Anthropic Buys Mithril, the One-Person Drug Company That Ran Its Entire Pipeline on Claude

Anthropic has quietly crossed from selling models to running drug pipelines. On Tuesday the company disclosed that it acquired Mithril Medicine, a biotech company whose entire staff was one person, and published early results from research partnerships with six drug developers, including Bristol Myers Squibb, BigHat Biosciences, Divergence Labs, Enveda, Generate Biomedicines and The Column Group. Everything described in the announcement is preclinical, and every number comes from Anthropic and its partners rather than independent reviewers.

The acquisition is the headline. Mithril's founder, Cody Tranbarger, used Claude to organize contracts, logistics and experiments across 11 external labs, and took a possible treatment for one form of muscular dystrophy from an idea to promising results in mice in under 18 months. Anthropic, which bought the company in July and is only now disclosing it, calls the model "Pipeline-in-a-Person." Tranbarger has joined Anthropic and plans to start several new programs. If the candidates hold up, Anthropic expects a drug ready to enter human clinical trials as soon as late 2028.

The partner results come with concrete numbers. Divergence Labs produced roughly 35,000 antibody designs in about a month; three bound the allergy-related protein IgE, and three of twenty that dislodged IgE appeared to outperform the standard medication. BigHat Biosciences tested about 1,010 variants in roughly ten weeks, with the best binding more than 10 times more tightly than the originals. At Enveda, Claude reviewed 1,061 pages of documentation, catching all 11 issues human reviewers had found plus seven they had missed, and cut review time from over 25 hours to under three; one regulatory filing draft took four hours instead of about three months. Generate Biomedicines reports a CAR-T design with a sequence about 70 percent shorter that produced roughly 10 times as many receptors in lab tests, work done by six scientists building a system of agents in Claude Code.

The disease Mithril targeted, LAMA2 congenital muscular dystrophy, is brutal. Children born with the complete form lack a protein that anchors muscle fibers, and per research cited in Anthropic's technical report, only one of 42 such children can walk unaided; there is no treatment that changes the course of the disease. Mithril's approach is a two-headed antibody that re-pins a related residual protein to muscle fibers. In a small mouse experiment, treated animals showed roughly twice the forelimb strength and grip of untreated ones.

The caveats matter, and Anthropic's own report lists them. The mouse study had seven treated animals against four controls, and the difference fell just short of conventional statistical significance; the company says rigorous replication is needed before the result can be trusted. The Mithril report is not peer-reviewed. Anthropic also documented Claude's error rate in a week of Mithril's operations: the author found six errors, five of which Claude caught itself, but at least three had already gone out in human-approved emails before being discovered, two of them more than four days later. Treat all figures here as company-reported and not independently verified.

Around the research, Anthropic rolled out a life-sciences platform push: Claude Science, an AI workbench for researchers; Claude credits for life scientists; a Life Sciences Verification Program that gives vetted organizations access to its most powerful models with more permissive safeguards; and an open call for guest investigators proposing drug discovery research.

The pattern Anthropic is pointing at is that the hard part of early biotech is often not the science but the coordination: vendors, protocols, contracts and regulatory paperwork spread across a dozen organizations. That is exactly the long, document-heavy work that agents have gotten good at. A one-person pharma pipeline was a punchline two years ago; on Tuesday it was an acquisition. With OpenAI having bought Orchestra Bio earlier this month to push its Rosalind model into drug development, the frontier labs are now competing for the same prize: a seat inside the pharmaceutical value chain, not just an API into it.