Themes become conclusions
A shared topic can be written as agreement even when each paper reports a different result.
Synthesize patterns without losing the papers behind them.
Generate themes, agreements, disagreements, limitations, and gaps from verified matrix cells. Every retained item cites source cells from at least two papers and remains open for researcher review.
Free includes 10 papers per month, the complete evidence workflow, and portable exports.Workflow
Move from reviewed cells to an auditable synthesis structure.
The synthesis stage begins only after source evidence has been checked across at least two papers.
- 01
Verify the matrix
Resolve review-required cells and confirm that the exact quotes and claimed pages match the source PDFs.
- 02
Generate cross-paper items
Create conservative themes, agreements, disagreements, limitations, and gaps from the verified evidence set.
- 03
Review each relationship
Open the cited source cells, accept supported items, flag weak relationships, and regenerate when the matrix changes.
A fluent summary can hide a weak evidence relationship.
Cross-paper writing often compresses several studies into one sentence. When citations and source passages are separated from that sentence, it becomes difficult to see whether papers agree, conflict, or simply discuss different topics.
Absence becomes a gap
A paper that does not mention an issue cannot by itself support a field-level research gap.
Sources become decorative
A citation list adds little protection when the exact cells supporting each clause are not available for review.
Inside EvidenceGrid
A synthesis claim should open back into its evidence.
Review the statement, inspect the supporting cells from each paper, accept it, request revision, or keep a category empty when the evidence is insufficient.
Pending reviewFeedback structure shapes revision behavior
Kim et al. report deeper revisions with structured peer review, while Morgan et al. associate actionable comments with substantive revision.
Kim et al., 2024Findings · p. 12
Students using structured peer review revised more substantively across three drafts.
Morgan et al., 2022Findings · p. 11
Actionable and specific comments were associated with more substantive revisions.
Make every relationship earn its place.
The Synthesis view is a review structure with explicit evidence links and conservative empty states.
Five distinct relationship types
Themes, agreements, disagreements, limitations, and gaps use different evidence rules instead of one generic summary prompt.
At least two papers per item
A retained cross-paper item points to evidence cells from multiple papers and shows each quote and page.
Fail-closed empty states
When verified evidence cannot support a defensible relationship, the category can remain empty rather than being filled with a weak claim.
Human review states
Accept supported items, mark them for revision, and regenerate when a source cell changes.
Plans
Synthesis is part of the core evidence workflow.
Free includes evidence-linked Synthesis with starter capacity. Researcher increases monthly papers, AI processing tokens, and private storage for recurring and larger reviews.
Compare Free and ResearcherQuestions researchers ask.
Does EvidenceGrid write literature review paragraphs?
No. It generates a reviewable structure of cross-paper items. Manuscript paragraphs remain a separate researcher-controlled writing stage.
Which cells are used for synthesis?
The generator uses verified matrix cells and their source quotes. Cells still marked for review or lacking matched evidence are excluded from trusted synthesis input.
How are disagreements identified?
A disagreement requires directly comparable findings with conflicting directions or conclusions. Different topics or a missing statement do not count as conflict.
Can I open the source behind a synthesis item?
Yes. Each evidence card links to its matrix cell, quote, page, and highlighted PDF source view.