Google grounds AI agents in wet labs, discovers nanomolar cancer-target binders in closed loop
aigclink · x · 2026-10-19
Google Research published three projects bringing AI agents to the wet-lab bench: agents not only design molecules but operate liquid-handling robots, analyze real experimental results, and adjust the next design cycle in a closed loop, discovering single-digit nanomolar protein binders for cancer targets. Key pieces:
- VERID: uses digital twins to ground agent instructions in real lab equipment state, tackling frontier models' unreliable reasoning about 3D geometry and physical state in lab robotics automation;
- Agentic aptamer sandbox for agentic aptamer design;
- Epitope excision experimental methods.
The team notes a single 100-aa protein spans at least 20^100 sequences, demanding dynamic multi-candidate exploration, and that wet-lab throughput—not hypothesis generation—is now the bottleneck. Three papers accompany the post.
More from AGI Musings
- Viral theory claims RSI-capable models can't be open-sourced, DeepSeek dates are frozen snapshots — teortaxesTex · 2026-10-19
- Victor Taelin speculates on AGI-era stroke first aid: recovery falls from 72% to 40% within 6 hours — louisvarge · 2026-10-19
- ICLR reviewer slams AI safety papers as CV padding; critic says Goodharting wrecked the field — JacquesThibs · 2026-10-19
- Anthropic reveals Claude now drives 26% of its own AI R&D — emmanuelvivier · 2026-10-19
- Stanford Opens Free CS336 Course: 18 Lectures on Building LLMs from Scratch — mdancho84 · 2026-10-19
- Designer Pitches Herself Publicly With 'Design Is Dead' Film Amid AI Anxiety — NirantK · 2026-10-19