Consensus logo
Academic Research · Rag Semantic Search

Consensus

Consensus is an AI academic search engine that answers research questions from more than 220 million peer-reviewed papers, returning cited syntheses, study snapshots and an evidence agreement meter rather than an unsourced answer.

Active Free plan Freemium API available 13+ Verified by Guidaio
Overview

What is Consensus?

Consensus is an AI academic search engine built on a corpus of more than 220 million peer-reviewed papers. Rather than returning a ranked list of links, it takes a research question written in ordinary language and answers it with a synthesis drawn from the literature, each claim tied back to the study it came from. Its publisher, Consensus NLP, Inc., positions the product as an AI workspace for scientific research — a research operating system rather than a search box.

Two search depths sit at the centre of the product. Pro Search analyses up to twenty papers and returns a summarised, cited answer; Deep Search, also called Deep review, widens the sweep to fifty papers and is aimed at genuine literature reviews. Around them sit several reading aids: the Consensus Meter, which shows how far published evidence agrees on a yes/no question; Study Snapshot, which surfaces a paper's methodology, sample size and duration; Ask Paper, which lets you question a single article in conversation; and a Citation Graph mapping how papers relate to one another.

Results can be filtered in ways that matter to researchers rather than to general web users: by study type — meta-analysis, systematic review, randomised controlled trial, observational study, literature review — and by citation count, SJR journal quartile, publication year, open-access status and whether preprints are included. Findings are kept in bookmarks and custom lists, and citations export to RIS, BibTeX and CSV.

Beyond the web app, Consensus ships an API, available by application only and priced from $0.10 per call plus a platform fee, and a MCP server alongside a Consensus connector for ChatGPT, so the same searches can run inside Claude, inside ChatGPT or inside an internal tool.

The company was founded in 2021 by Christian Salem and Eric Olson, began fully remote and is now based in San Francisco. It reports more than 10 million researchers, students and clinicians among its users, 12,500+ universities represented, and over 150 million research questions handled to date.

What it does

  • Ask a research question in plain language and get an answer built from peer-reviewed papers
  • Run a literature review across up to 50 papers with Deep review
  • Measure how far published evidence agrees on a yes/no question with the Consensus Meter
  • Extract a study's methodology, sample size and duration with Study Snapshot
  • Interrogate a single paper in conversation with Ask Paper
  • Organise findings into collections and export citations to RIS, BibTeX or CSV
  • Plug peer-reviewed search into ChatGPT, Claude or an internal tool via MCP and the API
Audience

When to use Consensus / When not to

A quick filter to help you decide if Consensus is the right fit.

When to use Consensus

  • Graduate students, PhD candidates and postdocs running literature reviews
  • Faculty and academic researchers who need answers tied to peer-reviewed sources
  • Clinicians and healthcare professionals looking for evidence-based findings
  • R&D, biotech and pharma teams scanning the literature for discovery work
  • Science journalists and fact-checkers verifying claims against primary studies

When not to use Consensus

  • Writers who expect the tool to draft an essay or a paper for them
  • Anyone after general web results, news or non-academic content
  • Patients seeking medical advice rather than research literature
  • Teams that need an interface or processing language other than English
  • Organisations requiring a signed DPA, a published subprocessor list or EU data residency
Get started

How to use Consensus

A typical end-to-end flow, from setup to results.

  1. Open consensus.app and start searching, or create an account through the sign-up page
  2. Sign in directly with a Google or Apple account if you prefer not to set a password
  3. Type a research question in plain language instead of a keyword query
  4. Read the cited answer and the list of peer-reviewed papers behind it
  5. Switch to Pro Search for a synthesis of up to 20 papers, or Deep review for up to 50
  6. Check the Consensus Meter when your question can be answered yes or no
  7. Narrow the results by study type, citation count, SJR quartile, year, open access or domain
  8. Open Study Snapshot for methodology and sample size, or Ask Paper to question an article
  9. Save what matters to bookmarks and custom lists, then export citations as RIS, BibTeX or CSV
  10. For programmatic use, apply for API access or connect the MCP server to ChatGPT or Claude
Quick read

Pros & Cons

Pros

  • Answers are drawn exclusively from peer-reviewed literature and always carry citations
  • Explicit, repeated commitment never to train language models on user data, in-house or third-party
  • No sale of personal information and no monetisation of user data, on a stated subscription-only model
  • Permanent free plan with unlimited searches, not a time-limited trial
  • Methodological filters that are rare elsewhere: study type, SJR quartile, citation count, preprints
  • Native connectors for ChatGPT and Claude through MCP, plus a documented search API
  • Broad adoption reported: 2.5 million monthly active users and 12,500+ universities represented

Cons

  • No mention of the GDPR anywhere, and no Article 27 EU representative designated
  • No Data Processing Agreement published and no subprocessor list disclosed
  • No security certification displayed — neither SOC 2, ISO 27001, HIPAA nor FERPA
  • Data hosted solely in the United States, with US jurisdiction accepted by using the service
  • No postal address published for the company, and no legal notice page
  • No iOS or Android app, and an English-only interface
  • Pricing page renders in JavaScript, so amounts are not readable without a browser
Pricing

Pricing & Plans

Consensus offers a permanent free plan at $0, which includes unlimited searches together with monthly quotas of Deep Searches, Pro Analyses, Study Snapshots and Ask Paper messages. The lowest paid entry point is the Pro plan at USD 20.00 per month, or USD 144 per year when billed annually, which is equivalent to USD 12 per month. A Deep plan is listed at USD 65 per month, and the Teams and Enterprise plans are quoted individually according to team size. API access is priced separately, from USD 0.10 per call plus a platform fee, and is granted by application only.

Plan 1
  • Free — USD 0
Plan 3
  • Deep — USD 65 per month or USD 540 per year
Plan 4
  • Teams — custom pricing
  • volume discounts up to 200 seats
Plan 5
  • Enterprise — custom pricing for universities and large organisations
Special offers — Student and clinician discount, documented in the Consensus help centre and requiring a verified academic email address · Annual billing brings the Pro plan to USD 144 per year, the equivalent of USD 12 per month · Volume discounts on the Teams plan for up to 200 seats
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how Consensus handles your data.

GDPR overview

There is no mention of the GDPR anywhere on the site. The word does not appear in the Privacy Policy dated 15 May 2026, in the Terms of Service dated 25 June 2025, or in any other page collected. No Article 27 EU representative is designated, no Data Processing Agreement is published or offered, no subprocessor list exists, and no legal basis for processing is stated. The policy is written under United States law: it addresses California privacy rights while declaring the company outside the scope of the CCPA, and it states that the company is headquartered in the United States with all servers located there, users consenting to processing under US jurisdiction. Rights to review, amend or delete personal data are offered to every user on request to the privacy address, but as a contractual courtesy rather than as GDPR data-subject rights.

Who owns the data?

The Terms of Service state that the site, its content, code and intellectual property belong to Consensus NLP, Inc. or its licensors, and that users receive only the limited rights the Terms grant. The Terms contain no clause assigning ownership of the questions, prompts or content a user submits, which the Privacy Policy classes as Service Interaction Information alongside the output generated for that user. The publisher states plainly that it does not sell personal information and does not monetise user data in any form. Access is restricted to staff with a legitimate business need and to business-critical vendors bound by enterprise agreements. Users may ask to review, change or delete their personal data. In a merger or acquisition, data may transfer to the successor, which remains bound by the same policy.

Reuse rights

Consensus grants only a limited right to use the site; it does not hand users a reusable licence over the platform, its content or its code, and the Terms reserve all other rights to the company and its licensors. Search results point to third-party peer-reviewed papers, so reuse of any paper remains governed by that publisher's own licence, not by Consensus. In the other direction, the publisher commits to narrow use of what users put in: information is used to run and secure the service, manage accounts, communicate, personalise the experience, analyse usage on aggregated and de-identified data, and meet legal obligations. Three uses are explicitly ruled out — selling personal information, training large language models on user data, and passing user data to third parties for model training, a commitment the policy extends to every vendor relationship. Data drawn from Google APIs is handled under the Google API Services User Data Policy, Limited Use included. The company does not honour Do Not Track signals.

Data retention & training

Retention summary
No retention period is published. The Privacy Policy says only that personal and non-personal information may be kept for as long as is reasonably necessary for a legitimate business purpose, and that the company may delete it at any time unless a signed agreement or the law requires otherwise. There is no retention schedule by data category and no maximum duration. Users may ask to review, change or delete their personal data by writing to the privacy address with the subject line "Your Personal Information". Anyone under 18, or their parent or guardian, may request removal of content about them, using the subject line "Removal of Minor Information"; the policy warns that such removal does not guarantee complete erasure.
Trains on customer data
No
GDPR contact

Hosting summary

The Privacy Policy is explicit: the company is headquartered in the United States and all of its servers are located there. Any personal information provided is therefore accessed or transferred to the United States, and by supplying it a user consents to storage and processing there. The policy also asks users to acknowledge that the data becomes subject to United States law, including the ability of US governments, courts and law enforcement or regulatory agencies to compel disclosure. No alternative hosting country or region is offered, and no data residency option is advertised. Information is described as encrypted both in transit and at rest using industry-standard protocols, with SSL named in the security section. Note that the domain resolves to a Cloudflare anycast address, which identifies the content delivery network rather than the location where data is stored. The governing law of the Terms of Service is that of Delaware, with arbitration seated in Kent County, Delaware.

Hosting countries
🇺🇸 United States
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting Consensus.

  • Treating a synthesised answer as a conclusion, when the underlying papers may disagree or be weak
  • Letting the Consensus Meter stand in for reading the studies, flattening nuance into a yes/no verdict
  • Erosion of the skill of searching and appraising literature unaided, particularly among students
  • Assuming coverage is exhaustive: a paper outside the indexed corpus simply will not appear
  • Using research findings as clinical or legal advice, which the tool is not intended to provide
  • Sending confidential or unpublished research material into a service hosted entirely in the United States
  • Relying on privacy commitments that are policy statements rather than audited or certified controls
Setup

Setup & Integrations

Technical difficulty

Very low for normal use. The web application needs no installation: create an account, or sign in with Google or Apple, and start searching. No technical skill is required. Connecting the MCP server is described as a three-step process, and adding the Consensus connector in ChatGPT means finding it under Apps and verifying your account. Only the API demands real engineering effort, and it is gated: access is granted by application through a form, after which a developer integrates it using the published documentation.

Deployment

Web appAPIPlugin

Integrations

ChatGPT Claude

Supported languages

English
Company

Behind Consensus

Company name
Consensus NLP, Inc.
Founded
24/08/2019
Country of origin
🇺🇸 United States
UBO
INFORMATION_NOT_FOUND
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇺🇸 United States
Support contact

Fundraising

11/05/2026 — USD 30 million round led by GreatPoint Ventures, whose partner DJ Patil, former US Chief Data Scientist, joined the board
Existing investors taking part in the same round: Union Square Ventures, NFDG and Draper Associates
At the time of the announcement the company reported 2.5 million monthly active users

Social

Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement Consensus.

E ElicitG Google Scholar
FAQ

Frequently asked questions

What does Consensus search?
A corpus of more than 220 million peer-reviewed scientific papers. It is an academic search engine, not a general web search tool, so every answer is built from published research.
Is there a free version?
Yes. The Free plan costs $0 and is permanent rather than a trial. It allows unlimited searches, with monthly quotas of 3 Deep Searches, 15 Pro Analyses, 10 Study Snapshots and 10 Ask Paper messages.
How much does a subscription cost?
The Pro plan is USD 20 per month, or USD 144 per year when billed annually. A Deep plan for heavy users is listed at USD 65 per month. Teams and Enterprise plans are quoted individually.
Will my data be used to train AI models?
No. The Privacy Policy states that queries, content and personal information are never used to train AI models, neither the publisher's own nor those of any third party, and that no user data is sent to third parties for that purpose.
Where is my data hosted?
Exclusively in the United States. The Privacy Policy states that the company is headquartered there and that all its servers are located there, and that using the service means consenting to processing under US law.
Is Consensus GDPR compliant?
The site never mentions the GDPR. No EU representative is designated, no Data Processing Agreement is published and no subprocessors are listed. European organisations should treat compliance as unverified and ask the publisher directly.
Is there an API?
Yes, a quick search API returning 20 papers per query with metadata and relevance scores. It is available by application only, with baseline pricing from USD 0.10 per call plus a platform fee and a custom quote.
Can I use Consensus inside ChatGPT or Claude?
Yes. A Consensus connector can be added from the Apps section of ChatGPT, and a Consensus MCP server connects the same peer-reviewed search to Claude and other MCP-capable assistants.
Is there a minimum age?
You must be at least 13 years old. Anyone under the age of majority in their jurisdiction but at least 13 needs a parent or guardian to accept the Terms on their behalf.
Is there a mobile app?
No iOS or Android application was found on the site. Consensus is delivered as a web application, together with its API and MCP server.
Conclusion

Should you pick Consensus?

Consensus is a mature, clearly scoped research tool. By restricting itself to peer-reviewed literature and citing every claim, it answers a question that general-purpose assistants answer badly, and its methodological filters — study type, citation count, journal quartile, preprint handling — show that it was designed by people who understand how researchers actually judge a source. The reported scale is substantial: 2.5 million monthly active users, 12,500 universities represented, a USD 30 million round raised in May 2026.

Its privacy posture is unusually explicit for the category. The publisher states that it never trains language models on user data, never passes that data to third parties for training, and never sells personal information, describing itself as fully subscription-funded. Taken at face value, that is a stronger commitment than most competing tools make.

The weakness is compliance transparency rather than product quality. The GDPR is never mentioned, no EU representative is designated, no Data Processing Agreement is published, no subprocessors are listed and no security certification is displayed. All data sits in the United States, and no postal address for the company appears anywhere on the site. A European institution planning a campus-wide deployment will need answers the website does not provide.

For an individual researcher, student or clinician, the free plan makes the tool easy to try and the USD 20 per month Pro plan is priced in line with the market. For an institutional buyer, the product deserves a look but the contractual paperwork deserves a conversation first.