MGI Tech and Shanghai AI Lab Introduce Physical AI Systems for Life Sciences

MGI Tech Co., Ltd.'s subsidiary Genoria AI and Shanghai Artificial Intelligence Laboratory announced in a press release the release of two systems, ProtoPilot and BioLab Bench, designed to integrate AI reasoning with physical laboratory automation in biology.

ProtoPilot is described as a self-evolving multi-agent framework meant to model real laboratory workflows. The system achieved a 52.38 percent score on the ProtocolQA benchmark, demonstrating its ability to interpret and operationalize complex experimental procedures. BioLab Bench acts as a complementary evaluation framework that measures agent performance from textual intent through to physical execution on laboratory devices.

The collaboration defines a concept referred to as Physical AI for life sciences. This approach connects AI models with hardware operations to produce reproducible laboratory results. According to the announcement, BioLab Bench covers a wide range of biological tasks, provides evaluation from planning to device action, and supports transferability across different automation systems.

The research supporting these releases was published as a preprint on arXiv in June 2026. Genoria AI was established in April 2026 as MGI's AI for Science division, following several years of AI-focused biological research beginning in 2019.