Stanford researchers unveiled Biomni, an AI “co-scientist” designed to autonomously run biomedical research workflows rather than only answer questions. The agent is described as capable of interpreting research prompts, planning experiments, selecting datasets, and writing code to execute parts of the workflow—aimed at compressing discovery cycles across data-intensive biology. The announcement comes as multiple groups push toward agentic systems for drug discovery, where the core value is shifting from information retrieval to task execution across scientific toolchains.