Summary
An encrypted Markdown knowledge base that connects notes to AI tools through a permission-controlled MCP server.
About
Granite combines encrypted, cross-device note management with controlled access for AI assistants. Its vault permissions, source citations, and team administration features position it for individuals and organizations seeking to use AI with private knowledge bases.
Granite is a Markdown-based note-taking and knowledge management tool with separate encryption keys for each vault and synchronization across phones, tablets, desktops, and the web. Its built-in MCP server allows Claude, ChatGPT, and Cursor to search, read, and update notes in authorized vaults, with responses citing the source note. Features include wikilinks, backlinks, a knowledge graph, offline search, live co-editing, and team controls including SSO, SCIM, and audit logs. The product states that user data is not used to train AI. It offers a free plan and a Pro plan priced at $3.99 per month, with no AI surcharge.
Key Features
Markdown notes encrypted with a separate key for each vault
Synchronization across phone, tablet, desktop, and web
MCP server for authorized AI search, reading, and note updates
Answers that cite the exact source note
Wikilinks, backlinks, graph view, offline search, live co-editing, SSO, SCIM, and audit logs
Real-World Use Cases
Personal knowledge management
Organize Markdown notes across devices and use supported AI assistants to retrieve or update information in selected vaults.
Team knowledge access
Maintain shared notes with live co-editing and administrative controls, while limiting AI access to approved vaults.
Source-grounded AI research
Search an existing note collection with AI and review cited notes to verify the basis of responses.
Business Opportunities
Secure AI access to internal documentation
Problem: Teams may want AI-assisted retrieval from internal notes but need to control which information assistants can access.
Solution: Use vault-level permissions and the MCP server to let supported AI tools search, read, and update approved notes, with source citations.
Estimated ROI: Potentially moderate through faster information retrieval and reduced manual searching; actual returns depend on adoption, content quality, and workflow fit.
Execution: Organize relevant documentation into vaults, define access boundaries, connect a supported AI assistant through MCP, and establish team administration practices using available controls.
Privacy-conscious personal knowledge workflows
Problem: Individuals may keep useful notes across devices but lack a controlled way to use them with AI tools.
Solution: Store notes in encrypted vaults and grant AI assistants access only to selected vaults for search and updates.
Estimated ROI: Potentially modest through improved retrieval and reuse of existing notes; value depends on the volume and frequency of note-based work.
Execution: Create a structured Markdown note collection, separate sensitive material into distinct vaults, and connect only the vaults needed for AI-assisted tasks.