Skip to Content
Overview

Knowledge Raven

Knowledge Raven is not a competitor to Confluence, Notion, or GitHub MCPs — those are data sources. Knowledge Raven is the aggregation layer that connects all your tools and makes their knowledge searchable through one unified interface for any AI agent.

Connect your company’s knowledge sources once. Any AI agent — Claude, ChatGPT, Gemini, Cursor — instantly searches everything via the Model Context Protocol (MCP).

What is Knowledge Raven?

Knowledge Raven is a SaaS platform that centrally stores and indexes your company knowledge, making it accessible to AI agents through a standard MCP interface. Instead of building separate integrations for each knowledge source, you connect once to Knowledge Raven and your agents can search everything.

How it works:

  1. Connect your knowledge sources: Confluence spaces, Notion pages, GitHub repositories, Dropbox folders, Google Drive
  2. Knowledge Raven indexes all content automatically — text extraction, chunking, embedding, and hybrid search are fully handled for you
  3. Your AI agent searches through all connected knowledge using 6 MCP tools with intelligent query routing

No embedding knowledge required. No vector database to manage. No pipelines to maintain.

Key Differentiator: Agentic RAG with 3 Query Types

Most knowledge tools offer basic vector search. Knowledge Raven uses Agentic RAG — the agent dynamically selects the optimal retrieval strategy per query:

Query TypeStrategyWhen Used
PrecisionVector-heavy (70% semantic + 30% BM25)Specific questions with known keywords
ExplorativeBM25-heavy (30% semantic + 70% keyword)Broad topic exploration
DiscoveryMetadata-onlyBrowsing available documents without search

This means your agent finds relevant knowledge even when queries are vague, exploratory, or cross-document — not just when they contain exact keyword matches.

MCP Tools

6 tools give your agent full control over retrieval:

  • search_knowledge_base — Precision search within a specific knowledge base
  • broad_search — Explorative search across all knowledge bases
  • fetch_document — Retrieve a specific document (preview, full, or by chunks)
  • list_knowledge_bases — Discover available knowledge bases
  • list_documents — Browse documents with summaries
  • get_document_metadata — Retrieve metadata without loading full content

Full MCP Tools Reference

Connectors

ConnectorStatus
ConfluenceLive
NotionLive
GitHubLive
DropboxLive
Google DriveOAuth verification pending

Connector documentation

Pricing

PlanDocumentsUsersQueriesPrice
Free503100 / user / month$0
Pro50015Unlimited$29 / workspace / month
EnterpriseUnlimitedUnlimitedUnlimitedCustom

All plans include all connectors and all 6 MCP tools. Upgrade anytime from your workspace settings.

Supported Formats

Knowledge Raven uses Gemini Embedding 2 — a natively multimodal embedding model. Supported content types:

  • Text: PDF, DOCX, TXT, Markdown, CSV
  • Images: PNG, JPEG (via multimodal embedding — no separate OCR pipeline)
  • Audio: MP3, WAV, M4A (via multimodal embedding)
  • Video: MP4, MOV up to 120 seconds (via multimodal embedding)

Quick Start

Ready to connect your first agent? Choose your client →

Last updated on