Stanford's Paper2Agent turns a research paper into an AI agent

Stanford's Paper2Agent, described in Nature, converts a research paper into an AI agent. The system deposits a paper's main text, code and datasets onto a Model Context Protocol server, then lets a team of AI agents autonomously build tools that apply the paper's methods to fresh data. Any large language model can then call those tools.

In a test on the AlphaGenome paper, Paper2Agent 'built 22 tools in about 45 minutes for US $14, and all 22 passed validation without human intervention.' The resulting agent answered genetics questions with near-perfect accuracy and outscored Biomni, a tool developed by academic researchers that draws on dozens of databases.

James Zou, a computer scientist at Stanford University and a co-author, says the ability to convert static papers into dynamic sources of information 'can help us to reimagine what knowledge looks like in the future.'

When two paper-derived agents were connected to each other, they flagged an MPHOSPH9 variant tied to ADHD risk that Zou says had not previously been reported. Asked separately to explain why a DNA change is linked to bad cholesterol, the AlphaGenome agent named a different causal gene from the one pinpointed in the original AlphaGenome paper, though Zou noted both hypotheses were supported by AlphaGenome's data.