PaperAI is an open-source AI application for working with medical and scientific papers, built by NeuML on top of the txtai framework. It has 1,787 stars on GitHub and remains under active development in 2026. PaperAI turns large collections of articles into a question-answering and report-generation engine.
Researchers often need answers that span hundreds of papers, not one document at a time. PaperAI indexes a repository of articles into an embeddings database, then runs bulk LLM inference over it: a configuration file defines the questions, and retrieval-augmented generation pipelines gather the relevant passages from across the corpus before the model answers. Think of it as launching hundreds of grounded ChatGPT-style prompts over your literature collection in one batch job, with every answer backed by the source papers. PaperAI inherits txtai’s flexibility around models and embeddings, so a workflow can run against hosted APIs or stay on local infrastructure when the literature is sensitive. That makes it practical for systematic screening, where the same battery of questions must be answered consistently across an entire corpus.
PaperAI is free and open source under the Apache-2.0 license. Costs are limited to your own infrastructure and any LLM API usage you configure.
PaperAI is for research groups, medical reviewers, and evidence teams who need to extract structured answers from hundreds of papers at once, such as literature screening and evidence synthesis. Its batch orientation makes it especially practical when the same questions must be asked of an ever-growing corpus.