By Hamza Faisal
Anthropic has acquired Mithril Medicine, a Cambridge, Massachusetts biotech with just one full-time employee, in a deal that turns the Claude maker into a drug developer in its own right. The acquisition, which closed in July 2026 and was disclosed on October 27, gives Anthropic its first in-house drug programmes: treatments for two rare forms of muscular dystrophy.
Its only full-time employee was founder Cody Tranbarger. With eight paid part-time consultants and advisers, he used Claude to organise contracts, logistics and experiments across 11 outside labs, taking a potential treatment from idea to promising mouse results in under 18 months. Terms were not disclosed; Tranbarger has joined Anthropic to continue the work and start more programmes.
A one-person pipeline
Mithril focused on LAMA2-related congenital muscular dystrophy and alpha-dystroglycanopathy, rare conditions in which faulty muscle-fibre connections cause weakening from birth or infancy. Drug development is normally the work of large teams with deep pockets.
Tranbarger's approach was different. Mithril designed bispecific antibodies, proteins that latch onto two targets at once, bridging laminin-alpha4 to dystroglycan to re-anchor muscle fibres. Contract labs produced the antibodies and ran the assays and mouse studies, while Claude kept the moving parts organised. In a pilot study, seven mice given one antibody format showed about twice the grip strength and peak force of four control mice.
Anthropic calls the model a "Pipeline-in-a-Person" and plans to test it on more rare-disease programmes. Tranbarger, a Harvard MS/MBA biotechnology graduate who previously worked at Prime Medicine and Casdin Capital, had posted online as the anonymous commentator "Professor Oak" before revealing his identity with the deal.
Early results, honestly labelled
The mouse results come with caveats. The technical report has not been peer reviewed, the study was not randomised, and the treatment and control groups were tested months apart. The report calls for blinded replication, and Anthropic says a drug candidate could reach human trials as soon as late 2028, subject to far more testing.
This is not a cure announcement, but a test of whether AI-assisted organisation can shorten the road to human trials, not a shortcut past safety trials. For rare diseases, where small patient numbers often leave research underfunded, a faster route would matter only if the science holds up under stricter scrutiny.
Anthropic's bigger drug bet
The acquisition was disclosed alongside early results from Claude-assisted drug work with six outside organisations: BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines and the Column Group. Most of that work is still preclinical, meaning candidates have not yet been tested in humans.
At Enveda, Claude reviewed 1,061 pages of manufacturing documentation, flagged all 11 issues humans had caught plus seven they missed, and cut review time by 89 percent, from over 25 hours to under three. Generate Biomedicines used Claude Code to design a CAR T-cell component about 70 percent shorter than standard, which in lab tests let immune cells produce roughly 10 times as many cancer-detecting receptors. Divergence Labs generated about 35,000 antibody designs in roughly a month for a food allergy treatment, and BigHat Biosciences tested about 1,010 variants of two antibody fragments in about 10 weeks.
Anthropic also announced Claude Science, a workbench with dedicated compute credits for researchers, plus a verification programme and an open call for guest investigators. The Column Group is establishing a new biotech, Pleiades, to use Claude to evaluate drug targets. Anthropic no longer wants to supply AI tools to drug companies only. With Mithril, it is betting a tiny team with a capable model can run a drug pipeline of its own.
By Hamza Faisal



