Virtual biotech company puts thousands of AI scientist agents to work on drug discovery

The latest spin-off from a Stanford Medicine lab is a biotech company entirely composed of artificial intelligence agents, with no human employees, lab space, or payroll. Founded by associate professor James Zou and graduate student Harrison Zhang, the virtual biotech firm aims to replicate the entire drug discovery process through AI scientists. The company's goal is to tackle drug development, from identifying molecular targets to designing clinical trials.

To test their hypothesis, Zou and Zhang's AI agents were tasked with identifying biological signals that predict the success of drug candidates and designing cancer therapies. By analyzing 50,000 clinical trials in less than a week, the agents discovered that drugs targeting "switch-like" genes—genes with high bimodality—were more likely to advance through clinical trials and reach the market. These drugs also had fewer adverse events.

The virtual biotech company then identified the protein B7-H3 as a potential target for cancer treatment. Fibroblasts expressing B7-H3 seemed to be suppressing immune activity and cloaking tumors from the immune system. The AI scientists proposed an antibody-drug conjugate, which delivers a toxic chemotherapy payload directly to cells with high B7-H3 expression.

A pharmaceutical company independently developed the same treatment in August 2025, leading to a Food and Drug Administration breakthrough therapy designation. This success demonstrates the potential of AI-driven drug discovery, offering faster, more efficient, and cost-effective solutions to complex pharmaceutical challenges.