Anthropic enters healthcare to accelerate drug discovery with Claude

Entering the healthcare race

Anthropic has officially moved into the healthcare sector to assist in the development of new medicines. The company announced this shift on October 27 by revealing new research partnerships with six organizations, including Bristol Myers Squibb and BigHat Biosciences. These moves signal a transition for the AI provider from general tasks like coding and email writing to specialized scientific applications.

A major part of this strategy involves the acquisition of Mithril Medicine, a unique one-person biotech startup. The company is run entirely by its founder, Cody Tranbarger, who uses Claude as an AI assistant to manage the complexities of a biotech firm. By leveraging the AI, Tranbarger coordinates contracts, logistics, and experiments across 11 different external laboratories.

The goal of this partnership is to find treatments for two rare forms of muscular dystrophy. Using Claude to do the heavy lifting, the startup advanced a new drug candidate from a basic concept to successful testing in mice in less than 18 months. If these early results remain consistent, the company hopes to begin human clinical trials as early as late 2028.

Cutting months of work into hours

One of the most significant impacts of Claude in this space is its ability to process massive amounts of technical documentation. As India Today reported, the biotech firm Enveda used the AI to review a manufacturing document that was 1,061 pages long. The AI successfully flagged all 11 errors that human reviewers found, while also identifying seven additional issues that the humans completely missed.

This level of accuracy led to a massive increase in efficiency for the verification process. The time required to complete the review dropped by 89 per cent, falling from over 25 hours to under three hours. By automating the most tedious parts of the verification, the AI allows scientists to focus on higher-level tasks while maintaining a higher standard of oversight.

The AI is also being used to navigate the dense world of pharmaceutical regulations. For example, Claude drafted a complex regulatory document in just four hours, a task that normally takes about three months to complete by hand. Additionally, Bristol Myers Squibb uses the technology to evaluate clinical trial plans, generating initial success probability analyses from a single prompt instead of taking several days.

Optimizing drug and cell designs

Beyond administrative tasks, Claude is being used to physically design the structures of new medicines. BigHat Biosciences utilized the AI to optimize antibody fragments, which are parts of the immune system used to fight disease. The resulting designs bind to their targets over 10 times better than previous versions while showing significant improvements in overall stability.

Other companies are using the AI to generate high volumes of potential treatments in very short windows of time. Divergence Labs used Claude to generate approximately 35,000 different antibody designs for food allergy treatments in only one month. This allows researchers to screen a much wider variety of possibilities than would be possible through manual design alone.

The AI is also making strides in cell therapy design, which involves modifying cells to treat diseases. Generate Biomedicines relied on Claude to propose gene designs that were 70 per cent shorter than the standard models currently in use. These types of optimizations help scientists create more efficient and potentially safer biological treatments for patients.

Tools for the scientific community

To support these advancements, Anthropic has launched a dedicated workbench called Claude Science. This platform is built specifically for scientific researchers to help them navigate the complexities of drug discovery and laboratory research. Alongside this tool, the company introduced a Life Sciences Verification Programme to provide a framework for these new applications.

While these results are impressive, Anthropic has been clear about the current limitations of the technology. All of the achievements mentioned so far are at the preclinical stage, meaning they have only been tested in lab environments or on animals. The data comes directly from the participating companies' own experiments and has not yet been proven safe or effective in human clinical trials.

The move by Anthropic follows a broader trend in the technology industry toward medical AI. Just one week before Anthropic's announcement, rival OpenAI revealed it had acquired Orchestra Bio. This suggests that the world's largest tech companies view the intersection of artificial intelligence and medicine as the next major frontier for innovation.