Anthropic reports early results from six drug-development partnerships using Claude

Tuesday, 3 November 2026By Sridhar Mukkandi1 min read

Anthropic described on 27 October several months of work with six drug developers, and said all of it is at an early, mostly preclinical stage. With Enveda, Claude reviewed 1,061 pages of manufacturing documentation, caught all 11 issues that human reviewers had found and flagged seven more, cutting human review time by 89%, from over 25 hours to under three. A regulatory document it drafted in four hours, against about three months by hand, still went through standard human review, which found one minor error.

With BigHat Biosciences, about 1,010 antibody variants proposed with Claude were made and tested over roughly 10 weeks, and the best bound more than 10 times more tightly. Divergence Labs generated about 35,000 antibody designs in about a month. Anthropic also said it acquired Mithril Medicine, a one-person company working on two rare muscular dystrophies; if results hold, human trials could start as early as late 2028.

So what

The Enveda result shows how to trust a model reviewer: compare it against issues people already found, then keep a human check at the end. Lab tests, not the model, still decide what works.

SourceAnthropicwww.anthropic.com/research/early-progress-drug-development Go deeper · EssayCheck the checker before you trust it

Drafted with AI from the original source, then checked and edited by Sridhar Mukkandi. Spotted a mistake? Write to hello@tarkika.com and we'll correct it here.

More from The Signal

Models Mistral releases the open weights of Large 4, its 1-trillion-parameter modelAn open model at this size gives teams that can't send data to a hosted API a real option. Check the licence before you build a product on it, and test it on your own tasks: the figures are Mistral's. Mistral AI Policy Anthropic and GSA agree a 40% discount on Claude for every level of US governmentGovernment buyers get the same lesson as everyone else: price is only one input. The examples Anthropic cites, from incident response to code migration, still need someone deciding what the model may do without review. Anthropic Agents GitHub: agents started more than 40% of pull requests in SeptemberA merge rate measures what reviewers accepted, not what held up in production. If agents now write the first draft, track reverts and follow-up fixes on their changes, and spend review time where a mistake costs most. GitHub BlogAll stories