Overview
Amazon revealed Bio Discovery on Tuesday, a new platform intended to connect computational biology with laboratory execution for drug discovery. The service gives researchers access to a catalog of more than 40 AI biology models and supports user-supplied models as well as third-party licensed options.
Platform capabilities
Bio Discovery is designed to let bench scientists and research teams employ AI-driven design tools without needing direct computational expertise. Users can evaluate models from the catalog, assemble workflows, or upload their own models. The platform uses AI agents to guide antibody-design decisions and provide recommendations without requiring users to write code.
The system returns AI-generated summaries and presents pre-filtered antibody candidates that have already undergone optimization and liability checks. When candidates are selected, they can be sent directly to integrated lab partners with real-time cost and turnaround time estimates. Results from wet-lab validation automatically flow back into the platform, enabling direct comparison between experimental data and model predictions.
Integrated lab partners and validation
Amazon positioned integrated laboratory partners such as Ginkgo Bioworks and Twist Bioscience to perform the wet-lab validation portion of the workflow. The company announced Bio Discovery during its AWS Life Science Symposium on Tuesday.
As a specific example of the platform's approach, an antibody design workflow validated by Sloan Kettering processed 300,000 targets and produced 100,000 candidate molecules across a four-step sequence. That workflow is available within the platform's environment for evaluation and reuse.
Market reaction and analyst note
Jefferies analysts commented that projects previously stalled in queues can be advanced immediately on the platform. They also noted that more accurate, integrated workflows may allow computational biologists to support a larger number of programs because of improved efficiency.
Conclusion
Bio Discovery combines a catalog of AI models, no-code guidance for antibody design, and direct links to wet-lab execution through partner labs. The platform emphasizes streamlined handoffs between in silico design and experimental validation while delivering cost and turnaround transparency for selected candidates.