Key points
- Nature Medicine published the paper on an open, agentic AI co-scientist for clinical and biomedical work on November 16, 2026.
- The corresponding authors are Randall J. Bateman and Justin T. Reese.
- Kaleigh F. Roberts and Zachary B. Abrams are listed as equal contributors.
- Authors' affiliations include Washington University in St. Louis, Lawrence Berkeley National Laboratory, Harvard Medical School and the Broad Institute.
A large research team has described an AI "co-scientist" built to help with clinical and biomedical discovery. In a paper published November 16, 2026, in Nature Medicine, the authors call the system open and agentic.
Agentic AI generally means software that can plan and carry out several steps of a task on its own, such as calling up tools or data, rather than answering a single prompt.
The corresponding authors are Randall J. Bateman of the Department of Neurology at Washington University in St. Louis and Justin T. Reese of the Division of Environmental Genomics and Systems Biology at Lawrence Berkeley National Laboratory. Both are also listed with the Consortium for Biomedical Research & AI in Neurodegeneration (C-BRAIN) in St. Louis. Kaleigh F. Roberts and Zachary B. Abrams contributed equally to the work. Other co-authors are affiliated with Harvard Medical School, the University of Fribourg, the UK Dementia Research Institute at University College London, Sage Bionetworks, the Broad Institute of MIT and Harvard, Stanford University and the University of Washington.
The material DexScience reviewed lists the paper's authors and their institutions. It does not explain how the system works, what tasks it was tested on or what it found. As a result, we can't yet say how well the co-scientist performs or how it compares with other tools.
Why it matters
The paper's title presents the system as an open AI co-scientist for clinical and biomedical discovery, which suggests other researchers may be able to examine or build on it.
What we don't know yet
The source material includes only the paper's title, publication details and author list. It gives no methods, test results or performance measures, so the system's capabilities can't be judged from it.
Terms in this story
- Agentic AI
- AI software that can plan and carry out tasks with several steps, often by using tools or data sources, with limited step-by-step human direction.
- Corresponding author
- The researcher named on a paper to handle questions and correspondence about the work.
Sources and verification
DexScience articles are produced with AI assistance and an automated fact-check against the linked sources. Spotted an error? Tell us and we will correct it openly.


