AI Inference Level — how much interpretive distance sits between the source text and what you see: Non-generative shows structure only, no AI claims Extractive traceable to source Structured synthesis organized, still checkable Abstractive generated verdict, verify before trusting

Platform	Core Workflow	Extraction & Gap Features	AI Inference Level	Free Tier Reality for Students
Elicit Best for Data Matrices	Systematic extraction of study parameters into a structured matrix.	Custom columns for methodology, population, settings, and limitations. Highly reliable extraction directly from PDFs.	Extractive Cell values are pulled directly from the PDF and traceable back to their location in the source — low risk of ungrounded claims.	Strict Limits: Generous one-time credits, but heavy extraction burns through them fast. Paid plans are steep for casual use.
SciSpace Best for PDF Interrogation	Exploratory search across 280M+ papers and deep conversational PDF analysis.	Custom extraction tables similar to Elicit, plus line-by-line copilot explanations of methodology and equations.	Extractive + light synthesis Table extraction stays traceable, but the copilot's line-by-line explanations add a layer of generated interpretation — check those against the original text.	Generous, very robust free tier for reading and extraction; highly accessible for budget-conscious students.
Paperguide Best All-in-One	Bridges the gap between extracting paper data and actually writing the literature review.	Systematic column extraction, integrated citation manager, and an AI drafting assistant all in the same workspace.	Structured synthesis Extraction itself is fine, but the built-in AI drafting assistant can generate review prose directly — the highest-risk feature here for skipping your own inductive reasoning.	Generous: Offers more free extraction columns than Elicit, plus daily credits for drafting and AI interaction.
Consensus Best for Claim Verification	Rapidly checks the scientific consensus on specific yes/no research questions.	Synthesizes claims across millions of papers to instantly highlight whether the literature is mixed or leaning in a certain direction.	Abstractive Delivers a generated yes/no verdict across papers up front — easiest tool on this list to accept uncritically without opening a single source.	Moderate Free tier includes a set number of deep AI syntheses per month.
ResearchRabbit Best for Visual Mapping	Visualizing citation networks to find hidden or isolated subfields.	Does not extract text tables, but builds interactive graphs showing the temporal evolution of citations to spot missing links.	Non-generative No prose synthesis at all. The tool shows structure; you're the one who has to notice and name the gap — good for building the skill, not just the answer.	Completely Free Free for academic use.
Litmaps Best for Visual Mapping	Tracking how an idea spreads through the literature over time.	Overlays citation networks on a chronological timeline, making it obvious when a research topic went dormant.	Non-generative Dormancy is visually implied on the timeline, never asserted as an AI claim — you still draw the conclusion yourself.	Generous Excellent free tier for basic mapping and discovery.

AI Literature Review Platforms for Gap Analysis