The interface is a browser window, a text prompt, and a button labeled ‘Generate.’ You type a description of a protein you need,something that binds to a specific cancer marker, an enzyme that breaks down a stubborn plastic. You click. A few seconds later, a three-dimensional model appears, a digital blueprint for a molecule that has never existed. This is the promise of Latent Labs: to turn biology into a software design problem, where the lab bench is a web page and the first draft of a new drug is a few keystrokes away.
Founded in 2023 by Dr. Simon Kohl, a key architect of DeepMind’s Nobel-recognized AlphaFold2 system, Latent Labs is a frontier AI lab with a simple, staggering goal: to make biology programmable [Radical Ventures]. The company has raised $50 million in seed funding from a consortium of deep-tech investors, including 8VC, Radical Ventures, and Sofinnova Partners, alongside angels like Google’s Jeff Dean and Cohere’s Aidan Gomez [TechCrunch, Feb 2025]. Their product, Latent-X, is a generative foundation model for de novo protein and antibody design, a platform that aims to let scientists conjure therapeutic candidates and industrial enzymes directly from a computer [Radical Ventures].
From Protein Prediction to Protein Design
AlphaFold2 solved a fundamental problem in biology: predicting the three-dimensional structure of a protein from its amino acid sequence. Latent Labs is attempting the inverse, and arguably harder, task. Starting from a desired function or structure, its models generate a sequence that will fold into a viable, stable protein. This shift from prediction to generation is the core of the company’s technical wedge. It is the difference between reading a map and drawing a new one for a destination that does not yet exist.
Kohl’s pedigree is the company’s initial credential. His work at DeepMind placed him at the epicenter of the computational biology revolution. The team he has assembled in London includes AI researchers and computational biologists from DeepMind, Google, Microsoft, and leading academic institutions, a concentration of talent that is itself a form of early traction [Radical Ventures]. The company’s advisory board includes a former Merck CEO, signaling an intent to bridge the gap between frontier research and pharmaceutical pipelines [justainews.com, 2026].
The Platform as a Partnership Engine
Latent Labs is not building its own drug pipeline. Instead, it is adopting a platform-and-partnership model, licensing its generative AI capabilities to biopharma and industrial biology companies [TechCrunch, Feb 2025]. The go-to-market strategy hinges on two surfaces: a web-based platform for interactive design and an API for deeper integration into a partner’s own discovery workflows [Latent Labs].
The company claims its technology offers not just novelty, but dramatic efficiency. Internal benchmarks suggest Latent-X can generate designs over 10 times faster than prior methods like RFdiffusion, and that its Latent-Y tool allows a single researcher to run design campaigns 56 times faster than expert estimates [Latent Labs] [decodingbio.substack.com, 2026]. For an industry where bringing a single drug to market can cost billions and take over a decade, even a fractional improvement in early-stage speed and success rates represents an enormous economic incentive.
The Competitive and Commercial Landscape
The field of computational drug design is crowded with well-funded players, from public companies like Exscientia to private rivals like Generate:Biomedicines and Cradle Bio. Latent Labs enters this space with a specific focus on foundational models for de novo generation across both health and sustainability applications, a potentially broader remit than some therapeutics-only competitors.
The company’s early risks are less about technical plausibility and more about commercial translation and focus.
- The partnership puzzle. No specific, named biopharma partners have been publicly disclosed. The model depends on convincing large, conservative organizations to adopt a radically new, external discovery engine. A single flagship partnership would be a powerful proof point.
- The generalist gamble. Pursuing both drug candidates and industrial enzymes could spread resources thin. Success in either domain requires deep, specific domain knowledge and go-to-market motions that are not easily shared.
- The validation timeline. Computational designs must ultimately be synthesized and tested in physical labs. The true measure of Latent-X’s value will be the lab-viability and functional efficacy of the proteins it generates, a feedback loop that takes time to close.
The company’s answer appears to be depth disguised as breadth. By building foundation models that capture the ‘fundamentals of biology,’ the bet is that a single powerful engine can be fine-tuned for diverse applications, from antibody design to enzyme optimization [CB Insights].
The Next Twelve Months
For a company that launched with $50 million, the coming year is about moving from technical promise to commercial evidence. The milestones to watch are concrete and customer-shaped.
Pre-seed (2024) | 10 | M USD
Seed (2025) | 40 | M USD
The funding provides a long runway for a team estimated at 11-50 employees [RivalSense]. The open roles for Computational Protein Designers and Software Engineers point to continued investment in core model development [Latent Labs]. The critical near-term signal will be the announcement of a first major pharmaceutical or biotech partner. Furthermore, any published, peer-reviewed validation of a Latent-designed protein in a wet-lab setting would serve as a powerful credential for the platform’s real-world utility.
Latent Labs sits at the intersection of two profound trends: the industrialization of AI and the digitization of biology. Its platform is an attempt to build the definitive tool for the second trend using the methodologies of the first. The cultural question it implicitly answers is one of agency and acceleration. For centuries, biological discovery has been a process of observation, hypothesis, and laborious iteration. Latent Labs is betting that the next century will be defined by intention, generation, and speed. It is offering a browser for the code of life, and waiting to see what the world decides to build.
Sources
- [TechCrunch, Feb 2025] Founded by DeepMind alumnus, Latent Labs launches with $50M to make biology programmable | https://techcrunch.com/2025/02/12/founded-by-deepmind-alumnus-latent-labs-launches-with-50m-to-make-biology-programmable/
- [Radical Ventures] Latent Labs portfolio page | https://radical.vc/portfolio/latent-labs/
- [Latent Labs] Company website and platform pages | https://www.latentlabs.com
- [decodingbio.substack.com, 2026] Article on Latent-X performance | https://decodingbio.substack.com
- [justainews.com, 2026] Article on Latent-X2 and advisory board | https://justainews.com
- [CB Insights] Company profile | https://www.cbinsights.com/company/latent-labs-technologies
- [RivalSense] Company intelligence | https://rivalsense.co/intel/latent-labs/