# Latent-X2 & Latent-Y by Latent Labs

Canonical page: https://technooptimist.io/ai/latent-x
Frontier: [AI](https://technooptimist.io/ai.md) · Field: Biology models · Company: [Latent Labs](https://technooptimist.io/companies/latent-labs.md)
Stage: Pilot (3 of 5: Research → Proto → Pilot → Deploy → Scaling)
Updated: last update 2026-03-23, facts checked 2026-10-11
Goal: Design drug-ready antibodies for any target on the first try, turning months of lab screening into days, so more diseases get candidate medicines faster and cheaper.
Status: Latent-Y agent live with selected partners since March 2026; designs not yet in clinical trials.
Next milestone: Partner antibodies reach clinical trials

## Milestones

| Status | Milestone | Date | Slip |
| --- | --- | --- | --- |
| Done | Launch with $50M | 12 Feb 2025 |   |
| Done | Latent-X: binder design in a browser | 22 Jul 2025 |   |
| Done | Latent-X2: drug-like antibodies zero-shot | 16 Dec 2025 |   |
| Done | Latent-Y design agent for partners | 23 Mar 2026 |   |
| Planned | Partner antibodies reach clinical trials |   |   |

## Most important updates

- **23 Mar 2026**: Latent Labs launches Latent-Y, an agent that runs antibody design campaigns. Latent-Y takes a written goal, plan or paper, picks binding sites on the target, designs antibodies with Latent-X2, checks them computationally and iterates, logging its reasoning. ([our story](https://technooptimist.io/stories/latent-labs-launches-latent-y-an-agent-that-runs-antibody-design-2349289e.md) · [original source](https://www.venturebeat.com/business/latent-y-the-autonomous-ai-agent-for-drug-design-at-scale))
- **16 Dec 2025**: Latent Labs unveils Latent-X2, generating drug-like antibodies without optimisation rounds. Latent-X2 designs antibodies and peptides meant to be drug-ready on the first pass. ([our story](https://technooptimist.io/stories/latent-labs-unveils-latent-x2-generating-drug-like-antibodies-without-483034f8.md) · [original source](https://www.biopharmatrend.com/artificial-intelligence/latent-labs-unveils-ai-model-generating-antibodies-with-low-immunogenicity-and-drug-like-properties-1434/))
- **22 Jul 2025**: Latent Labs launches Latent-X, designing protein binders in the browser. Latent-X generates macrocycles and small protein binders atom by atom for targets it has never seen, through a no-code web platform with a free tier. ([our story](https://technooptimist.io/stories/latent-labs-launches-latent-x-designing-protein-binders-in-the-browser-1073485c.md) · [original source](https://sofinnovapartners.com/news/introducing-latent-x-a-frontier-generative-ai-model-for-protein-binder-design-accessible-via-no-code-platform-for-push-button-protein-design))
- **12 Feb 2025**: Latent Labs, founded by AlphaFold 2 alumnus Simon Kohl, launches with $50M. Latent Labs came out of stealth with $50 million, a $10M seed plus a $40M Series A co-led by Radical Ventures and Sofinnova Partners. ([our story](https://technooptimist.io/stories/latent-labs-founded-by-alphafold-2-alumnus-simon-kohl-launches-with-50m-e0a97c5b.md) · [original source](https://techcrunch.com/2025/02/12/founded-by-deepmind-alumnus-latent-labs-launches-with-50m-to-make-biology-programmable))

## Current obstacles

- **Binding is not a drug**: A good binder must also be safe, stable, manufacturable and work in animals and people, each a separate hurdle.
- **Hard targets still fail**: Latent-X2 found binders for half of 18 hard targets; the other half yielded nothing with the designs tested.

## Physics limits

- **Proteins move**: Models design against mostly static structures, but real proteins flex and change shape, which can break a predicted fit.
- **Biology must be tested**: Immune reaction and behaviour in the body cannot be fully predicted from structure; animal and human trials remain necessary.

## How it works

### Target: Pick the disease protein

A scientist names a target, such as a receptor on a cancer cell, or Latent-Y extracts it from a paper or goal.

### Generate: Write binder and shape together

The model generates sequence and 3D structure at once, shaped to grip a chosen spot on the target.

### Filter: Check in the computer first

Designs are scored for binding and for drug-like traits such as stability and low immune reaction before any lab work.

### Test: Few designs per target

Only a handful to dozens of designs per target go to the lab, instead of screening millions of random variants.

## Spec sheet

| Spec | Latent-X2 & Latent-Y | Provenance |
| --- | --- | --- |
| Latent-X macrocycle hit rate | 91-100% (7 targets) | reported |
| Latent-X mini-binder hit rate | 10-64% | reported |
| Latent-X2 targets with binders | 9 of 18, from 4-24 designs each | reported |
| Latent-Y target-level success | 67% (3 lab campaigns) | reported |
| Expert speed-up with Latent-Y | 56x (user study) | reported |

Provenance: reported = stated by the company; estimated = our estimate; sample = a sample figure.

## Papers and demos

- **Jul 2025** (paper): [Latent-X: an atom-level frontier model for de novo protein binder design](https://arxiv.org/abs/2507.19375). Technical report with lab hit rates and head-to-head tests vs AlphaProteo and RFdiffusion.

## About Latent Labs

Latent Labs builds AI models that design new proteins, binders and antibodies from scratch for drug makers. Founded by Simon Kohl, a member of DeepMind's AlphaFold 2 team.

- AlphaFold predicts the shape of existing proteins; Latent Labs designs new ones that bind a chosen target.
- Latent-Y (March 2026) is an agent that runs an antibody design campaign from a written goal to lab-ready designs.
- Launched with $50M (Radical Ventures, Sofinnova); models run in London, wet lab in San Francisco.

- Founded: 2023
- Headquarters: GB, US
- Status: private
- Raised: $50M
- Last round: Seed + Series A $50M, 12 Feb 2025, led by Radical Ventures, Sofinnova Partners
- Website: https://www.latentlabs.com
- People: Simon Kohl (Founder & CEO)
- Partners, customers, investors: Radical Ventures (investor); Sofinnova Partners (investor); GenScript (partner)

Company page: https://technooptimist.io/companies/latent-labs.md

## Related programs

Everything in AI: https://technooptimist.io/ai.md

---

The Techno Optimist tracks every frontier of technology, program by program: https://technooptimist.io/ (a Markdown version of any page: add .md to its URL; site map for agents: https://technooptimist.io/llms.txt).
Data API (JSON, realtime stream): https://technooptimist.io/developers. People read 5 pages free, then Reader $10/month; these Markdown pages are open to agents.