Consensus

What is Consensus?

Consensus is an AI academic search engine and research workspace for discovering and understanding scientific literature. A user can enter a research question, keyword query, paper title, DOI, or author name, and the service retrieves relevant papers from its scholarly database.

Consensus then helps interpret the results through citation-backed summaries, comparisons, study snapshots, and broader literature syntheses. Its main value is reducing the time needed to locate relevant research and identify patterns across papers, while still giving users links back to the underlying sources.

How Consensus supports research

Academic search and evidence discovery

Search is the foundation of the product. Users can ask questions in ordinary language or use keywords, Boolean operators, author names, paper titles, and DOIs. Filters and academic search features help narrow results by characteristics such as study design, population, intervention, and outcome.

The service is particularly useful at the beginning of a literature review, when a researcher needs to map a topic, find relevant papers, identify foundational work, or see whether the literature generally agrees or disagrees on a question.

Citation-backed summaries and comparisons

Consensus can summarize findings across papers and compare studies by their methods, sample sizes, populations, interventions, and reported outcomes. Answers are connected to citations, allowing users to inspect the research behind a generated claim instead of treating the response as an unsupported chatbot answer.

The Consensus Meter is designed for suitable yes-or-no questions and provides a way to examine agreement or disagreement across the available literature. It should be treated as an aid to evidence review rather than a definitive verdict on a scientific question.

Research Agent and Deep Review

For more involved questions, Research Agent can plan and execute multi-step research tasks using capabilities such as academic search, citation crawling, DOI lookup, author search, and related-paper discovery. Deep Review is intended for larger literature syntheses and can examine approximately 50–100 papers for complex research questions.

These features are useful when a simple search result is not enough, but their output still requires checking. Automated synthesis can miss context, misread a study, or flatten important differences between research designs and populations.

Paper, document, and library workflows

Users can save papers and searches in My Library, organize collections, and analyze saved research. Consensus also supports conversations with individual papers, uploaded documents, and collections when full text is available. Citation exports and connections involving Zotero, BibTeX, RIS, EndNote, Paperpile, and Mendeley make it easier to move results into an existing reference-management workflow.

Full-text access varies by paper, publisher, and institutional access rights. A citation in the search results does not necessarily mean that Consensus can provide or analyze the complete paper.

Who is Consensus for?

Consensus is aimed at academic researchers, graduate and undergraduate students, clinicians, educators, librarians, analysts, and professionals who need evidence-based information. It can help with forming a research question, preparing a literature review, comparing interventions, locating supporting evidence, finding research gaps, and tracing citation relationships.

It is less suitable for users looking for a general-purpose conversational assistant, creative writing tool, coding environment, or image and video generator. A general AI search service such as Perplexity may cover a wider range of web research, while Consensus is more narrowly focused on scholarly literature and research evidence. For academic evidence synthesis, it occupies a more specialized position alongside tools such as Elicit.

Platforms and integrations

Consensus is available through a web application and mobile apps for iOS and Android. It also provides API and MCP access, with integrations or connected access for ChatGPT, Claude, and Microsoft 365 Copilot. These options matter for researchers who want to bring academic search into an existing assistant, reference-management, or organizational workflow rather than use the standalone web interface exclusively.

Pricing and access

Consensus has a continuing free tier. The free plan includes unlimited paper searches, 10 Pro messages per month, three Deep reviews per month, 10 Study Snapshots per month, and 30 API or MCP calls per month. The limits apply to deeper functions rather than basic paper discovery.

The Pro plan costs $20 per month, or $144 per year when billed annually. The Deep plan costs $65 per month, or $540 per year when billed annually. Pro and Deep include higher monthly allowances for Deep reviews and API or MCP calls. Teams and Enterprise plans use custom pricing, with enterprise-oriented administration and security options such as SSO/SAML, centralized account management, audit logs, and administrative controls.

Pricing and usage allowances can change, so researchers with heavy review or API requirements should check the current plan details before committing to a workflow.

Limitations and reliability considerations

• AI interpretation still needs verification: Summaries and comparisons can misinterpret source papers or omit important qualifications. Important claims should be checked against the original studies.
• Coverage and full text vary: The availability of papers and full-text content depends on the scholarly database, publishers, and access rights.
• Advanced features are metered: Research Agent, Deep Review, Pro messages, Study Snapshots, and API/MCP access have plan-specific limits.
• The product is specialized: Consensus is designed for scholarly research, not broad web assistance, creative production, software development, or general task automation.
• Model details are not fully transparent: Consensus documents the use of OpenAI, Anthropic, Google, and Baseten service providers for content processing, along with fine-tuned open-source models for selected features, but exact model selection varies by feature and is not fully disclosed.

Privacy and data handling

Consensus states that user queries, content, and personal information are not used to train large language models, including models operated by Consensus or third parties. Its privacy documentation says that service providers may process content to return results under contractual restrictions and not for model training.

The service may collect account, usage, payment, search, conversation, and uploaded-document information as applicable. Its privacy policy states that information is encrypted in transit and at rest. Users can clear thread history or delete their account, and the product offers incognito mode and enterprise options intended to support separate or no query logging. Organizations should review the current privacy and enterprise terms before uploading sensitive research material.

Is Consensus a good fit?

Consensus is a good fit when the central task is finding and synthesizing scientific papers, especially when citation traceability and study comparison matter. It can accelerate early-stage literature review work and provide a structured way to explore agreement, disagreement, and related research.

It is not a substitute for reading important source papers, assessing methodology, or applying expert judgment. Users who need broad internet research, general conversation, content generation, or unrestricted automated workflows will likely find its academic focus limiting. For evidence-focused researchers who understand those boundaries, its combination of scholarly search, citations, research agents, library features, and deeper review tools makes it more targeted than a general AI assistant.