Deep Review: let an agent do your literature search
For questions that need more than one search query, Deep Review builds a research strategy, runs it, and hands you a fully-cited briefing.
Some questions don't have a single-paper answer. "What does the current evidence say about psychedelic-assisted therapy for treatment-resistant depression?" — that touches clinical trials, mechanistic neuroscience, regulatory guidance, and meta-analyses. You'd need 5-6 different searches to cover it properly.
Deep Review does those searches for you. It takes your question, breaks it into sub-questions, runs each one against our 300M+ paper index, reads the best hits, and writes you a structured briefing with inline citations. Basically a tireless research assistant that reads fast and always shows its sources.
The process
- Breaks your question into 4-8 targeted sub-queries, each capturing a different angle.
- Runs each sub-query with different filter strategies (RCTs for one, reviews for another, etc).
- Scores results by relevance, recency, citation count, and study quality.
- Reads abstracts and full text (where open-access) for the top hits.
- Writes a structured briefing organised by theme, with every claim cited inline.
You can see everything it's doing
This isn't a black box. You see every sub-query as it runs. You see which papers it's considering. You can step in at any point — refine a query, exclude something irrelevant, or ask a follow-up about a specific finding. The final output is a briefing you can audit, not a magic answer you have to trust.
It's especially useful for interdisciplinary questions, for mapping fast-moving research areas, or for getting up to speed on something outside your main field. Finds connections across subfields that you'd miss searching within your usual journals.
Run a Deep Review.
Try Deep Review →