Skip to content
NakodaAI

Consensus

Consensus

An AI search engine over 220 million peer-reviewed papers, cited sentence by sentence.

Visit official site (opens in a new tab)
  • Research
  • Education

What it does

An AI-powered academic search engine that reads peer-reviewed literature and answers a research question in plain language, citing the specific papers each sentence of its answer drew from.

The interesting part

Its architecture splits the work across four coordinated agents — planning, searching, reading and analysis — and it now ships an MCP server, so the same 220-million-paper index can be queried directly from Claude Desktop, Claude Code or another MCP client instead of only through its own web app.

Who it is for

  • Researchers and students who want a synthesized, cited answer instead of a list of search results
  • Clinicians and policy staff who need to check what the evidence base actually says
  • Anyone already working inside an MCP-compatible AI tool who wants cited literature search without leaving it

Who should not use it

  • Anyone needing full-text access to papers behind a paywall — Consensus indexes and cites, it doesn't unlock subscriptions
  • Researchers who need a fully manual, unfiltered search rather than an AI-synthesized answer

A typical use case

A policy analyst asks whether a specific public-health intervention shows a consistent effect across studies, and gets a synthesized answer with per-claim citations back to the underlying papers.

Limitations

  • Synthesis quality depends on how well a question is scoped — vague questions return vaguer answers
  • Coverage is peer-reviewed literature specifically, not grey literature or working papers
  • Deeper features (Deep Search, Scholar Agent) are gated behind paid tiers

Pricing

Free tier available with limited searches; paid plans unlock Deep Search and higher usage — see official pricing.

Privacy & data

Partnered with major academic publishers (Wiley, AAAS, Sage, APA, ACS and others) for licensed index access; see its own privacy policy for account and query data handling.

Integrations

  • MCP server for Claude Desktop, Claude Code and other MCP clients
  • ChatGPT Deep Research connector

Alternatives

Elicit

Similar goal, with a stronger focus on structured data extraction from individual papers.

Why Nakoda noticed it

Most 'AI search' tools for research quietly summarize from a model's training data. Consensus's entire premise is the opposite — every sentence traces to a specific paper — and its 2025 MCP release means it's now infrastructure other AI tools can call, not just a destination site.

Provenance

Verified · last checked 2026-09-24

Nakoda read the product's own site directly and confirmed the facts on this page against it.

Sources checked

← Back to all products