Routing index and targeted retrieval
Agents can inspect a compact index describing the articles in a brain, then pull the specific knowledge relevant to a task instead of relying only on broad search or chat history.
OzBrain
OzBrain is a hosted shared knowledge base that lets connected AI agents and teammates read and write structured, versioned knowledge. It helps individuals, developers, and teams keep decisions, research, plans, and operating context consistent across tools such as Claude, Claude Code, ChatGPT, and Cursor.
Shared knowledge infrastructure for AI agents
OzBrain is a hosted knowledge base designed for AI agents and the teammates who use them. It provides a shared layer of structured articles, links, provenance, and freshness information so connected agents can work from the same source of truth instead of isolated chat memories, project contexts, or copied documents.
Agents use a routing index to identify relevant articles, read the knowledge needed for a task, and stage new information back into the brain. Proposed changes are checked against existing knowledge before they become canonical, while article history records the agent and reasoning behind each accepted change. OzBrain can be connected to agents including Claude, Claude Code, ChatGPT, Cursor, and Gemini, with access granted to collaborators when a team needs a shared knowledge base.
Core features
Staged agent writes
Agents stage new knowledge in the article it belongs to rather than writing directly into the canonical version.
Conflict checking
When a proposed write contradicts existing knowledge, the change pauses and the conflict is surfaced. A user or another agent must approve a supersede or record an exception.
Provenance and change history
Article history records which agent made a change, what changed, and the stated reason, allowing users to review how the brain evolved.
Freshness and replacement tracking
Newer knowledge can mark older notes as replaced and point to the current version, reducing the need to find and update scattered copies manually.
Private storage and markdown export
Brains are encrypted at rest and sealed to the account. The site states that it does not train on brain data or sell it, and users can export their knowledge as markdown.
Practical use cases
Shared team context
Keep company vision, rules, skills, engineering plans, customer knowledge, research, and roadmap information in one brain that multiple team members' agents can consult.
Cross-repository development
Give a coding agent access to stack conventions, client details, and decisions when starting a fresh repository, while leaving repository-specific build commands and quirks in CLAUDE.md.
Retaining development knowledge
Let an agent stage decisions and implementation gotchas discovered during a build so they remain available after the session and can be read by other connected agents.
Maintaining current operational information
Use article freshness and replacement tracking for evolving plans, product details, or decisions so agents are directed toward newer knowledge rather than scattered older copies.
Controlled agent collaboration
Review proposed changes and their reasons before they enter shared knowledge, which supports multi-agent work without silently overwriting existing facts.
Pros and Cons
Pros
- Provides one shared knowledge layer for multiple AI agents and collaborators.
- Stages writes and checks contradictions before new information becomes canonical.
- Records provenance, reasoning, and article history for changes.
- Supports article-level retrieval through a routing index rather than depending solely on search or chat memory.
- Offers markdown export and states that brain data is encrypted at rest and not used for training.
Cons
- The documented connector setup and detailed workflows are most explicit for Claude and Claude Code; behavior for other listed agents is less fully described in the available source.
- Custom connector setup cannot be initiated from the Claude mobile app, although an existing connection can be used there.
- The product is centered on agent-readable structured knowledge, so it does not replace human-oriented workspaces such as Notion or repository-specific files such as CLAUDE.md.
FAQ
What is OzBrain?
OzBrain is a hosted shared knowledge base that AI agents can read and write. It stores structured articles with links, provenance, and freshness information, providing a common knowledge layer behind connected agents rather than separate memory in each AI product.
How do I connect OzBrain to Claude or Claude Code?
Add OzBrain through Claude's connector menu using the MCP URL https://ozbrain.com/api/mcp, then sign in and approve the connection. Claude Code uses the command `claude mcp add --transport http ozbrain https://ozbrain.com/api/mcp`, followed by approval through `/mcp`.
What can an agent do after connecting to OzBrain?
Connected agents can read a routing index, retrieve the articles relevant to a task, and stage new knowledge. Writes are checked for conflicts before they become canonical, and each change is attributed in article history.
Can teammates and multiple AI tools use the same brain?
Yes. Grant a collaborator access to a brain and their agents can read it too. Knowledge staged by one connected agent can then be available to other connected agents, including Claude, ChatGPT, Cursor, and Claude Code, subject to the shared brain's access.
Does OzBrain replace AI projects or repository instruction files?
No. OzBrain is described as a layer underneath tools such as Claude, ChatGPT, Cursor, Gemini, and Claude Code. It is intended to hold shared structured knowledge, while projects and repository-specific files continue to serve their existing roles.
Quick Facts
- Category
- AI productivity and knowledge management
- Primary users
- Individuals, teams, and developers using multiple AI agents
- Agent connections
- Claude, Claude Code, ChatGPT, Cursor, and Gemini are named on the product site
- Connection method
- MCP connector; Claude and Claude Code setup instructions are documented
- Knowledge format
- Structured articles with links, provenance, freshness, and article history
- Export
- Markdown export is available, including after cancellation according to the site
Alternativas a OzBrain
Falconer
Falconer is an AI-powered company brain that keeps internal documentation accurate, searchable, and up to date by syncing with tools like GitHub, Slack, and Linear. It helps engineering and cross-functional teams centralize knowledge and answer questions from a shared source of truth.
Goldfish
Goldfish is an AI memory app for macOS that helps you reply, write, summarize, and continue using the context already on your Mac. It stores memory locally and writes in your tone.
Lasso
Lasso is an ecommerce product data platform for enriching catalog records, processing supplier files, generating product content, and monitoring competitors. It combines a web app with a REST API, SDK, and MCP server for teams and developers.
ArtDeck
ArtDeck is a visual reference board app for iPhone, iPad, and Mac that lets artists collect mixed reference material on one canvas and study it with ToolBox lenses. It works offline, supports optional iCloud sync, and is sold as a one-time purchase with no account required.
Struere
Struere is an AI-native platform for turning spreadsheet data into structured operational software with dashboards, alerts, and automations. It is aimed at teams that want to replace manual spreadsheet workflows without building custom tools from scratch.
garden-md
garden-md is an open-source Node.js CLI that turns meeting transcripts into a local company wiki. It links entities across transcripts, keeps the original text intact, and renders the result as a browsable wiki.