Anthropic’s life sciences team put numbers on the table for the first time on October 27. For several months it has been working with six organizations on new medicines: BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines and The Column Group. And there is something in the post that was not public before: Anthropic has bought a drug company.
What came out of the six partnerships
Generate Biomedicines wants to rebuild CAR T-cell therapy so it also works against solid tumors. Six of its scientists used Claude Code to propose new designs for one stretch of the genetic instructions. The best one ran about 70% shorter than the standard version and did better in the lab: roughly ten times as many receptors, faster killing of tumor cells, immune cells that stayed active longer.
Divergence Labs is working on food allergies, which needs an antibody with two arms. Claude turned the goals into a design plan and then, largely on its own, generated around 35,000 antibody designs in about a month. Early tests produced three hits for one arm and 20 for the other.
At BigHat Biosciences, Claude proposed changes to two drug leads, and the lab built and tested about 1,010 variants in ten weeks. Halfway through, Claude and an Anthropic researcher decided the usual heat-stability and clumping metrics were misleading, proposed different ones, and turned out to be right. The best versions bound more than ten times more tightly, one stayed in solution at a temperature 18 degrees higher, and one clumped more than a hundred times more slowly.
Enveda is the paperwork case. Its manufacturing chemists were spending up to 75% of their time on documents. Across 1,061 pages, Claude found all eleven issues human reviewers had spotted and seven they had not. Review time dropped 89%, from over 25 hours to under three. Claude also wrote first drafts of two regulatory documents, one of them in four hours instead of roughly three months by hand.
The Column Group is setting up Pleiades, a company built around Claude from day one. Its first Claude skill judges whether a drug target is worth chasing, tested on 45 target-disease pairs. Bristol Myers Squibb recently started using Claude to assess trial plans and estimate their chance of success.
Mithril Medicine: a one-person company, now inside Anthropic
Mithril Medicine was developing treatments for two rare forms of muscular dystrophy, and it was one person. Founder Cody Tranbarger ran contracts, logistics and experiments across eleven external labs through Claude, and handed much of the company’s operations to it. Getting from an idea to promising results in mice took him under 18 months. He is continuing the work at Anthropic, and if the candidates hold up, a human trial could start as early as late 2028.
The biggest lever is the paperwork
The molecule numbers read well, but they are preclinical and early. What is measured solidly so far is something else: 89% less review time, three months of writing down to four hours. Drug development is documentation to a large degree, and that is where Claude pays off today. Anthropic making the same models broadly available through Claude Science fits the picture: the bottleneck is not only in the lab.