AutoDNA: agent-run biomolecular laboratory reported in Nature
Nature published AutoDNA on 4 November 2026, an AI-native autonomous laboratory for biomolecular engineering built by a Shanghai Jiao Tong University group led by Haibo Chen. Six specialised language-model agents interpret a natural-language request, design the protocol, write the instrument control code and drive more than 20 types of laboratory instrument, executing over 11,000 autonomous hardware steps. The system ran end-to-end nucleic acid testing, optimised enzymatic DNA synthesis to yields matching expert practice, and completed a DNA data-storage cycle in which the original digital content was recovered exactly. Nature ran an accompanying News and Views piece, 'AI agents enter the wet lab'. Human staff still replenished reagents and consumables and intervened on hardware faults.
Why It Mattered
Self-driving laboratories have existed for a decade, but they have largely been single-purpose: one instrument class, one objective, a fixed workflow. AutoDNA is notable because the planning layer is a general language-model system that reasons about instrument capabilities abstracted as services, which lets it compose multi-step, multi-objective experiments across heterogeneous hardware from one prompt. That shifts the bottleneck in automated experimentation from workflow engineering to orchestration, and it is the architecture most likely to be copied. The DNA data-storage round trip is the strongest single demonstration: synthesis, storage, sequencing and exact recovery is a closed loop with an unambiguous pass or fail, not a benchmark score. The honest limits matter as much as the result. Humans remained in the loop for physical maintenance, and the companion commentary records hardware errors that required hands. The entry belongs in a history of AI as an early, peer-reviewed data point on where autonomous experimentation actually stood in 2026, and as a marker that the leading demonstration of agent-run laboratory science came from a Chinese university group publishing openly, with code and data deposited. It also supplies the concrete counterpart to the year's many claims about AI scientists: a system whose outputs were physical, reproducible and reviewed.
Who Built It
Shanghai Jiao Tong University (Haibo Chen and colleagues)
Applications
- Synthetic Biology
- Laboratory Automation
- DNA Data Storage