AI literature review
AI literature review without hallucinated citations.
Most "AI for literature review" tools invent papers that don't exist. Symmathy doesn't. Every answer is grounded in records returned from a federated search across PubMed, Europe PMC, Cochrane, ClinicalTrials.gov, LILACS, J-STAGE, HAL, DOAJ, OpenAlex, and CrossRef — and every sentence links back to the source.
How source-grounded AI lit review works
- Search runs first. Your query hits 10+ biomedical databases in parallel. Duplicates are merged across sources.
- AI reads only what was retrieved. The model sees real records — titles, abstracts, full text where available — and nothing else. It cannot cite a paper that wasn't returned.
- Every claim is linked. Each sentence in the answer carries an inline citation pointing to the original source. You can click through and read the paper.
- Non-English sources are translated inline. LILACS (Spanish/Portuguese), J-STAGE (Japanese), HAL (French), ThaiJO (Thai) — translated on retrieval so they're searchable and readable.
What AI lit review tools should not do
We don't grade studies, rank evidence quality, or tell you which paper is "best." That's your clinical and methodological judgment. We surface the literature, link to sources, and stay out of the interpretation.