A former DeepMind scientist who helped Alpafold, the AI model that brought a revolution into the prediction of protein structure, has collected $ 50 million to launch latent laboratories, a startup aimed at using artificial intelligence to design new proteins.
Simon Kohl, a physicist and machine learning expert born in Germany, founded Latent Laboratories in 2023 with the ambition of “achieving computational control of biology.” The company wants to collaborate with biotech companies to speed up the development of medicines by replacing traditional experimental methods with AI-driven protein design.
The financing round, completed in December, was led by Radical Ventures, an AI-oriented investment fund, in addition to Sofinnova Partners, known for its life-sized investments. Pacific backers include: Jeff Dean chief scientist from Google, Aidan Gomez-MEDE founder of generative AI company Cohere and Mati Staniszewski-MEDE founder of AI Audio Startup Elevenlabs.
With an earlier $ 10 million in starting capital, Latent Labs has now obtained a total of $ 50 million to scale up its activities.
Kohl, 34, previously worked at DeepMind, where he played a key role in the second iteration of Alphafold – technology that is recognized with a Nobel Prize for Chemistry. Although Alpafold enables scientists to map out existing protein structures, Kohl believes that the next step is to completely redesign new proteins with the help of generative AI.
‘[AlphaFold] This allows you to look at the catalog of existing natural proteins, but you can’t make a new one, “said Kohl.
His goal with latent laboratories is to develop AI models that eliminate the need for traditional laboratory techniques, making drug discovery faster and more precise.
Latent Labs assembles a team of top AI and biotech researchers in London and San Francisco, including four former DeepMind scientists. The company actively hires more specialists as it expands.
By applying generative AI to protein and molecular design, Kohl believes that latent laboratories can reduce errors in clinical studies, an important challenge in the pharmaceutical industry.
“There are still many failures in clinical tests, and in some respects you can play the problem back to how the medicine was originally discovered,” he explained.
“If you can design proteins and molecules with a higher degree of control and precision, the idea is that we will see fewer failures in the future.”
Kohl recognized concern about the ethical implications of proteins generated by AI and called it an “important question” that requires a broader discussion. However, he emphasized that current AI technology is not yet advanced enough to surpass what human experts can achieve in a well-equipped laboratory.
Regarding when Latent Labs’ AI will perform better than traditional methods, Kohl admitted that it was difficult to predict.
“Our North Star is that our AI systems will eventually be so advanced that they no longer require experimental laboratory validation,” he said. “At that time you will have a fairly powerful technology in your hands.”
Latent Labs enters a competitive but still rising field where multiple startups are investigating how AI can reform the development of medicines. Kohl compared the landscape with a space competition.
“It feels like a good number of rockets are being launched now, and they can all reach a job,” he said.
Although latent Labs have not announced its current projects or customers, the rapid financing and controversial support from the startup are suggesting that it is an important player in the future to be an important player of AI-driven drug discovery and protein technology.