Combine papers from search & uploads in Elicit Notebooks

Elicit Notebooks adds a workflow for merging papers found inside Elicit with papers already in a researcher’s library or uploaded from other sources—so analysis can span multiple databases without losing context. The core move is to start with a question that searches Elicit’s large paper database, then selectively add additional papers (via PDF upload or imports like Zotero) into the same notebook workflow for side-by-side analysis.

A typical flow begins by asking a targeted question over Elicit’s database—here, a lung-cancer question about biopsy frequency among lung-cancer patients in the US. The search returns an initial set of papers (eight to start), and the researcher can keep loading more results as needed. From there, the notebook workflow supports expanding the dataset with papers outside Elicit: an “extract data from uploaded papers” step lets users select papers they’ve collected externally and bring them into the notebook’s working set. Uploads can be done via PDF, and imports can come from Zotero; selection can be broad (including selecting all papers in a folder) or narrow (choosing a few specific papers).

Once the combined set is assembled, the interface distinguishes between public Elicit papers and private uploaded papers. Users can then choose which papers matter most—whether that means focusing on a subset from the public search results, the private uploads, or a mix of both. With that curated selection, the notebook can generate a new table that becomes the basis for deeper analysis.

From the new table, Elicit Notebooks supports multiple “go deeper” actions: summarizing abstracts for the selected papers, creating structured views that help compare methods and findings, and enabling chat-style interaction with the papers themselves. The workflow also supports iterative refinement: after drilling into the most relevant papers, users can return to broader search with slightly different query framings, load additional papers, and repeat the narrowing process.

The practical payoff is a two-stage research loop—go broad by aggregating papers from Elicit’s database (described as about 125 million papers across open access and paywalled sources) plus user-collected materials from other systems, then go deep by narrowing to a high-quality subset and running analysis on that set. The notebook also allows adding columns to extract data across the selected papers, including columns based on user-defined criteria or columns that help filter papers related to a specific query. In short, the feature is designed to keep multi-source literature organized and analyzable in one place, enabling theme-finding and method comparison across both discovered and user-supplied papers.