Latent Labs Launches AI Protein Design Tool

Latent Labs Launches AI Model for Biological Programming

Approximately six months following its emergence from stealth mode and a $50 million funding round, Latent Labs has unveiled a web-based AI model specifically designed for programming biology.

Protein Binding Performance

According to Latent Labs CEO and founder Simon Kohl, the model has demonstrated “state-of-the-art” results in laboratory testing for protein binding. Kohl, a scientist formerly co-leading the protein design team at DeepMind’s AlphaFold, explained that “state-of-the-art,” or SOTA, signifies the highest level of performance achieved to date within the AI industry for a given task.

He further elaborated to TechCrunch that the company possesses computational methods for evaluating the quality of these designs.

A significant proportion of the proteins generated by the model are anticipated to be viable upon physical laboratory testing.

Introducing LatentX

Latent Labs’ core biology model, designated LatentX, empowers academic institutions, biotechnology startups, and pharmaceutical companies to design innovative proteins directly within a web browser.

LatentX transcends the limitations of naturally occurring proteins, enabling the creation of entirely novel molecular designs with precisely defined atomic structures.

This capability promises to accelerate the development of new therapeutic interventions.

Distinction from AlphaFold

Kohl emphasizes that the ability to design completely new proteins is the key differentiator between LatentX and AlphaFold.

“AlphaFold functions as a model for protein structure prediction,” he clarified. “It facilitates the visualization of existing structures, but does not enable the generation of new proteins.”

Business Model and Licensing

Unlike AI-driven drug discovery companies such as Xaira, Recursion, or Isomorphic Labs (a DeepMind spinout), which concentrate on developing proprietary medications, Latent Labs operates on a licensing model.

This allows external organizations to utilize their model.

Kohl noted that not all companies have the resources to independently construct their own AI models, infrastructure, and dedicated teams.

Future Plans and Availability

Currently, LatentX is available without charge.

However, the company plans to introduce fees for advanced features and functionalities as they become available.

Competition and Backing

Other organizations offering open-source AI foundational models for drug discovery include Chai Discovery and EvolutionaryScale.

Latent Labs has secured backing from investors including Radical Ventures, Sofinnova Partners, Google’s Chief Scientist Jeff Dean, Anthropic’s CEO Dario Amodei, and ElevenLabs CEO Mati Staniszewski.