Grounding AI agents in biological reality: Building blocks for autonomous wet labs
Google Research introduces the VERified Infrastructure for Discovery Laboratories (VERID) framework, a semantic digital twin that provides live 3D state tracking, fast virtual dry runs, hardware safety checks, and synchronized video‑action feedback to safely ground AI agents in wet‑lab robotics. Built atop the open‑source PyLabRobot framework, VERID lets agents simulate and validate liquid‑handling protocols before execution, reducing physical errors and enabling closed‑loop molecular design cycles. This bridges the gap between high‑level AI reasoning and the error‑prone physical reality of laboratory automation, accelerating discovery of high‑affinity protein binders.
⚡ Key Takeaways
- VERID is a live, stateful 3D volumetric entity tree that tracks robotic arm geometry, well volumes, reagent identities, and pipette tip inventory.
- Core capabilities include Live 3D state tracking, Fast virtual dry runs, Hardware safety checks, and Synchronized video and action feedback.
- Built on the open‑source PyLabRobot framework, enabling integration with existing lab robotics stacks.
- Safety checks block physically impossible commands (e.g., lowering an 8‑channel pipette into a too‑short trough) and logistically invalid actions (e.g., aspirating from a depleted well).
- Virtual dry runs simulate protocols in seconds, catching collisions before reagents are touched.
By providing a reliable simulation and safety layer, VERID lets AI agents iterate rapidly on experimental designs, cutting down costly wet‑lab failures and speeding up the discovery of nanomolar protein binders for therapeutic targets.
✅ Practical Steps
- Integrate VERID with your lab’s robot by adopting the PyLabRobot framework as the underlying control layer.
- Use VERID’s live 3D state tracking API to query deck layout, well volumes, and tip inventory before issuing any liquid‑handling commands.
- Run VERID’s fast virtual dry‑run simulation to validate protocols and catch collisions or empty wells prior to execution.
Want the full story? Read the original article.
Read on Google Research ↗