A model developer has bought a drug company
Anthropic said on 27 October that it has acquired Mithril Medicine and is running deep research partnerships with six drug developers, in a progress report from its science team. The company says the work is still at an early stage and mostly preclinical.
Mithril’s founder, Cody Tranbarger, used Claude to coordinate work across 11 external laboratories. Anthropic says a possible treatment for one form of muscular dystrophy went from idea to promising results in mice in less than 18 months, and that if those results hold, a drug could be ready for human clinical trials as soon as late 2028.
That date is the claim to hold onto, and also the one that cannot be checked yet. Everything in the report is reported by Anthropic or by its partners. None of it has been published in a peer-reviewed journal or evaluated by a regulator.
What the six partners report
Generate Biomedicines, working on CAR T-cell therapy, had six scientists use Claude Code to propose shorter designs for a genetic component. One design was about 70% shorter than the standard version. In lab tests, Anthropic says, it let immune cells make roughly ten times as many receptors, killed tumour cells more efficiently and stayed active longer.
Divergence Labs, working on food allergies, had Claude turn the team’s goals into a design plan and generate about 35,000 antibody designs in roughly a month. Early tests produced three designs that bind IgE and 20 that dislodge it. Three of those 20 dislodged IgE more than the standard medication does.
BigHat Biosciences had Claude propose changes to two antibody-fragment leads. The lab made and tested about 1,010 variants over roughly 10 weeks. The best versions bound more than ten times more tightly; one lead stayed in solution at a temperature 18°C higher, and the other clumped more than 100 times more slowly.
The paperwork result is the most checkable one
Enveda put Claude on 1,061 pages of manufacturing and regulatory documentation. Anthropic says it caught all 11 issues human reviewers had found and flagged seven more, and that human review time fell 89%, from over 25 hours to under three. Claude also drafted two regulatory documents; one took four hours against about three months by hand.
Of everything in the report, this is the result a reader can most easily reason about. A document review has a ground truth — the 11 issues humans found — and a measurable time. The seven additional flags are the part that still needs adjudication: a flag is not yet a finding.
The Column Group is forming a company called Pleiades to run drug programmes with a small team. Its first Claude skill, for assessing drug targets, was evaluated on 45 target-disease pairs. Bristol Myers Squibb is using Claude to assess early-stage trial plans and estimate their likelihood of success, and the partnership is planned to expand in 2027.
What to watch
Two things. First, whether the muscular dystrophy candidate files an investigational new drug application on the stated timeline — late 2028 is close enough to be falsifiable. Second, whether any of these results reach peer review. A mouse result and a binding assay are the stage at which most drug programmes look promising; the attrition comes later.
