Enter any topic and Paper Digest finds the most influential work around it - papers, patents, grants, clinical trials and the experts behind them - from hundreds of millions of research documents. Each selected document is analyzed individually, and the results are written up as a fluent literature review. Every reference in the output is a real, indexed document - nothing is cited that does not exist.
The system is built on Paper Digest's own semantic analysis, text summarization and machine learning, developed since 2018. By default, large language models help write the final prose, grounded in the selected documents. If you have zero tolerance for hallucinations, use the no-LLM version, where no large language model is involved at any step and nothing can be hallucinated.
Limit a review to a subject area, a date range, a specific conference or journal, or a single expert - or review patents, grants, clinical trials and experts instead of papers. Paper reviews can be exported as RIS or Markdown for Zotero, Mendeley, EndNote and Obsidian, and any review can be shared: click Share to get a permanent link that reruns the same review for colleagues. To see worked examples of every option, read how the AI literature review works.