An AI-native experimental laboratory for autonomous biomolecular operations.

AutoDNA, an AI-native autonomous biomolecular engineering laboratory, orchestrates 20+ instrument types across 11,000+ steps to run nucleic acid synthesis, transcription, amplification, and sequencing with minimal human input.

• Why it matters: Autonomous labs are still mostly limited to single-goal workflows and simple setups, leaving a gap for multi-objective biomolecular engineering under engineering-oriented constraints.
• What they did: AutoDNA co-designs models, experiments, and instruments in an end-to-end multi-user framework, autonomously executing over 11,000 steps while humans mainly handle hardware maintenance like replenishment.
• The result: In four applications, it matched expert-optimized enzymatic DNA synthesis yields, accelerated concurrent protocol execution by 3.2-fold, achieved exact recovery in an end-to-end DNA data storage cycle, and identified and validated a clinically detected KRAS G13D mutation.

AI summary of the abstract · may contain errors · not medical advice · read the paper.