Simon Kohl

Simon Kohl founded Latent Labs in London in 2023 to apply generative AI to the design of new proteins. He came from DeepMind, where he was a senior research scientist on AlphaFold 2, the protein structure project that earned Demis Hassabis and John Jumper a Nobel Prize in Chemistry, and then started and co-led DeepMind’s protein design team. Latent Labs came out of stealth in February 2025 with $50m (£37m) in total funding, including a $40m Series A co-led by Radical Ventures and Sofinnova Partners.

From a model to an agent

In December 2025 the company released Latent-X2, a model for designing antibodies and peptides. Latent Labs says it produced hits against half of 18 deliberately difficult targets, needing between four and 24 designs for each.

In March 2026 it built an agent on top of it. Latent-Y takes a therapeutic goal written in plain language, or a scientific paper, and works through the steps a structural biologist would take, from choosing where on the target to bind to designing and checking candidates. It can run end to end or pause for a scientist to review each stage. The company reports a 67% target-level success rate across nine targets with no human filtering, and says experts using the agent completed design campaigns 56 times faster than independent estimates of the time they would otherwise take.

On 15 July Latent Labs opened Latent-Y to approved researchers worldwide, with a free quota of 250 designs a day and paid credits for larger campaigns. “We built Latent-Y to be a force multiplier for researchers,” Kohl said at the launch. “Just as coding agents are accelerating software development, Latent-Y accelerates drug development”, he added.

Compute and channels

The company runs on rented infrastructure. In June 2026 it reported more than 4x faster training and more than 60% faster inference after moving its models to NVIDIA Blackwell GPUs on Nebius’s cloud. Since May 2025 it has had a multi-year collaboration with Amazon Web Services (AWS), which it describes as one of its channels to scientists. Enterprise customers can reach the platform in a browser, through an API or as a private deployment, and the company says it does not train on user data or outputs.

Why it matters to UK buyers

For pharmaceutical and biotech research teams, Latent Labs turns frontier protein design into a service bought by the design rather than a capability each company has to build. The agent format also means one scientist can run many campaigns in parallel, which changes the staffing sums for smaller UK biotechs. Its route through AWS and private deployments puts it within reach of cloud partners already serving life sciences customers.

Since February 2025 Kohl has moved Latent Labs from a stealth research lab to a product that any approved researcher can use in a browser. The next measure is commercial: how many of those researchers become paying enterprise partners.