Adaptyv Bio Raises $40M to Automate the Lab That Validates AI-Designed Proteins

An AI model can now design a new protein sequence in an afternoon. The harder part comes next: proving that design actually folds correctly and binds its target still runs through slow, fragmented wet labs. It’s the reason AI drug discovery keeps hitting the same wall — design got fast, validation didn’t.

Adaptyv Bio, based in Lausanne, Switzerland, is building automation to remove that bottleneck. Customers submit a protein design as a digital sequence — through a web platform or API — and Adaptyv makes the DNA in-house, expresses the protein cell-free, measures binding via surface plasmon resonance (SPR), and returns real, QC’d experimental data.

The company said it raised a $40 million Series A led by Highland Europe, with ACE Ventures — which led its $8 million Seed round in November 2024 — doubling down, alongside byFounders and Y Combinator.

Co-founders Julian Englert (CEO) and Daniel Nakhaee-Zadeh Gutierrez (CTO) met at EPFL in Lausanne and started Adaptyv in 2021 alongside Moustafa Houmani and Amir Shahein. The team went through Y Combinator, then raised CHF 2.5 million from Swiss VC Wingman Ventures to refine the technology in its Lausanne lab.

In the 18 months since the Seed round, revenue grew roughly 10x and the team tripled. Adaptyv brought DNA synthesis in-house and ran “Proteinbase,” a competition that drew 680 participants submitting 10,000 protein designs. Its public API is now integrated with Boltz, Chai, Cradle, Tamarind, Latent Labs, and Benchling. The company says it now serves 100+ customers, including frontier AI labs, top-5 pharma companies, and AI-native drug discovery startups.

The headline case study is Anthropic. Anthropic ran “Claude Science” — using its Claude models to design proteins — across 16 targets pulled from Adaptyv’s public design competitions, sending the sequences to Lausanne anonymized so Adaptyv couldn’t tell which model produced which design. Published last week, the results showed 95% of 1,320 designs expressed successfully, and 354 (26.8%) bound their target — with Claude matching or beating expert human hit rates on most targets, and beating them by more than 3x on one. Google DeepMind runs the same design-validate loop at Adaptyv, Chai Discovery validated its zero-shot antibody model Chai-2 there, and Roche and Novo Nordisk say they now run therapeutic design cycles roughly 4x faster using Adaptyv’s lab.

The funding will go two directions. Horizontally, Adaptyv plans to triple lab capacity by the end of 2026 and open a new lab and office in London in Q4 2026, alongside its Lausanne base — growing the team from 25 to roughly 60, focused on production, automation, and software. Vertically, the company wants to move beyond binding validation — currently a market worth a few hundred million dollars a year — into developability and full biophysical characterization, then next-generation modalities like peptides, ADCs, and degraders, and eventually cell-based function, building toward what the company calls a “biological gigafactory.”

Adaptyv isn’t the only one trying to automate the lab itself. Recursion Pharmaceuticals recently unveiled “LabClaw,” an architecture combining five AI agents and 28 skill modules to run the full loop from target discovery to wet-lab execution, while the University of Toronto’s Acceleration Consortium — backed by a CAD 200 million grant — now operates a fleet of 50 self-driving robots across multiple university labs. On the design side, Chai Discovery (which raised $400M in July), Latent Labs (led by a team with a Nobel Prize in Chemistry pedigree), and Alphabet’s Isomorphic Labs are scaling up the design layer — Adaptyv is carving out the wet-lab validation layer underneath it.

Anthropic’s and Google DeepMind’s prominent roles here echo a broader pattern of Big Tech reaching into drug discovery.