What Chai Discovery does
Chai Discovery (private; founded in 2024) is an AI-for-science company building a computer-aided design suite for molecules, focused primarily on antibody and related biologics design. Its core product direction is to let scientists start from biological targets (e.g., antigens, epitopes, and antibody framework/format constraints) and then use multimodal foundation-model capabilities to generate and refine candidate molecules rather than relying on traditional trial-and-error or large-scale wet-lab screening.
On its website, the company describes the suite as supporting de novo protein (and specifically de novo antibody) design workflows with controls for antibody format/framework sequence, epitope targeting, and advanced design constraints (including agonism/antagonism intent and handling of chemical modifications like glycans). The company also positions its approach as “predict and reprogram” interactions between biochemical molecules, and repeatedly links its technical progress (Chai-1/Chai-2/Chai-3) to increased success rates and reduced iteration time for partners in preclinical drug discovery.
Business model: Chai Discovery appears to sell access to its platform/models to biopharma organizations for applied R&D. Multiple official announcements describe partners leveraging Chai’s models and platform in their drug-discovery pipelines (e.g., deployment licensing / partner access), including collaborations and trial access programs.
Strategic position (as of 2026-08): the company’s fundraising and partnership cadence in 2025–2026 indicates scaling from research breakthroughs toward broader commercial deployment across large biopharma customers. In mid-2026 it announced a $400M Series C at a $3.8B valuation and described Chai-3 as materially improving target success rates and binding affinity. It also announced multiple large-industry collaborations across major pharma and immunology/biologics players (Bristol Myers Squibb, argenx, Novartis, Eli Lilly) and a license agreement with Pfizer.