Large law firms are paying to train AI on their own deal and trial files
Kirkland & Ellis agreed this year to pay Palantir $500 million to build technology for its private equity fundraising advice, Business Insider reported. With Harvey saying it has users at 80% of the largest 100 firms, one consultant says the edge has to come from what the models know.
Reality
- Evidence35
- Adoption50
- Hype gap+30
- Incentives70
- Confidence50
buildOne report1 publisher Z.ai's GLM Legal claims fewer invented citations with no independent accuracy data behind it
Z.ai launched GLM Legal to cut hallucinated citations in legal work, but it published no accuracy figures anyone can verify independently. Teams weighing it against Harvey or CoCounsel will have to build their own citation test, because no public number supports the comparison.
Reality
- Evidence22
- Adoption
- Not scored
- Hype gap+30
- Incentives
- Not scored
- Confidence45
One in five UK and Irish law firms deep into AI report trouble meeting billable-hour targets
About one in five UK and Irish law firms that have widely adopted AI report difficulty meeting billable-hour targets, legal software company Clio found. Firms must now decide who keeps the hours AI saves, and who pays to train juniors once software does the document review.
Reality
- Evidence50
- Adoption65
- Hype gap+20
- Incentives70
- Confidence50
GPT-6 Astra scores 37 points lower on ARC-AGI-3 without OpenAI's proprietary harness
OpenAI's GPT-6 Astra scored 99.9% on ARC-AGI-3 through a proprietary harness but 62.7% on the benchmark's neutral one, Live Science reported. Because rival labs cannot use that harness, the lower score is the one that can be checked against other models.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+60
- Incentives70
- Confidence45
Repeat buyers push AI-to-AI startup acquisitions 14% past last year's total
Venture-backed AI companies bought 195 AI startups through Sept. 29, Crunchbase data shows, 14% more than all of 2025, while buyers rose just 2%. The same companies are buying more often, so sellers face a pool of acquirers that has barely grown.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+10
- Incentives40
- Confidence60
Clio buys Learned Hand on a bet that judges will need AI of their own
Clio, last valued at $5 billion, has bought Learned Hand, a startup that builds AI tools for judges, on undisclosed terms. Jack Newton is betting that litigants without lawyers will push courts toward AI, and that the bench can lead Clio to Big Law.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+30
- Incentives60
- Confidence50
Anthropic opens a public storefront for more than 2,000 Claude connectors and plugins
Anthropic's Claude Marketplace is live with more than 2,000 connectors and plugins, including ones published by Atlassian, Google, Microsoft and Salesforce. Each connector enabled from it adds a supplier that belongs in third-party and access reviews.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+15
- Incentives60
- Confidence45
Tenet is post-trained on Moonshot's Kimi K3, converting per-call payments to OpenAI, Anthropic and Google into a fixed training bill. The counterparty risk moved rather than disappeared.
Reality
- Evidence55
- Adoption15
- Hype gap+30
- Incentives60
- Confidence55
Lightspeed and Diffusion co-led $550m at a $15.6bn mark, which added $4.6bn of paper value in six months and only reads as ordinary software if $400m of ARR becomes about $1.56bn. Eighty of the Am Law 100 already buy.
Reality
- Evidence55
- Adoption70
- Hype gap+30
- Incentives60
- Confidence60
Astra for Law is GPT-6 Astra with a legal search index and analysis instructions wrapped around it. OpenAI will also sell that index through an API to Harvey and Legora, at a price it has not published.
Reality
- Evidence50
- Adoption30
- Hype gap+20
- Incentives70
- Confidence60
buildOne report1 publisher Anthropic lets committed Claude spend pay for partner software
Claude Marketplace groups connectors, partner products and consultants in one catalog, and eligible customers can direct part of an existing Anthropic commitment at the software listed there.
Reality
- Evidence52
- Adoption30
- Hype gap+8
- Incentives72
- Confidence60
Nearly half of one newsletter's 25 product openings ask for eval experience
Lenny's newsletter counted eval experience in nearly half of 25 product manager postings it listed, and the two practitioners it commissioned say most teams skip error discovery and write metrics for failures they never found.
Publishers:lennysnewsletter.com
Reality
- Evidence32
- Adoption55
- Hype gap+33
- Incentives85
- Confidence58
Legora's CEO calls a European frontier AI lab wishful thinking
Max Junestrand told Politico that Europe's opportunity is in applications, and in the same conversation he agreed that European customers waiting longer for new American models is a competitive disadvantage.
Reality
- Evidence55
- Adoption70
- Hype gap+18
- Incentives80
- Confidence58
Harvey's cost of serving a dollar of revenue tripled in six months
A March 2026 update lifted usage and inverted Harvey's gross margin, from about 50% to minus 50%. Its answer is Tenet, a legal model post-trained on the open-weight Kimi K3 base and run on compute it controls.
Reality
- Evidence26
- Adoption44
- Hype gap+34
- Incentives55
- Confidence37
buildConfirmed3 publishers The legal toolkit ships as its own entry in the ChatGPT model picker and its own API string. Firms weighing it against retrieval stacks they already built still have no price to compare.
Perspective Coverage
3 publishers
- Builder
- Builder 38%
- Operator
- Operator 30%
- Investor
- Investor 32%
Reality
- Evidence55
- Adoption35
- Hype gap+25
- Incentives70
- Confidence62
Rechtman's 30-100x market claim restates software's 1-3% revenue capture
Yoni Rechtman of Slow Ventures says AI lets vendors sell the outcome instead of the workflow. His 30-100x market expansion is software's 1-3% revenue share inverted, and the one priced case in the framework moves 10x.
Publishers:insights.euclid.vc
Reality
- Evidence30
- Adoption20
- Hype gap+42
- Incentives78
- Confidence60
Raindrop replays real production traffic against every pull request to test agent changes
The San Francisco company now has $50m in total funding and a research preview, Simulations, that runs anomaly detection over replayed user traffic on each pull request. It works only if you already have traffic.
Reality
- Evidence33
- Adoption26
- Hype gap+31
- Incentives82
- Confidence56
Beacon folds Haize Labs into an in-house research group for its 45 software companies
Haize Labs sold red teaming and evaluation tooling to frontier labs and large enterprises. Its engineers are now an internal research group inside a Toronto holding company that buys small-business software.
Reality
- Evidence45
- Adoption30
- Hype gap+12
- Incentives70
- Confidence52
buildOne report1 publisher AIUC raises $40M to put an insurance policy behind its AIUC-1 agent audit
Rune Kvist's startup wants to sell enterprises an audited, insured answer to who pays when an agent fails, and the round was announced in a podcast episode that names Cursor, Harvey, Lovable and ElevenLabs.
Reality
- Evidence42
- Adoption20
- Hype gap+35
- Incentives78
- Confidence45
Ribbit leads a $40M round into an AI agent standard seven vendors have certified against
The Series A led by Ribbit Capital takes AIUC's total funding to $55M, paying for quarterly audits of a standard that Cursor, Harvey, KPMG and four other named vendors certify against, and for a push up to frontier models.
Reality
- Evidence42
- Adoption38
- Hype gap+33
- Incentives90
- Confidence55