OzBrain

OzBrain is a hosted knowledge base that acts as a shared brain every AI agent you use can read and write. Instead of each assistant keeping its own private memory, OzBrain provides one structured source of truth that sits underneath Claude, ChatGPT, Cursor, Claude Code and other connected agents. It is built for people and teams who already use several AI tools every day and want the work they have already done to be available the next time any agent starts a task. The company describes it as your Dropbox for agent knowledge: a place where structured articles with links, provenance and freshness live behind the connector menu that Claude and ChatGPT already show you. Today, context lives in your chats and your teammate's context lives in theirs, and the two never meet. People share a doc, drop a message and paste the same things over and over again. Meanwhile copies of the same plan sit in Drive, on laptops, in Downloads, in email, and no one is sure which version is current. Platform memory does not solve this either: as the site puts it, that memory is a few preferences and a thin summary of past chats. OzBrain's answer is to hold the work itself — your projects, decisions, research and the thinking you have already done — so an agent starts with what the task needs instead of whatever fits in a profile. The centrepiece is a single shared brain. Point everyone at one OzBrain and your agents and their agents read and write to the same place, so what one person works out, everyone's agents already have. The site illustrates this with a team view: a company brain holding company vision, rules and skills, engineering plans, customers, research and the roadmap, with individual people connected through Claude, Cursor, Codex and Claude Code. Because the knowledge sits outside any single chat product, it is not owned by whichever assistant happened to be used first. Sharing on brains you own is included in every plan, including the free one. OzBrain breaks your knowledge into nested pieces so that an agent pulls the exact slice it needs — the email body, not the whole launch plan. The site shows a launch plan at 11,842 tokens and 47.3 KB nesting into launch campaigns at 1,486 tokens and 5.9 KB, which in turn nests into launch email at 218 tokens and 0.87 KB. The argument is straightforward: the less an agent has to load, the faster and cheaper it answers, and the less it invents from context it never needed. Alongside this, OzBrain keeps the latest version out front. Instead of copies of the same plan scattered across Drive, laptops, Downloads and email, every agent decides from the current version rather than an old copy. When newer thinking lands, OzBrain goes back through your knowledge on its own, marks the old notes as replaced and points to the latest. You never have to hunt down every place an old decision lived, and no agent answers from a version you have moved past. The site gives the example of a website launch plan being updated when the company acquired a new domain, with related company details marked as updated to ozbrain.com. Every change is also on the record: which agent made it, what changed and the reasoning. Changes are proposed before they land, so several agents can work at once without writing over each other, and a recent-changes table shows the time, agent, article and reason for each edit. Privacy and ownership are handled explicitly. Your brain is encrypted at rest and sealed to your account, so nothing leaks between tenants. The site states that it never trains on your data and never sells it, describing OzBrain as a sovereign place for your data. Export is available at any time — everything as markdown, including after you cancel — on the principle that your knowledge should not be locked in anywhere, including OzBrain. Deleting your account removes your content: delete means deleted. The company also notes that it runs OzBrain on OzBrain, with its own data sitting next to yours, because it wanted the safest and easiest tool for itself as well as for users. Getting started does not require any coding. You add OzBrain from the connector menu in Claude or ChatGPT, sign in and approve it; there is nothing to install. Connect guides exist for Claude and ChatGPT, and the same brain can be reached from Claude Code, Cursor, OpenClaw, Hermes Agent, Gemini Spark where Google makes it available (US, Spark eligibility) and any client that supports connectors. Because it is one URL, anything that speaks the protocol can hold the same brain. Rather than being a memory API that developers code against, OzBrain is described as a brain you connect — structured articles with links, provenance and freshness, and one source of truth rather than a separate memory in each product. The stated outcomes for users are practical. Agents make fewer mistakes because they pull only the slice of knowledge a task needs. Answers are faster and cheaper because less context has to be loaded. Nobody has to maintain or "work" the brain, since replaced notes are marked automatically and the current version stays out front. Teams stop pasting the same context into several separate chats, and a brand-new chat can know what everyone has already worked on. Because everything is exportable as markdown, users keep an exit route and can move their knowledge to whichever tools serve them best. Concrete scenarios appear throughout the site. In one, a team connects individual agents through Claude, Cursor, Codex and Claude Code so the whole team's agents work together on the same knowledge — company vision, rules and skills, engineering plans, customers, research and roadmap — instead of each person's context living in a separate chat. In another, a website launch plan is updated when the company acquires a new domain, and OzBrain marks related articles, such as company details, as updated so no agent answers from the old domain. The changelog example shows several agents, including Claude, making coordinated changes to articles such as a website launch plan, an MCP setup guide and a surfaces page with a stated reason for each edit. A further scenario is a new chat starting with what a team has already worked on rather than from a blank slate. OzBrain is aimed at individuals and teams whose work already runs through multiple AI agents, from a single person's venture and personal knowledge to an organisation where agents run the operation from one brain. Supported integrations include Claude and ChatGPT through their native connector flows, plus Claude Code, Cursor, OpenClaw, Hermes Agent, Gemini Spark where Google makes it available (US, Spark eligibility) and any client that supports connectors. Pricing starts with Free at $0 forever, with up to 50 articles, sharing on brains you own, unlimited brains, reads, writes and connections, write-time size discipline and markdown export anytime. Pro is $20 per month with up to 500 articles, described as room for one venture plus personal knowledge. Max is $99 per month with up to 5,000 articles, for when agents run the operation from one brain. Enterprise sits above the Max ceiling with org-owned shared brains and per-seat pricing designed with you. Every plan includes unlimited reads and writes, and free plans mean you can start without a credit card. The takeaway is that OzBrain turns scattered, agent-specific context into one shared, encrypted and exportable brain that every agent and teammate can read and write. It keeps the current version out front, prunes replaced notes automatically, records who changed what and why, and lets you walk away with everything in markdown whenever you choose.