Chai Discovery

Chai Discovery is an artificial intelligence biotechnology company that develops foundation models for biochemical structure prediction and molecular design. It describes its goal as building a "computer-aided design suite" for molecules: models that learn the underlying principles of biochemical structure and interaction so that scientists can design new medicines computationally rather than through iterative laboratory trial and error. Its products include Chai-1, an open-source structure prediction model released in 2024, and Chai-2, a generative model that takes an antigen structure as input and produces antibodies that bind it, supporting de novo design of custom antibodies rather than modification of existing ones.

The company positions its technology against the two prevailing antibody discovery approaches, rational design and directed evolution, arguing that generative design compresses the antibody discovery cycle from 12-24 months to four to eight weeks. Contrary Research reports Chai-2 achieved 20% experimental hit rates across 52 de novo designed targets, described as a 200x improvement over prior computational methods. Public materials also cite work on drug-like antibody design against challenging targets with atomic precision.

Chai Discovery was founded in March 2024 and is headquartered in San Francisco, California, with reporting that the company began in New York. It is privately held, at Series B stage as of April 2026, with roughly 31 employees (Startup Intros lists a range of 11-50).