Anthropic is going all in on drug discovery, and every biotech with a big team should be worried. This week, the AI lab shared early results from six drug development partnerships, including one with Bristol Myers Squibb. Anthropic also quietly bought Mithril Medicine back in July and now runs Mithril's drug programs in-house. Mithril had one full-time employee, founder Cody Tranbarger. Part-time consultants, 11 outside labs and Claude did the rest. From Mithril's technical report: Mithril took an antibody for a rare childhood muscular dystrophy from idea to proof of principle in mice in under 18 months in a small early study, treated mice showed roughly twice the grip strength of untreated ones Claude read a 341-page lab report and proposed a next step for all 113 antibody constructs in under 8 minutes Claude drafted 17 emails in one week, and the team sent 12 after editing them Claude flagged 41 details nobody had confirmed in writing, so the team could check each one before sending» Mithril's work is still preclinical. Human trials are late 2028 at the earliest, and only if the drug candidates keep holding up. So Claude did the reading and the drafting, and the team made the decisions. When a team member overruled one of Claude's recommendations, Claude updated the memo, the run order and every affected draft in four minutes. If one person can run a drug program like that, a big team stops being much of an advantage. Unless the mouse results don't hold up in humans, in which case all that speed just gets Mithril to a dead end faster. How many drug programs can one person run with AI?