Yoneda Labs, the Y Combinator startup building a foundation model for chemists working in drug discovery, today announced it has raised $4 million in seed capital from Khosla Ventures, 500 Emerging Europe, 468 Capital, Fellows Fund, and Y Combinator.

More than 400,000 chemists are working on chemical manufacturing around the world. While scientists have begun to leverage AI in the discovery and design of drugs, the chemists whose job it is to synthesize the drugs are forced to run trials in wet labs without much, if any, automation, simulation tools, or other computational support. As a result, the pharmaceutical and chemical manufacturing industries waste billions of dollars on failed chemical experiments each year.

“When a chemist wants to couple two molecules together, they are left to old-school literature search and trial-and-error methods in a lab,” said Michal Mgeladze-Arciuch, founder and CEO of Yoneda Labs. “At Yoneda, our vision is to build a foundation AI model that analyzes and predicts what will happen before a chemist has to run their experiment. This would increase their productivity by an order of magnitude and potentially enable the creation of new drugs that isn’t currently possible to do cost-effectively at scale.”

"Machine learning and generative AI models have already begun to accelerate physics-oriented fields like aerospace engineering,” said Jon Chu, partner at Khosla Ventures. “Chemistry will be no different and the team at Yoneda Labs has a novel approach to creating a foundation model for chemistry that could change the way chemicals are manufactured and improve the drug discovery process."

Founding Team

Yoneda Labs was founded by a group from the University of Cambridge with backgrounds in chemistry, robotics, and machine learning. The idea for Yoneda Labs was born when one of the founders, Jan Oboril, was working on drug development at a major pharmaceutical company. He felt he wasted hundreds of hours running trials in the lab that could be replaced with machine learning algorithms that could run the same experiments in silico.

After joining forces with co-founders Michal Mgeladze-Arciuch, who worked on machine learning algorithms at Jane Street and researched large AI models at UC Berkeley, and Daniel Vlasits, who won an international robotics competition, the team built an initial prototype that improved Jan’s process in the lab and was accepted to Y Combinator.

During their time at YC, the team identified 20,000 chemical reactions to generate proprietary data for their foundation model. In a small-scale trial, the models suggested good conditions in 95% of cases, a stark improvement to the industry standard in which the majority of the experiments fail.

With this funding, they plan to build out a robotics lab capable of running experiments on the remaining 20,000 reactions to continue training the first-ever generalizable AI model for chemists. Later this year, they plan to be able to run and analyze 200 experiments a day, the equivalent output of roughly 20 full-time chemists.

About Yoneda Labs

Yoneda Labs is building the world’s first foundation AI model for chemical manufacturing. The company was founded by Michal Mgeladze-Arciuch, Jan Oboril, and Daniel Vlasits, who met at the University of Cambridge and identified an opportunity to bring machine learning to outdated pharmaceutical practices.

View source version on businesswire.com: https://www.businesswire.com/news/home/20240425073708/en/