Consensus

An AI academic search engine for peer-reviewed literature. You can retrieve evidence from 250 million papers by asking questions in natural language, visually present the consensus and divergence of research conclusions, and support Deep Search automated reviews and medical modes.

AI Search Engine#Academic Search#Evidence#Literature

Updated: 2026-10-26

What is Consensus?

Consensus is an AI academic search engine for peer-reviewed literature, officially positioned as "Research Starts Here". Unlike ordinary academic search tools, it allows users to ask a direct question in one sentence. The system searches for evidence in more than 250 million papers and intuitively shows whether the research community "mostly supports" a certain question or "has disagreements" on it. Therefore, it is often hailed as the standard configuration for a new generation of academic research. According to its official website, 10 million researchers, students and clinicians are already using it, and more than 170 university libraries have partnered with it to provide access for students and faculty. Its coverage includes full-text content licensed from major publishers.

Who is it for?

Graduate students and researchers who need to quickly determine "what the scientific community thinks about a certain issue"; evidence-based medical practitioners such as clinicians and pharmacists; students who need to find evidence to support their papers or reviews; and analysts who need to cite authoritative peer-reviewed literature in their reports. It is especially user-friendly for university users—if your library has partnered with Consensus, you can directly use it through institutional access without personal payment.

Key features

Natural Language Querying

No need to construct search queries; start your search by describing your research question in a conversational sentence. For example, simply ask "What is the impact of intermittent fasting on metabolic health?" and the system will automatically match relevant papers.

Consensus View

The search results present the distribution of the research community's attitudes toward the question, allowing you to quickly see whether the evidence is consistent and avoid one-sided conclusions from only reading a single paper. This is Consensus's most unique selling point.

Deep Search Deep Search

Automates a single round of literature review — the system independently formulates search strategies, screens and synthesizes multiple research papers, and outputs organized review results, making it ideal for quickly getting to grips with an unfamiliar research area.

Natural Language Filtering

Describe filtering criteria in everyday language (such as "within the past five years" or "randomized controlled trials") without manually combining fields and filters, reducing the threshold for retrieval.

Medical Mode

A dedicated medical search mode designed for clinical and evidence-based practice scenarios, helping healthcare-related researchers quickly locate high-quality evidence.

Draft Reports

Based on retrieval and analysis, it organizes evidence into a draft report that can be directly edited further, shortening the distance from searching for materials to writing a paper.

How to use Consensus

Treat it as a "starting point for finding evidence" rather than an endpoint: first use the consensus view to judge the direction of the conclusion, then use Deep Search to quickly conduct a round of literature scanning, and finally return to the original text to conduct in-depth reading of key papers.
University users should first check whether their library has activated institutional access, which can save a lot of costs.
developing the habit of "using abstracts to determine the direction and original texts to determine the conclusion" in daily retrieval will make the research process more rigorous.

Things to know

The consensus view and AI summaries are "general signals", and the rigorous conclusions of the papers shall still be subject to the original text. Before formal citation, be sure to open the original text to check the research design, sample size and statistical methods.

Although the search coverage reaches 250 million papers, not all full texts are available; for some literature, you can only view the abstract or obtain the full text through institutional subscriptions.

Free usage usually has limits on the number of times or available functions. For high-frequency use, it is recommended to first confirm whether your institution has purchased it and then evaluate whether you need a paid tier.

AI-generated content may contain occasional errors; in critical scenarios such as clinical decision-making, do not rely on a single tool alone for judgment.