Google Research announced new advancements in closed-loop molecular design by introducing the VERified Infrastructure for Discovery Laboratories (VERID) framework. VERID creates a semantic 3D digital twin of laboratory workcells, allowing AI agents to plan, simulate, and execute automated liquid-handling protocols on physical lab robots while tracking variables like reagent volumes, pipette tip inventory, and robotic arm geometry in real time.

By grounding AI agents in physical state tracking, the system bridges the gap between high-level generative hypothesis design and real-world wet-lab experimentation. In initial case studies, the framework enabled agents to automatically detect liquid dispensing anomalies, recover from hardware equipment failures mid-run, and dynamically replan constrained robotic movements during automated drug discovery campaigns.

Why it matters

  • Accelerates drug discovery by linking generative AI models directly to automated real-world lab execution.

  • Reduces scientific bottlenecking through dynamic digital twins that detect hardware errors and replan experiments.

Source: research.google