttons and the closing marketing line that more than 1 million businesses are achieving meaningful results with OpenAI. It is sales material, not a product announcement, and no release, version, or new capability is described as shipping. No second source, no transcript, and no independent verification of any number was used or exists in this file. Every statistic below is self reported by Legora, or by Legora's own unnamed customer, to Legora's own model vendor.(Legora is building the agentic operating system for legal work)
- HOOK sentence one, "Legora says AI cut its own NDA review work by over ninety five percent", traces to the page's Results at a glance block, "95%+ reduction in Legora legal-team time spent reviewing NDAs, from ~20% of team capacity to under 1%", and to the same figure in the page's stat strip. The figure is rendered as spoken words per the narration rules, and the sentence is attributed to Legora in the channel's voice because Legora is the reporting party. HOOK sentence two, "That number comes from Legora", is the required self reported frame, carried in sentence two rather than buried later.(Legora is building the agentic operating system for legal work)
- SCENE 1 both sentences trace to the section titled Giving Legora's own lawyers room to do higher-value work: "Before the workflow was introduced, Kadri Bergstrom estimates that reviewing these boring documents consumed around 20% of the legal team's capacity. Today, it's under 1%." The attributor named out loud is "Legora's general counsel", matching the page's byline for Emma Kadri Bergstrom, General Counsel, Legora. The page's own word is "estimates", so this is an internal estimate and not a measurement; the narration therefore says "she says" for the second figure as well. The publication word "today" is not narrated, per the date ban, and is replaced with "now". The surname diacritic is dropped because the name is not spoken.(Legora is building the agentic operating system for legal work)
- SCENE 2 sentence one, "Legora's CEO cites a Fortune one hundred customer with ninety percent faster contract turnaround", traces to "Junestrand cites one Fortune 100 customer that has: reduced contract turnaround time by 90%". The page's own verb is "cites", which the narration reuses, and the attributor is Max Junestrand, CEO and co-founder, Legora. SCENE 2 sentence two, "He says it returned nine and a half hours per lawyer weekly", traces to the same bulleted list, "returned 9.5 hours per lawyer each week for higher-value work". Fortune 100 is spoken as "Fortune one hundred" because a brand glued to digits is mishandled by the voice engine. The customer is never named on the page, so no reader can check either figure.(Legora is building the agentic operating system for legal work)
- SCENE 3 sentence two, "Its CTO says agents reach a working fix in about five minutes", traces to "Coding agents can now search millions of lines of logs and pursue many possible explanations simultaneously, sometimes taking Legora from bug report to working fix in around five minutes", attributed on the page to Jacob Lauritzen, CTO at Legora. The page hedges with "sometimes", so this is a best case and not a typical case. The word agentic, which is the page's own branding, is never spoken because the voice engine garbles it in every tested configuration.(Legora is building the agentic operating system for legal work)
- SCENE 3 sentence one, "Legora says that customer also saved more than twenty million dollars on outside legal work", traces to the same Junestrand bulleted list and to Results at a glance, "$20M+ saved on external litigation and diligence for the same customer". This is the only money figure spoken in the entire narration, it sits alone in its own sentence, and the only adjacent number is the small figure five minutes, so no two large figures are spoken close together. The narration compresses external litigation and diligence to outside legal work for the word budget, and the phrase "the same customer" keeps the figure attached to the Fortune one hundred customer introduced in scene 2 rather than to Legora itself. The sentence order within scene 3 was set deliberately so the customer figures finish before the engineering claim begins.(Legora is building the agentic operating system for legal work)
- SCENE 4 is the required provenance scene and is not a claim from the page. "Legora reported these figures to OpenAI for OpenAI's own customer story page" states the publishing relationship visible in the page's template and header. "No outside party checked them, and this is not a product launch" states the two facts that the page itself establishes by omission: it names no auditor, no third party, and no methodology for the 95%+, 90%, 9.5 hour, 20 million dollar, or five minute figures, and it describes no release. DELIBERATELY NOT NARRATED, with reasons: the Tabular Review claim that thousands of contracts across multiple questions can be analysed in minutes was cut for the word budget and because no baseline is given; the CTO's "20 solutions at the same time" prototyping claim was cut in favour of the more concrete five minute bug to fix figure; the growth from three offices to 18, three financing rounds and five acquisitions in the past year was cut because it describes Legora's own scaling rather than a legal work outcome; the leadership lesson that median usage is 3x higher when more than 75% of leaders use Legora was cut because the page gives no sample, no time frame, and no definition of active use; the Legora plugin for ChatGPT, the founding in Stockholm in 2023, the GPT-3.5 origin anecdote, and the forward looking quotes about overnight diligence and real time collaboration were all cut for length, and model names were avoided entirely because brand tokens glued to digits are garbled by the voice engine. SCOPE LIMIT: no fact from outside this one page appears anywhere in this file, in narration, on screen, or in sources, and no company financing detail beyond the page's own mention of three financing rounds was considered.(Legora is building the agentic operating system for legal work)