Cradle

Cradle is a technology company that integrates artificial intelligence-assisted design software for protein engineering with a wet laboratory infrastructure that continuously validates its models. The platform is designed to enable researchers to generate protein candidates using experimental data and simultaneously optimize multiple targets. The company regularly tests its models under near-real-world conditions through laboratory work conducted at its headquarters in Amsterdam, providing feedback to its software.

Cradle does not limit itself to computational approaches; since its founding, it has operated its own wet laboratory in Amsterdam. This choice enables machine learning models to be regularly subjected to A/B testing with experimental data and validated against dozens of measurements and properties with every release. The company aims to improve the general performance of its default models by developing targeted large datasets that encompass fundamental sequence-property relationships such as expression level, stability, and resolution.

Cradle implements comprehensive automation to accelerate laboratory workflows and enhance data quality. Repetitive tasks are delegated to robotic systems, reducing human error and shortening iteration cycles; turnaround times that traditionally spanned months have been reduced to weeks. Experiments are miniaturized and high-throughput screening is enabled through the use of nanoliter-scale precision liquid handling systems such as the Formulatrix Mantis in its workflows. The company regularly shares its methods and principles for automation through blog posts and guides.

The Cradle platform enables researchers to import their own experimental data to generate molecular candidates and optimize multiple properties simultaneously within a single round. Users can define targets such as activity, binding, stability, resolution, and expression; the platform selects plate-level candidates from millions of computational designs for laboratory testing. The system attempts to model variations in experimental conditions and potential batch effects by integrating data from multiple rounds and generates variants with multiple mutations from the first round to capture epistatic interactions. Workflows can be configured either through a simple interface with default assumptions or via API with detailed constraints; mutation rules can be specified for active regions or user-defined regions.

The platform aims to accelerate protein engineering in industrial biotechnology and biopharmaceutical research pipelines. It supports objectives such as catalytic turnover, heterologous expression, and resilience under harsh process conditions in enzymes; affinity maturation, developability, immunogenicity, and avoidance of liability motifs in antibodies. Data-driven improvement workflows are provided for diverse biological modalities including vaccine antigens and peptides. Cradle also enables projects with no prior data to be initiated with a single sequence and works with data from diverse formats such as existing NGS, FACS, ELISA, and chromatography outputs.

Custom models are trained exclusively on experimental data provided by the institution. Data security is prioritized through applications such as SOC 2 compliance, single sign-on support, and active monitoring. The platform offers a structure that allows custom predictive models to be used within reports and provides a fully managed, scalable GPU infrastructure. When outputs are reported, predicted performance scores of candidates and three-dimensional visualizations of specific mutations can be examined; laboratory rounds are tracked via “round status” monitoring.

Cradle’s solutions are used by research teams at industrial biotechnology companies such as Novonesis, Corteva, and IFF, as well as biopharmaceutical firms including Novo Nordisk, argenx, and Johnson & Johnson. Public competitions and benchmarking efforts include examples such as the Cetuximab optimization reported in the Adaptyv Bio protein design challenge and the documented increase in EGFR binding affinity. Cradle has published studies evaluating enzyme engineering using benchmark-focused “Align to Innovate” approaches and fully automated generative models.

Cradle is structured as a multidisciplinary team of experts in biology, machine learning, software engineering, and design. Founders and team members bring prior experience in product and technology development from various technology and biotechnology companies. The company provides technical guidance to users on experimental design and platform usage through customer support teams and scientific advisors.