Anthropic buys Mithril Medicine, a drug startup with one full-time employee

A company of one, bought by an AI lab

As tmtpost.com reported, Anthropic disclosed that it acquired a biotech startup called Mithril Medicine in July. The company has a single full-time employee and two drug programs aimed at rare genetic diseases. Anthropic did not say what it paid.

The employee is the founder, Cody Tranbarger. Until now he was known online mainly as an anonymous account called Professor Oak, and he revealed his identity in a long post. He joked that the first surprise was the acquisition and that the second was who the one person is: "Surprise number two: that person is me."

How one person ran a drug company

Mithril was not literally a one-person effort. It was backed by the life-science investor Atlas Venture, used part-time scientific advisers, and sent its real lab work to eleven outside laboratories.

Tranbarger set the research direction, judged results and managed the relationships. He handed the daily grind to Claude. According to Anthropic's materials, that meant organizing handoffs between labs, tidying data, planning experiments, sorting contracts, tracking progress and drafting work emails.

He calls the approach "Pipeline-in-a-Person". From the first idea to early animal results took less than a year and a half.

What the drugs aim to do

The two programs target LAMA2-related congenital muscular dystrophy and a condition called dystroglycanopathy, as the source describes it. Both are rare inherited diseases in which muscle cells lose their link to the tissue around them, causing serious weakness and damage.

Mithril's approach uses a bispecific antibody, a protein built to grab two targets at once. Here it holds the muscle cell and its surrounding scaffold together, like a bridge.

In early mouse tests for the LAMA2 disease, animals on the highest dose had roughly double the grip strength of untreated ones. Four of the eight untreated mice died during the observation period, against one in the top-dose group. These are small animal results, not evidence that it works in people.

A pattern, not a one-off

The deal fits a run of Anthropic moves in life sciences. It launched Claude for Life Sciences, bought another biotech AI company, Coefficient Bio, and opened research tools for labs. Tmtpost.com also notes a report that it set up its own wet lab, meaning a physical lab with real biological samples.

Anthropic also shared early results from six partners. One partner, Enveda, used Claude to review a large set of drug-manufacturing documents, and the combined AI-plus-human review took far less time than a human-only review.

The caveat is the calendar. AI can speed up research steps, but it cannot skip clinical trials. The lead Mithril program is still a few years from human testing, so any medicine for patients is a long way off.