Adaptyv

Adaptyv Bio (legal entity Adaptyv Biosystems Sàrl) operates an automated, AI-native wet laboratory for experimental validation of designed proteins. Customers submit protein sequences digitally through a web platform or an API, the company's automated lab performs the assays, and results are returned as a structured data package, with a stated turnaround of about three weeks from sequence design to experimental data. The workflow synthesises DNA in-house, expresses proteins cell-free, and measures binding using SPR (and BLI) with built-in quality control.

The service catalogue covers protein expression analysis, binding/affinity characterisation, and thermostability characterisation, offered at screening or full-affinity resolution against targets from Adaptyv's own library or targets shipped by the customer. Alongside experiment ordering, the platform provides experiment management, progress notifications, results analysis, raw-data export and tracking of design-round performance across time, positioning the lab as the "test" step in a design–build–test–learn loop for generative protein design.

The company also publishes community and benchmarking work: protein design competitions (EGFR, TREM2, Nipah), the BenchBB "Bench-tested Binder Benchmark", the Proteinbase repository of experimental protein design data including negative results, and case studies validating externally developed models such as Microsoft Research's EvoDiff, BindCraft, ProtRL and Pro-1.

Founding story

The founders positioned the company around the bottleneck that AI can generate thousands of candidate protein designs in silico while experimental validation remains slow, costly and technically complex; CEO Julian Englert has described the company as building the link between computational design and real-world validation, noting that biology AI models are only as good as the experimental data they are trained on.

Business model

Fee-for-service laboratory validation sold per protein, with published price points starting at $49 per protein for expression, $99 per protein for binding or thermostability, and from $149 per protein for configurations including two replicates; orders are configured by number of proteins (24–768), assay resolution and target source, and placed via a self-serve web platform or API.

Per-protein service revenue from expression, binding and thermostability assays, with pricing scaling by protein count and assay resolution and additional cost for targets supplied from Adaptyv's library; revenue was reported to have grown roughly 10x between the 2024 seed round and 2026.

Traction

Reported metrics include over 10,000 proteins tested in 2025 and hundreds of experiments run during the beta period, more than 30 companies using the service in 2025 rising to over 100 customers by 2026, more than 10 preprints published in 2025 using Adaptyv data, roughly 10x revenue growth since the seed round, a team that more than doubled in six months during 2025 and tripled by 2026, and a competition platform that attracted 680 participants and 10,000 protein designs to a single competition. In an international competition the company reported raising protein success rates from 2.5% to 13% in two months. In the Anthropic benchmark, 95% of 1,320 designs expressed and 354 bound their target, an average hit rate of 26.8%, with only one target yielding no binders. A TREM2 hackathon with muni produced 141 designs from 16 teams, of which 100 went to the wet lab and 37 bound, with the best human binder at 1.11 nM and the best agent binder at 3.64 nM.

Latest developments

In August 2026 Adaptyv announced a $40 million (€34.35 million) Series A led by Highland Europe with existing investors ACE Ventures, ByFounders and Y Combinator participating, to roughly triple lab capacity by the end of 2026, open a London lab and office in Q4 2026, expand in Lausanne, and grow the team from 25 to roughly 60 in production, automation and software. It also plans to move beyond binding measurements into developability and full biophysical characterisation, next-generation modalities such as peptides, ADCs and degraders, and eventually cell-based function. In the same month it published a benchmark run with Anthropic in which Claude models designed binders against 16 targets from Adaptyv's public competitions for anonymised wet-lab testing.