Chai Discovery Inc., a startup using artificial intelligence to accelerate drug development, today announced that it has closed a $130 million funding round led by Oak HC/FT and General Catalyst.

The two investment firms were joined by more than a half dozen other backers. The consortium included OpenAI Group PBC and Thrive Capital, one of the ChatGPT developer’s largest investors. Chai Discovery is now valued at $1.3 billion.

The investment comes about six months after the company introduced Chai-2, a multimodal AI model optimized for medical research. It’s designed to help scientists develop new drugs faster by automating some of the manual work involved in the process. The algorithm can provide a “100-fold improvement over previous computational methods” in certain areas, the company claims.

Many drugs are based on molecules known as mAbs antibodies. Those are large proteins that attach themselves to one specific section of a disease-causing virus or bacterium and neutralize it. According to Chai Discovery, competitors’ AI models often don’t design entire proteins but only protein fragments, which have limited usefulness. Chai-2, in contrast, can generate full-length mAbs antibodies. 

The company says the model can also generate other molecules with potential therapeutic applications. It’s capable of designing nanobodies, which are antibodies derived from llamas and alpacas, along with miniproteins. Those are proteins that are a fraction of the size of mAbs antibodies.

An antibody can only neutralize disease-causing viruses or bacteria after it successfully binds to their surface. As a result, so-called binding affinity is one of the main factors scientists evaluate when developing new drugs. However, it’s not the only consideration: Researchers also weigh details such as an antibody’s manufacturability, safety and stability. Chai says that its AI can account for those factors more effectively than other models.

One of the contributors to Chai-2’s performance is that it evaluates antibodies with a high degree of granularity. According to Chai Discovery, the model “reasons about structures at sub-angstrom scale.” An angstrom is a unit of measurement that corresponds to one ten-billionth of a meter, which is slightly more than the length of a hydrogen atom.

The company tested Chai-2’s capabilities by having it design a set of antibodies with potential therapeutic applications. From there, it evaluated the antibodies across four criteria that determine whether a molecule could be turned into medicine. It says 89% of Chai-2’s suggested designs contained either a single issue or none.

“What looked like five-year problems just months ago are now getting solved in weeks,” said Chai Discovery founder and Chief Executive Officer Josh Meier. “Our latest models can design molecules that have properties we’d want from actual drugs, and tackle challenging targets that have been out of reach.”

The company will use its newly raised capital to support its research and commercialization efforts.

Image: Chai Discovery