data-to-paper is an AI-driven framework designed to guide the complete scientific research process, from raw data to human-verifiable research papers. It utilizes LLM and rule-based agents to navigate through tasks such as creating hypotheses, conducting literature searches, interpreting results, and writing comprehensive manuscripts. Key features include field-agnosticism, open or fixed-goal research options, data-chained manuscripts, human oversight through a GUI app, and transparent replay of the entire process. Users can try out data-to-paper with their own datasets and contribute to its development. However, users must accept all risks associated with the software and ensure compliance with laws and ethical standards.
https://github.com/Technion-Kishony-lab/data-to-paper