AI agents run microbiology labs: a robotic system that reads DNA and creates it

For many years, possessing "skilled hands" remained the decisive standard for evaluating researchers in molecular biology laboratories, where sample processing and the use of precise glass pipettes required exceptional focus and manual craftsmanship. Today, that laboratory equation has shifted after researchers at Shanghai Jiao Tong University, in a study published in the journal Nature, revealed a fully automated genetic laboratory run by software and AI agents, carrying out complex experiments from start to finish without the need for a human touch in the experimental path itself.

The system named "AutoDNA" represents a step forward toward what is known as self-driving laboratories.It relies on robotic arms that transfer samples and coordinate work among more than twenty different types of laboratory equipment, completing thousands of precise operations in a single project.Launching an experiment requires only a simple initial direction from the human supervisor, after which a team of intelligent agents plans the steps and programs them, then operates the devices, monitors the results, and addresses any protocol deviation, while technicians are limited to replenishing consumable reagents and maintaining mechanical hardware.

Biological laboratories have long known varying degrees of automation through machines that perform specific tasks such as mixing or heating, but building a self-driving laboratory for molecular biology remained a complex challenge due to the multitude of devices and the need for extreme precision in adjustment. Hybo Chen, a computer scientist who co-authored the study, explained that previous systems added AI algorithms to handle only one stage while experts remained constrained to calibrating each device individually; therefore his team chose to design the new system from scratch on the basis of full integration with intelligent agents to manage a laboratory environment rich in diverse equipment.

The system proved its practical capabilities through a series of rigorous field tests, successfully screening DNA samples and detecting a genetic mutation linked to a type of cancer with high accuracy, in addition to its ability to sequence and synthesize gene strands. In another demonstration of encryption efficiency, the system stored a text message within the four nucleotide bases (A, C, G, T) borrowed from a popular quote attributed to physicist Stephen Hawking that states intelligence is the ability to adapt to change, prompting Professor of Chemical Engineering at North Carolina State University, Milad Abu Al-Hussni, to emphasize that the future of scientific work will clearly depend on expanding such independent laboratories.

This fundamental shift in wet-lab environments redraws spending and operational priorities in research and medical institutions across the Gulf, Egypt, and the Levant.Large regional projects aimed at mapping genomes and precision oncology programs often clash with the scarcity of laboratory expertise capable of spending thousands of hours transferring liquids and manually sorting. With the entry of software agents into the heart of these processes, the investment burden shifts from expanding traditional laboratory staff to building layers of software supervision and securing supply chains for chemical reagents and robotic spare parts, raising laboratory operational capacity to round-the-clock operation at a lower cost per sample.

This new reality forces universities and health centers in the region to redefine the skills required of graduates in biology and biotechnological sciences; the critical skill is no longer confined to manual practice in the lab but extends to supervising independent software teams and setting quality standards for data produced by machines. If health institutions race toward rapid genomic diagnosis, their ability to adopt a unified laboratory architecture that integrates AI software will determine the difference between laboratories that wait weeks for analysis results and those that complete sequencing and synthesis in a few hours.