Paper2Agent turns any research paper into an AI agent

Paper2Agent, described in a Nature news piece on 16 September 2026, converts a research paper into an AI agent that can field questions about the work, run its methods on new data, and collaborate with agents built from other papers. The team's Nature paper reports the conversion runs in about 45 minutes on a personal laptop for about $14.

The authors, Jiacheng Miao, Joe R. Davis, Yaohui Zhang, Jonathan K. Pritchard and James Zou, cast the output as a "virtual corresponding author." In the write-up by reporter Kaia Glickman, they say Paper2Agent makes it "easier for researchers to reproduce papers and understand work in unfamiliar fields." The pipeline parses a paper's text, supplementary materials, code and datasets, deposits them on an MCP server, and lets a team of AI agents build callable tools from the described methods.

The numbers land in computational biology. On an AlphaGenome case study, Paper2Agent produced 22 MCP tools and scored 98.7% on tutorial queries and 100% on novel queries. Applied to 100 papers from bioRxiv, it turned 74 into working agents comprising 593 validated tools, scoring 91.2% on a 300-question benchmark at $0.20 per query. In one collaborative test, three Paper2Agent instances nominated GPR137 as the probable causal gene at the psoriasis rs887314 locus, a result the team validated through signature-correlation analysis rather than wet-lab work.

The code sits on GitHub; three of the researchers we track shared the paper.