Anthropic’s life sciences team is systematically moving beyond general-purpose text generation and embedding its models directly into the capital-intensive pipelines of drug discovery. Six organizations developing new medicines, including Bristol Myers Squibb and BigHat Biosciences, have engaged in deep research partnerships with Divergence Labs, Enveda, Generate Biomedicines, and The Column Group to connect Claude to active R&D infrastructure.
Outsourcing primary molecule design to language models is no longer an isolated experiment in computational biology. It represents a structural shift in how pharmaceutical pipelines operate under strict capital constraints. As Anthropic deploys its models across target identification, manufacturing optimization, and regulatory filing preparation, the underlying economics of drug development face a fundamental re-engineering.
Integrating Models into Core Operations
Anthropic’s life sciences team has engaged in deep research partnerships with six organizations developing new medicines: BigHat Biosciences, Bristol Myers Squibb, Divergence Labs, Enveda, Generate Biomedicines, and The Column Group. On 25 September 2026, in our archive report, Anthropic revealed that its Claude model now autonomously handles up to 26 percent of the research and development tasks for its next-generation models. The current expansion moves language systems past isolated ideation into active, multi-stage pharmaceutical workflows.
Generate Biomedicines deployed six scientists using Claude Code to propose new designs for a specific segment of genetic instructions for CAR T-cell therapy. One design produced using Claude Code was about 70 percent shorter than the standard version of the CAR T-cell therapy genetic instructions.
Meanwhile, Divergence Labs is developing drugs designed to detach IgE from immune cells entirely rather than solely neutralizing the IgE circulating in blood. Claude converted Divergence Labs’ project goals into a detailed design plan and generated roughly 35,000 antibody designs in about a month largely autonomously. Divergence Labs found hits for both arms in early tests, including three designs that could bind to IgE and 20 that could dislodge it.
Scaling Autonomous Drug Discovery
BigHat Biosciences and an Anthropic researcher worked with Claude to improve three properties in two leads. Claude, connected to BigHat’s lab data system Reccy, proposed changes to each molecule, and BigHat’s lab made and tested about 1,010 variants in about 10 weeks.
Anthropic has also initiated proprietary drug discovery efforts of its own, including programs targeting rare childhood forms of muscular dystrophy where Claude handles much of the scientific and operational workload. This dual approach of partnering with established market players while building proprietary pipelines tests whether language architectures can resolve distributed operational bottlenecks across the entire lifecycle of a therapeutic asset.
Anthropic plans to scale this work up substantially from here, forcing traditional R&D finance teams to evaluate whether autonomous iteration cycles can offset the immense costs of clinical trial attrition.
What’s behind it
Early drug discovery cycles and capital burn shrink dramatically when a language model delivers tens of thousands of antibody variants in roughly a month instead of years of wet-lab screening. While in early July the platform integrated over 60 skills in genomics and chemoinformatics, six therapeutics organizations are now embedding the system directly into core candidate engineering pipelines. Will the 35,000 antibody candidates generated for Divergence Labs genuinely lower attrition costs in clinical trials, or will therapeutic economics remain constrained by physical wet-lab bottlenecks?
Our stories: Claude Science: Anthropic’s Vertical AI for R&D (July 1); Anthropic Deploys 30,000 Autonomous Agents for Claude R&D (September 25).