Onava offers an autonomous orchestration platform that links AI‑native protein engineering models with high‑throughput functional assays, creating a continuous learning loop across drug discovery programs. By aggregating variant data—from thousands to billions per campaign—into a cross‑program data flywheel, the system reuses experimental results to accelerate candidate optimization for biopharma and biotech firms.
Funding
Funding not disclosed
Founders
Product
Problem
The drug discovery workflow creates huge volumes of experimental data, yet most programs operate in isolation and treat each experiment as a one‑off event. Consequently, insights from failed candidates are rarely reused, causing duplicated effort and extended development timelines.
Solution
Onava provides an autonomous orchestration layer that connects model training with experimental feedback across multiple discovery programs, forming a continuous learning loop. Its AI‑native protein engineering models are trained directly on functional biological outcomes, guiding each optimization cycle toward more viable candidates. The company’s proprietary high‑throughput functional assay platforms generate from thousands to hundreds of billions of variant data points per campaign, supplying the scale needed for effective AI training. By aggregating data across programs, a cross‑program data flywheel ensures that knowledge from earlier experiments accelerates later ones, compressing timelines without compromising quality. Human expertise is integrated to steer AI decisions and interpret results, while the scalable infrastructure treats every experiment as reusable data for future projects.
Target Audience
Biopharma companies and biotech firms that need accelerated biologic and protein‑based therapeutic discovery, particularly those requiring large‑scale functional screening and AI‑driven optimization.
Features
- Autonomous orchestration layer that automates model training and experimental feedback across programs
- AI‑native protein engineering models trained on functional outcomes rather than surrogate metrics
- Proprietary high‑throughput functional assay platforms capable of generating thousands to hundreds of billions of variants per campaign
- Cross‑program data flywheel that compounds learning, allowing each new candidate to benefit from prior experiments
- Integrated human‑in‑the‑loop expertise to guide AI decisions and interpret results
- Scalable infrastructure that treats every experiment as reusable data for future projects