Anthropic speeds up drug discovery as Claude spots issues humans miss, cuts review time by 89%

From writing emails and code to designing real medicines, Anthropic has officially entered the race to transform healthcare with AI.

In Short

• Anthropic acquired Mithril Medicine and announced partnerships with six research organisations.
• Claude AI is saving scientists months of work by quickly reviewing and drafting complex medical documents.
• Early lab tests show the AI can design highly effective medicines that could reach human trials by 2028.

Anthropic, the company behind the popular Claude AI chatbot, is moving beyond writing emails and coding. On October 27, the AI giant announced it is stepping directly into the healthcare sector to help develop new medicines. To kickstart this mission, Anthropic acquired Mithril Medicine, a unique one-person biotech startup, and revealed new joint research partnerships with six organizations, including BigHat Biosciences and Bristol Myers Squibb.

The one-man biotech army

Mithril Medicine is run entirely by a single employee, founder Cody Tranbarger. By using Claude as his AI assistant, Tranbarger coordinated contracts, logistics, and experiments across 11 external laboratories. His startup focuses on finding treatments for two rare forms of muscular dystrophy.

With Claude doing the heavy lifting, the company advanced a new drug candidate from just a concept to successful mice testing in under 18 months. If these promising early results hold up, human clinical trials could begin as early as late 2028.

Cutting months of work down to hours

The most striking data from the announcement shows how much time AI is saving scientists. Enveda, a biotech firm, used Claude to review a massive 1,061-page manufacturing document. The AI easily flagged all 11 errors that human reviewers had found, and discovered seven extra issues that humans had completely missed.

This slashed the verification work by 89 per cent, dropping the time required from over 25 hours to under three hours. In another major win, Claude drafted a complex regulatory document in just four hours, a tedious task that manually takes about three months.

Pharmaceutical giant Bristol Myers Squibb also used the technology to evaluate its early-stage clinical trial plans. According to the company, estimating the success probability of a medical trial usually takes several days, but Claude could generate an initial analysis from a single prompt.

Designing smarter, stronger treatments

Claude is also actively helping scientists design the drugs themselves. BigHat Biosciences used Anthropic’s AI to optimise antibody fragments, creating designs that bind to their targets over 10 times better while showing major improvements in stability.

Divergence Labs used the AI to generate around 35,000 antibody designs for food allergy treatments in just a month. Meanwhile, Generate Biomedicines relied on Claude for cell therapy design, resulting in proposed gene designs that were 70 per cent shorter than standard models.

The road ahead for AI in healthcare

To support this growing sector, Anthropic has launched "Claude Science," a dedicated AI workbench built specifically for scientific researchers, alongside a Life Sciences Verification Programme.

However, Anthropic clearly notes that all these achievements are currently at the preclinical stage. The data comes directly from the companies' own lab experiments and has not yet been proven safe or effective in human trials.

Still, the race to merge artificial intelligence with medicine is accelerating fast. A week before Anthropic’s announcement, on October 20, 2026, rival OpenAI revealed it had acquired Orchestra Bio, proving that the world's biggest tech companies see drug discovery as the next major frontier for AI.

- Ends