This post is co-authored with the Biomni group from Stanford.
Biomedical researchers spend approximately 90% of their time manually processing massive volumes of scattered information. This is evidenced by Genentech’s challenge of processing 38 million biomedical publications in PubMed, public repositories like the Human Protein Atlas, and their internal repository of hundreds of millions of cells across hundreds of diseases. There is a rapid proliferation of specialized databases and analytical tools across different modalities including genomics, proteomics, and pathology. Researchers must stay current with the large landscape of tools, leaving less time for the hypothesis-driven work that drives breakthrough discoveries.
AI agents powered by foundation models offer a promising solution by autonomously planning, executing, and adapting complex research tasks. Stanford researchers built Biomni that exemplifies this potential. Biomni is a general-purpose…
