AutoDNA: AI-native lab automates biomolecular engineering

Researchers have developed AutoDNA, an AI-native autonomous laboratory for biomolecular engineering that can conduct complex experiments across diverse instruments with minimal human oversight, according to a new study. The system autonomously coordinates over 20 types of instruments across more than 11,000 steps for core nucleic acid operations like synthesis, transcription, amplification, and sequencing. Human intervention is limited to hardware maintenance such as reagent replenishment. In tests, AutoDNA matched expert-optimized DNA synthesis yields, accelerated concurrent protocol execution by 3.2-fold, completed a DNA data storage cycle with exact digital content recovery, and identified a clinically detected KRAS G13D mutation. The research, led by teams at Shanghai Jiao Tong University, addresses a limitation in prior autonomous systems that used AI as a task-specific tool rather than providing global experimental understanding. AutoDNA's co-design of models, experiments, and instruments enables AI and automation to co-evolve, establishing a blueprint toward science-as-a-service for synthetic biology and information storage applications.