Frontier AI systems could transform biology research and drug discovery by navigating enormous search spaces. For example, a single 100-amino-acid protein spans at least 20100 possible sequences, which calls for dynamic, multi-candidate exploration. Google's long-term vision is for AI agents to not only ideate and analyze, but also run lab experiments, learn from the results, and self-improve, all driven by human scientific agency.
Realizing this vision requires solving two linked challenges. First, frontier models still do not reason reliably about 3-dimensional geometry and physical state, which complicates automating the liquid-handling robots that move reagents, track volumes, and execute molecular biology protocols. Second, while in-silico hypothesis generation and design are advancing rapidly, real-world wet-lab experiments have not kept pace, making experimental readouts the bottleneck more than ever.
To address the first challenge, the team presents the VERified Infrastructure for Discovery Laboratories (VERID) framework. VERID is a semantic digital twin of the laboratory workcell: a live, stateful 3D model that tracks everything from robotic arm geometry to individual well volumes, reagent identities, and pipette tip inventory. Built on the open-source PyLabRobot framework, it offers live 3D state tracking, fast virtual dry runs, hardware safety checks, and synchronized video and action logs.
In case studies, synchronized video caught uneven dye stratification caused by dispensing too close to the well floor, and the agent refined its mixing strategy. When the simulator rejected an 8-channel aspiration that would not fit in a narrow trough, the agent replanned and designed a targeted factorial diagnostic. In a closed-loop self-play campaign, an agent improved its accuracy more than fivefold by identifying its own optical and fluidic artifacts.
Source: Google AI Blog · Summarized by HeadlinesBriefing