(52.390999, 4.835315)
Cradle Amsterdam HQ
Many AI protein engineering companies work purely in silico. We're different.
From day one, we’ve run our own wet lab in Amsterdam—because it’s the only way to build AI scientists can truly trust. Real experiments drive reliable results, and our in-house lab lets us move faster by continually improving the AI models that power the Cradle platform.
Each machine learning model iteration undergoes rigorous A/B testing against its predecessor. Our lab team systematically validates performance improvements across dozens of assays and properties. This way, we consistently validate and improve our models.
Beyond validation, our scientists develop targeted large datasets that capture fundamental sequence-property relationships. We systematically map out how sequence changes affect properties like expression, stability, and solubility. Datasets like these help make the default machine learning models in Cradle better.
Having in-house scientists means we understand the daily challenges of protein engineering. The constant exchange between our software, ML, and bio teams ensures that we deeply understand our customers’ needs. On top of this, we leverage our lab and in-house expertise to benefit the broader scientific community by frequently sharing findings, methods, and more on our blog.
Biology has always moved on slow feedback loops. The faster we can generate lab data, the faster our AI models get better.
At Cradle, we optimize and automate our lab workflows to speed up iteration and improve our machine learning models. Automation also reduces human error, giving us higher-quality data. Our robots take care of the repetitive tasks, freeing scientists to focus on the hardest questions. As a result, we’ve cut the traditional turnaround time of multiple months down to weeks.
Lab automation at Cradle
3 min
We're big on automation—and sharing our knowledge.
Most labs keep their work locked behind layers of protection. Because we don’t develop proprietary assets, we take a different approach—openly sharing our workflows and insights with the broader community.
Bringing machine learning into R&D means turning data around much faster, yet many research cycles still take months. By sharing what we learn in real time, we hope to help teams adopt these tools more quickly and make their research move faster.
