Hacker News new | ask | show | jobs
Consensus Deep Search – AI that reads 1,000+ papers for you like a researcher (twitter.com)
11 points by aakashadesara 351 days ago
4 comments

Consensus Deep Search is a research agent that produces full literature reviews across 200M academic papers—in under two minutes.

It breaks down complex questions, conducts structured multi-step searches, and returns AI-generated reports with citations, visuals, and evidence mapping.

Key features:

- Structured literature search across 6,000+ scientific domains

- Analyzes and filters 1,000+ papers per query

- Outputs a cited report with intro, methods, results, and discussion

- Visualizes research consensus, key authors, and knowledge gaps

Built to accelerate discovery by turning questions into high-quality answers—fast.

Just tried it - this is pretty cool. How is it generating the search strategy? There were some questions in there that I hadnt thought of
The search strategy is generated by decomposing your initial query into sub-questions using custom prompting and retrieval heuristics— it tries to surface angles a researcher might ask but you might not think of initially.
Thats cool, yea I really like how you can actually see all the questions its asking. I'm actually learning alot through that on improving my own searches
what databases does this search over?
Same question ^ is it just research papers? If so, thats really helpful. How does it compare to google scholar?
powered by multiple databases including Semantic Scholar, PubMed, and others
pretty sick - is the student discount just for undergrad or masters/phd too?
Looks like you just need a .edu so I'm assuming phd also