Latent Labs’ cover photo
Latent Labs

Latent Labs

Biotechnology Research

Frontier models for all molecules of life.

About us

We are making biology programmable to transform health and sustainability for the benefit of all. We are a frontier AI lab building foundational models that capture the fundamentals of biology. We empower partners with breakthrough generative AI to gain unprecedented control over molecular biology, creating new antibodies, optimizing existing enzymes, and advancing genetic engineering. Contact us to work with us today.

Website
https://www.latentlabs.com/
Industry
Biotechnology Research
Company size
11-50 employees
Type
Privately Held

Employees at Latent Labs

Updates

  • Latent Labs reposted this

    For Latent Labs, the shift in antibody discovery is no longer just about finding better binders faster. It is about replacing iterative screening with generative design that starts closer to a drug-like candidate. In his talk, Simon Kohl argued that the field is moving from lab-based discovery toward prompt-driven molecular design. He framed Latent-X2 as a frontier model spanning peptides, cyclic peptides, minibinders, and drug-like antibodies. The strongest claim from the talk was not only binding performance, but also developability: Kohl said the model’s designed antibody formats were already developable 47% of the time without optimization, based on measures including thermal stability, monomericity, yield, polyreactivity, and hydrophobicity. He also described Latent-Y, a protein design agent intended to turn that modeling capability into a working design workflow. In the example shown on stage, the agent moved from a text prompt to lab-validated nanobody recommendations, compressing work that would otherwise take weeks of computation into an afternoon. The broader case was clear: antibody design is starting to look less like library search and more like programmable biology. #SynBioBeta #SyntheticBiology #Biotech

    • No alternative text description for this image
    • No alternative text description for this image
  • We'll be at SynBioBeta next week, with a keynote from Simon!

    I'll be speaking at SynBioBeta twice this year. Keynote, May 5th: 'Designed, Not Discovered: From Prompt to Drug-Like Antibody'. From Latent-X2 designing drug-like antibodies zero-shot, to Latent-Y running lab validated agentic design campaigns from a text prompt. May 6th: 'Programmable Molecules: AI and the Rise of Context-Aware Therapeutics', with Ashoka Madduri (Sanodi) and Jacob Becraft (Strand). See you in San Jose. Full programme: https://lnkd.in/gMSBizcg

    • No alternative text description for this image
  • Latent Labs reposted this

    Amazing afternoon with Jensen Huang and his leadership team yesterday as they hosted Radical Ventures and several of our portfolio companies at NVIDIA HQ. Spending hours discussing the future of AI with those building it is so inspiring. Thank you to Jensen and the entire Nvidia leadership team for your time, insights, and genuine support of the founders who are defining the AI frontier. A truly memorable day with a partner who shares our vision for an extraordinary future.  DatologyAI, Firsthand, Ribbon, Latent Labs, P-1 AI, Prime Intellect, Reka AI

    • No alternative text description for this image
  • Genetic Engineering & Biotechnology News is covering the launch of Latent-Y in their latest piece on how agents are starting to automate scientific discovery. Latent-Y designs antibodies from a text prompt, achieving a 67% target-level success rate without human filtering or intervention. As Simon Kohl puts it: a "force multiplier" that completes design campaigns 56-time faster. Thank you Fay Lin, for the article! Read it here: https://lnkd.in/eY8SeUZu

    Have we reached the agentic AI inflection? In response, many point to OpenClaw, which went from a solo developer’s side project to one of the fastest-growing open-source frameworks in history. More researchers are customizing agents for domain-specific use. While at NVIDIA GTC, I compiled a list of autonomous systems augmenting the life science lab, including: ▪️ 𝐊𝐨𝐬𝐦𝐨𝐬 (Edison Scientific): AI scientist that performs hundreds of research tasks in parallel to compress months of work into a single day ▪️ 𝐋𝐚𝐛𝐎𝐒 (Le Cong and Mengdi Wang): Extended reality (XR) operating system that unites computational reasoning with physical experiments ▪️ 𝐋𝐚𝐭𝐞𝐧𝐭-𝐘 (Latent Labs): Agent that designs therapeutic antibodies from a text prompt ▪️ 𝐃𝐲𝐧𝐨 𝐏𝐬𝐢-𝐏𝐡𝐢 (Dyno Therapeutics): Agent that combines protein binder design with data-informed filters to identify candidates most likely to succeed By no means is this list exhaustive. Additional players to watch include, Lila Sciences, Potato, Phylo, and more (comment anyone I'm missing!) Andrew Beam, Marinka Zitnik, Rory Kelleher, Andrew White, Samuel G. Rodriques, Simon Kohl, Sam Sinai, Eric Kelsic, Nick Edwards, PhD, Kexin Huang Read my full round-up at GEN (Genetic Engineering & Biotechnology News): https://lnkd.in/gsu4RMXP

  • View organization page for Latent Labs

    10,277 followers

    Today we're launching Latent-Y: the world's first autonomous agent for drug design, lab-validated end to end. Give it a research goal. Latent-Y reasons like a PhD drug design expert, reads the literature, identifies the target, designs candidates, and iterates until it hits. We validated Latent-Y across three campaign types and nine targets, achieving lab-confirmed nanobody binders with a 67% target-level success rate and single-digit nanomolar affinities confirmed by SPR. Every confirmed binder is a novel molecule. In user studies, experts working with Latent-Y completed design campaigns 56 times faster than independent expert time estimates. When the agent runs campaigns in parallel, those gains compound further. Our own protein design team has switched to Latent-Y for their campaigns. Latent-Y is the first system of its kind, and every campaign produces a full reasoning trace so researchers can audit, steer, and intervene at any point. Two highlights from our validation: in our cross-species campaign, Latent-Y extended its own capabilities from a high-level prompt, writing custom generative code to solve a problem it had never been explicitly designed for, producing lab-confirmed binders against both human and cynomolgus targets. In a separate campaign seeded entirely by a scientific publication, it inferred the mechanism of action, identified the target and epitope, and designed confirmed binders, mirroring how scientists work in practice. We regard this as a first step. Latent-Y already runs at scale, and we are building toward a scientific engine that closes the loop with automated laboratories and reaches discoveries no single team could achieve alone. Blog post: https://lnkd.in/eZU_PJZH Technical report: https://lnkd.in/eEsHGbW4  Apply for access at platform.latentlabs.com or reach out at partnerships@latentlabs.com.

  • Nature Biotechnology put out a review article on the rapid progress in de novo antibody design today. It's the magazine's anniversary issue too, good way to celebrate! Thanks for the thoughtful work going into this, Lisa Melton and Andrew Marshall. More soon.

    View profile for Simon Kohl

    I've spent years working toward a simple but seemingly impossible goal: designing biologics the way engineers design chips, from first principles, before anything is fabricated. Today, Nature Biotechnology ran an overview of how close the field is getting. As founder of Latent Labs, I'm featured alongside our advisor Stefan Oschmann (former CEO at Merck Group)— and some of the other key players driving this shift — discussing what's actually behind the leap. Two things came together: algorithmic advances in sequence–structure modeling, where foundation models learn chemistry broadly rather than antibody-specifically, and wet-lab loops that continuously stress-test and inform the models. The question has shifted from "Can you do it?" to "When will it matter clinically?". We intend to answer that with data, including what the same model can do beyond antibodies, across biologics and on hard targets, like cyclic peptides against KRAS(G12D). 🔗 Nature Biotechnology, March 2026: https://lnkd.in/eCEeugME

  • View organization page for Latent Labs

    10,277 followers

    Today we announce results from our deployment of NVIDIA Blackwell B300 GPUs in partnership with NVIDIA and Nebius, a leading AI cloud provider for training and inference, achieving over 4x training speedups and more than 60% faster inference for our generative models powering drug discovery. "These performance gains aren't just benchmarks, they translate directly into faster research cycles and shorter paths to therapeutic candidates for our partners," - Simon Kohl, CEO at Latent Labs. “Latent Labs moves with remarkable focus and speed, and we’ve been especially impressed by the accuracy and performance of their models in production-grade research workflows. It’s exactly the kind of company we’re excited to support on Nebius AI Cloud,” - Ilya Burkov, Global Head of Healthcare and LifeSciences at Nebius. Read the full release: https://lnkd.in/e9bQjyAZ

  • Our CEO Simon Kohl will be speaking at NVIDIA GTC as part of a panel on 'Beyond the Model: Driving Cutting-Edge Discovery'. The panel includes David Ruau (Nvidia), Eric Durand (Owkin) and Robin Roehm (Apheris) and discusses the scale of AI deployments in the industry today and what's coming next. More to come soon. Tune in here, next Wednesday, March 18: https://lnkd.in/edJP_JB2

    Joining NVIDIA GTC next week for a panel where research meets reality: what does it take to deploy AI that doesn't just assist drug discovery, but acts as a true force multiplier for the researchers driving it? Latent-X2 can design drug-like antibodies and macrocyclic peptides zero-shot, with developability and low immunogenicity from the first attempt. That changes what's possible at the bench. The next frontier is bringing that to real deployments at scale, where researchers can pursue more targets, more strategies, and more ambitious science in parallel. That's the conversation I'm looking forward to having with David Ruau, Eric Durand, and Robin Roehm, how real-world deployments are evolving, what it means to give research teams the infrastructure to move at a different pace, and where the field is heading next. There's more to share on this very soon. I'll be in San Jose and happy to connect if you're around. Register here: https://lnkd.in/edJP_JB2

    • No alternative text description for this image
  • Simon Kohl will speak to how AI is collapsing timelines and increasing scale in drug discovery, this Thursday at the Science Museum in London.

    "What if biology were programmable?" Next Thursday, Simon Kohl is in our Speakers' Corner to share his 15 minute vision for the future. About Simon: - Co-developed AlphaFold (building its widely used uncertainty prediction, 'PLDDT') - Co-led team Google DeepMind's Protein Design team. - Founder & CEO of Latent Labs, a Pillar VC portfolio company. .... and an Encode Fellowship Advisor. This event is part of Pillar VC's AI for Science Series: 3 cities, 12 speakers, 450+ builders. Join us in London as we bring together founders, researchers, theorists, field-builders, and funders - all building at the intersection of AI and science. 🔗 https://lnkd.in/ezV2FduT

    • No alternative text description for this image

Similar pages

Funding