Early progress on accelerating drug development with Claude

Anthropic argues that AI could accelerate drug development by assisting across the entire process, not just inventing molecules: identifying targets, improving drug candidates, supporting manufacturing and regulatory work, and planning clinical trials. To discover where Claude helps and where it falls short, the company has partnered with six biotechnology and pharmaceutical organizations and begun running drug programs itself. The reported projects span cancer, food allergies, rare diseases, manufacturing, and company operations; most remain preclinical, so the results are early and do not establish that any candidate will become an effective medicine.

The examples include a redesigned genetic component for CAR T therapy that performed better in laboratory tests; tens of thousands of antibody designs, with early candidates that detached allergy-related IgE from immune cells; and improvements to two antibody-fragment leads, including stronger binding and greater stability. At Enveda, Claude identified issues in manufacturing records and cut review time by 89%, while drafting regulatory documents much faster than manual work. Other efforts apply Claude to clinical-trial decisions and to running a small drug-development company. Anthropic also describes a muscular-dystrophy program that reached promising mouse results in under 18 months, with human trials potentially possible in late 2028. The company plans to expand partnerships and provide research tools and access, while acknowledging that clinical validation remains ahead.