Paper2Agent converts research papers into interactive AI agents

A team led by Jiacheng Miao and James Zou has published a system in Nature that converts a research paper into a working AI agent in about 45 minutes for roughly $14 on a personal laptop. Their framework, called Paper2Agent, walks a paper's codebase through six steps: identifying it, setting up the environment, discovering tutorials, running them, extracting tools, and assembling a Model Context Protocol server. The output is an agent a chatbot can call directly.

The authors describe that output as a "virtual corresponding author, exposing its manuscript, supplementary materials, datasets, code and workflows as active, agent-native knowledge" rather than static text.

Applied to 100 computational biology papers, the pipeline agentified 74, producing 599 tools of which 593 passed automated validation, with 91.2% accuracy on benchmark questions. A case-study agent built from AlphaGenome generated 22 validated tools and answered novel queries with 100% accuracy, against 82.7% for a Claude-plus-repo baseline and 37.3% for Biomni, and it ran 1.9x faster than the Claude baseline.

Five of the AI researchers we track had already shared the paper by the time it hit our alert queue, which fits its own framing: Paper2Agent "transforms research output from passive artifacts into active systems that can accelerate downstream use, adoption, and discovery." Human researchers, the authors note, remain responsible for hypothesis selection and evidence evaluation.