Encountering a paper can alter a research idea’s trajectory: a researcher might find their contribution already addressed, or spot an unresolved problem that opens a new direction. The authors call this a literature-initiated pivot, and argue it connects two activities often managed separately—developing an idea and deciding which literature to measure it against. With collaborators at the University of Pennsylvania, they built LitPivot, a system that organizes literature around the specific part of a proposal being edited, offers critiques supported by excerpts from selected studies, and suggests revisions that researchers can accept, edit, or reject.

Why pivoting is hard: analysis of meeting notes from four completed computer science research projects identified 21 literature-initiated pivots, falling into two types: differentiating an idea from existing work to establish novelty, and aligning an idea with existing work to build on it. In a formative study, five researchers using a text editor and standard search tools struggled to develop ideas while critically evaluating them against the literature; when a paper challenged an idea’s novelty, they wanted concrete revision suggestions supported by prior work. That shaped LitPivot‘s assessments, which pair literature-supported critiques with specific suggestions for revising the selected part of an idea.

How LitPivot works: a researcher starts by writing a short proposal and enabling facet highlighting, which segments the idea into problem, contribution, and evaluation. Clicking a segment brings up an adjacent panel grouping related papers by that facet, for example grouping by evaluation methods when the evaluation facet is selected. The researcher selects papers, requests an assessment, and reviews a suggested revision alongside supporting excerpts. The literature is then reranked based on relevance to the revised facet. LitPivot separates a research idea into its problem, proposed contribution, and evaluation plan; each facet has its own set and relevance ranking of related literature. For a problem statement, it looks for evidence the problem exists and reasons it matters. For a contribution, it compares what selected papers have addressed with limitations and future work, looking for gaps. For an evaluation plan, it checks whether the proposed tests address the research question and how similar work has been evaluated. Changing one facet can leave another out of date, so LitPivot flags passages that may need revision. Participants accepted, edited, or rejected suggestions, and could request new papers or refresh the literature ranking.

Findings: in a lab study, 17 researchers developed one assigned idea with LitPivot and another with a baseline tool—a text editor and an AI tool for asking questions about selected papers, using the same paper collection. Two domain experts, unaware of which system was used, rated proposals on how well they used relevant literature to support and justify the idea. Proposals developed with LitPivot received higher ratings. Participants also reported stronger understanding of the surrounding literature and greater self-reported improvement in novelty and usefulness. One participant started with a proposal to generate consistent, factual synthetic medical notes; LitPivot identified consistency across multiple hospital visits as a potential gap, and after removing papers about single-visit notes and another assessment, the participant narrowed the proposal to detecting contradictions across notes from multiple visits. In a follow-up study, five researchers used LitPivot on their own ongoing projects in 30-minute sessions, across multiple project phases: two used it to vet new ideas against the literature, while three used it to refine the framing of more mature ideas. One researcher highlighted how seeing parallel concepts from different domains helped reframe their contribution.

Boundaries: LitPivot is intended for early-stage research ideation, but the idea could possibly extend to longer writing tasks such as introduction sections. The authors place it alongside scientific-writing environments like Prism and agentic assistants like Claude Code, exploring a workflow where researchers control both the draft and the evidence used to assess it. At Ai2, it extends work on theory generation, synthesis, and hypothesis testing—Asta‘s Theorizer and AutoDiscovery, HARPA, Scideator, and IdeaSynth—and follows their study Old Ideas, Novel Problems, which examined the limits of automated novelty evaluation. LitPivot makes the choice of relevant literature part of the revision process. The system will be presented at UIST 2026 in Detroit.