Consensus is an academic search engine with an AI layer on top. You ask a question in plain language and it returns relevant peer-reviewed papers, along with a short summary of each. Its signature feature, the Consensus Meter, shows whether the studies it found tend to answer a yes-or-no question with yes, no, or a mix, which is a quick way to see where the weight of evidence sits.

There is a free level, with paid plans for deeper search and more features. Details change, so confirm on the Consensus pricing page.

What is different about it

Most AI assistants give you an answer and maybe a source. Consensus starts from the literature and shows you the studies first. For a question like whether caffeine improves athletic performance, the meter is a fast reality check, and each paper card lets you open the abstract or the full text.

You can also ask a follow-up about a single paper and export citations in formats such as RIS, BibTeX, and CSV, which saves fiddling when you draft a report. Seeing the papers first, rather than just an answer, matters when you are deciding whether a claim is worth repeating.

Good ways to use it

Use it early, to find out whether a topic is even settled, and to collect the handful of key papers worth reading. Use the meter to avoid over-claiming: "some evidence suggests" is often the honest summary, not "science proves". Then read the papers themselves for methods and sample sizes, because a big study and a tiny one agree on the screen but not in weight. When a result surprises you, check whether the studies are recent and whether they studied people like the ones you care about.

What to watch

The summaries are generated and can simplify away important caveats, and the meter reflects only the papers it retrieved, not the whole literature. For health, legal, or financial decisions, treat it as a starting point and consult a professional. It is strongest in English-language science, less so for Chinese sources.

For building a structured review, our Elicit guide goes further; to organize papers you already have, try NotebookLM.