Consensus is a search engine for scientific literature. You ask a question, and instead of answering from a language model's memory, it searches peer-reviewed papers and reports what the research actually says — including the useful cases where studies disagree, which a plain chatbot tends to smooth over. It also surfaces the papers themselves, so every claim has a source you can open.

That grounding makes it far more useful than a general assistant for evidence-based questions: whether a supplement works, what the literature says about a habit, whether a claim is supported. The consensus meter, which summarizes the balance of findings across studies, is a genuine shortcut for judging how settled a question is.

The limits follow from its scope. It only knows what is in its indexed literature, so questions outside scientific publishing — business, current events, practical how-tos — are the wrong fit. Coverage also varies by field, with some disciplines far better represented than others. And it is a starting point for research rather than a conclusion: it summarizes studies without assessing their quality, and reading the underlying papers remains the work.

The free tier is unusually generous for this category, with unlimited paper searches and monthly allowances for deeper reviews and snapshots, while paid plans expand the synthesis features.

A practical way to use it: start with the consensus summary to get your bearings, then open the papers behind the strongest findings and read the methods and limitations yourself. That sequence is faster than searching a database cold and more honest than trusting a single summary. For anything you intend to act on — a health decision, a claim in published work — the citations are not decoration; they are where the actual answer lives.

Best for researchers, students, and anyone evaluating scientific claims. Not for non-scientific questions, niche fields with thin literature coverage, or work where the answer must be authoritative rather than summarized.