Stanford and Mount Sinai won the Alzheimer’s AI Prize and pledged to make Biomni-AD freely available on a shared Alzheimer’s disease data platform

Stanford and Mount Sinai won the Alzheimer’s AI Prize and pledged to make Biomni-AD freely available on a shared Alzheimer’s disease data platform

On March 20, the Alzheimer’s Disease Data Initiative awarded $1 million to the Biomni-AD team, and on March 27, Mount Sinai separately announced that it wants to deploy the system for researchers worldwide through AD Workbench — a free cloud environment for working with Alzheimer’s disease data. Biomni-AD is designed as an AI assistant that takes a plain-language question and returns a ready-made analysis plan, code, charts, and a report. For now, this is an announced rollout, not an already established standard.

According to the prize organizers’ announcement, Biomni-AD became one of two winners of the Alzheimer’s Insights AI Prize. The same announcement says the tool is supposed to be made free for all researchers through AD Workbench. That would move the system from a competition project into a shared working environment for Alzheimer’s research.

“We want researchers around the world to move from data to insight in minutes, not months,” Mount Sinai geneticist Kuan-lin Huang said in the team’s team announcement.

That same text describes the system in fairly concrete terms. It is meant to bring genetics, single-cell data, CRISPR screens, proteomics, biomarkers, and clinical datasets into one workflow. The team writes about 180+ specialized tools, reproducible output with code and charts, and human oversight at every step: the researcher reviews the plan first and only then approves the run.

Alzheimer’s is a fitting proving ground for this approach. There is already a huge amount of data, but it is scattered across different platforms, formats, and access agreements. AD Workbench was built for exactly this problem: a dataset catalog, workspaces, virtual machines, change auditing, and code sharing. If the integration works, a researcher could come in with a question like “which signals in single-cell data line up with clinical biomarkers?” and get not just an answer, but the full analytical trail.

Earlier in March, we wrote about the broader Biomni stack as infrastructure for biology. This new step is different in both scale and form: Biomni is being tied to one specific disease, where there is already a dedicated data repository, working environment, and international research community. Alzheimer’s becomes the testbed where AI is being turned into a shared tool for the entire field.

For now, this story has clear limits. The organizers have not yet shown a public independent benchmark comparing Biomni-AD with the best general models specifically on Alzheimer’s tasks. Free deployment through AD Workbench remains a launch plan. If the team gets it into working shape, researchers in the field could start with a question to the system, while the platform assembles a reproducible analysis inside a shared data environment.