Consensus

Consensus is an AI academic search engine and research workspace that sits on top of more than 250 million peer-reviewed papers, including licensed full text from major publishers such as Taylor & Francis, SAGE, the American Chemical Society, and the AAAS Science family of journals. Instead of generating an answer from model memory, every response is tied back to real studies, so you can see which paper supports which sentence. The product ships several ways of interrogating the literature. Plain Papers search is free and unlimited, returning ranked studies with abstracts, journal names, citation counts and study metadata. Pro Analysis uses AI to summarize key findings from the top papers in a search and produce a high-level synthesis with inline citations, typically drawing on up to 20 papers per query. Deep Search, sold on its own Deep plan, automates a full literature review: it builds a search strategy, expands key terms, follows the citation graph, surfaces conflicting findings, and works across up to 50 papers in one pass, turning days of manual screening into minutes. The Consensus Meter answers yes-or-no research questions by classifying papers as supporting, opposing or neutral and visualizing the balance of evidence, which makes it obvious when a field is split rather than settled. Medical mode narrows results to roughly 50,000 clinical guidelines and about 8 million articles from the top 1,000 medical journals. Natural-language filters let you specify timeframes, populations and study designs directly in your prompt, so Consensus applies the right filters without a structured query builder. Supporting tools include Study Snapshot, which extracts methodology, sample size, duration and outcomes from a paper; Chat with Full Text for asking questions across a paper, a collection or your own uploads; Citation Grounding, which shows the exact quoted passage and highlights it in the PDF; My Library and shared collections for saving papers; Threads for follow-up questions; a Research Agent that plans and executes multi-step tasks; and a Research Gaps Matrix that maps under-covered combinations of population and method. Consensus also runs inside ChatGPT, Claude and Microsoft 365 Copilot, and its API and MCP server are out of beta with a self-serve dashboard, so developers can wire the same corpus into their own agents.