Overview of all the new features in Elicit Notebooks

Elicit Notebooks get a redesigned workspace and a more modular workflow system that lets researchers run multiple queries, combine selected papers, and build deeper analysis step-by-step—while keeping everything organized and auditable. The new homepage shifts workflows from a horizontal layout to a vertical, “step” stack, making it easier to move through stages like searching for papers, extracting data from PDFs, and generating concept lists or topic summaries across multiple studies.

A key upgrade is notebook-level organization. Users can add a notebook title so the same notebook can hold multiple related research queries—such as “RCTs on cash transfers” or a dissertation topic—without forcing everything into separate sessions. When running “find papers” (for example, a question like the long-term effects of screen time on children), Elicit returns a familiar table of results, but with added controls: a full-screen view for deeper inspection, and the ability to delete papers one-by-one as relevance judgments are made. That combination is meant to speed up curation, so irrelevant studies can be removed early rather than carried through later steps.

The real power comes from adding new steps that build on the current state. After an initial search, users can add an independent query (another “ask a question and find papers”) that aggregates results into the same notebook. Alternatively, they can upload PDFs or pull from an existing library, then run extraction or generate a list of concepts from the selected papers. Multiple actions can run in parallel: a concept-extraction step can produce a concept list while the paper table remains available for review.

Once papers are selected, additional step options unlock. Users can create a new table from selected papers—useful for consolidating “best” or “most relevant” studies across many queries and sources. They can summarize abstracts for the chosen set, and they can “chat with papers” to explore what the selected papers are about in a flexible question-and-answer format. The chat interface also tracks which papers were used as inputs, and it supports navigation to open-access full text or abstracts for non-open-access papers. A toggle controls whether chat uses full text (when available via open access or uploads) or relies only on abstracts, with an explicit note that using full text can increase credit consumption.

To manage complexity over time, the interface includes a table-of-contents style navigator for jumping between steps, the ability to collapse sections, and the option to add columns (such as a methodology column) to existing tables. Users can export individual steps, while the notebook itself functions as a long-running session: every action is logged step-by-step, notebooks are saved for later continuation, and users can start fresh by creating a new notebook. Credit behavior remains similar to prior Elicit: uploading papers is free, but extracting data, processing more papers/columns, and using more text in chat increase credit usage. The overall effect is a more structured, iterative research workspace designed for multi-month projects rather than one-off queries.