Consensus

The meter is the product

Consensus answers research questions from peer-reviewed literature instead of the open web. Type a question, and it searches a corpus of 200M+ papers, reads them, and returns a synthesized answer with a Consensus Meter — the share of papers that support, oppose or are mixed on the claim — plus citations back to the individual studies. Where a general chatbot might summarize whatever it finds online, Consensus is bounded to published research, which is the point: for evidence questions, the answer should come from papers, not blog posts.

The meter is the product's core idea: instead of one AI's opinion, you get the distribution of what published studies actually found, with the underlying papers one click away. For questions where the literature is thin or contradictory, the meter honestly shows "Mixed" or a small N rather than manufacturing certainty — that transparency is the feature.

A search in practice

A search for a typical research question — Does regular physical exercise improve memory in older adults? — returns an answer and the evidence behind it:

Yes, with the strongest evidence for aerobic exercise. Consensus Meter, N = 18 papers: 89% Yes · 6% Possibly · 6% Mixed · 0% No

The answer text is grounded in that distribution — "benefits are most consistent for aerobic exercise, while resistance, mind-body and multicomponent exercise also help, though effects vary by memory type and dose" — and each claim links to the studies behind it. The search also surfaces related queries with paper counts (physical activity and episodic memory: 9.2M results) that hint at the size of each evidence base. That structure is what separates it from a chatbot: the answer carries receipts, and the receipts are the point.

Free to ask, paid to go deep

The core search — asking questions and reading meter results — is free. The paid tier expands the workflow around it: more searches per month, AI-powered paper summaries, deeper filters (study type, journal, sample size), and features like building comparison tables across papers. Pricing starts around $9 a month. The honest split: a student or researcher asking a handful of evidence questions a week may never need the paid tier; someone running systematic reviews or building evidence tables across dozens of papers will hit the free limits and find the paid tier pays for itself in the bibliography time it saves.

Who should trust it

• ✓ Works for: students and researchers who need evidence-based answers with citations rather than AI summaries; anyone checking "what does the research actually say" on a health, psychology or social-science question; writers and policy people who need to cite real studies, not plausible ones.
• ✗ Not a fit for: open-ended general knowledge, where a search engine is faster and broader; cutting-edge topics with thin literature, where the meter will honestly show a small N; anyone needing deep full-text analysis across huge corpora on the free tier, which is where the paid plan becomes necessary.