AI Pulse · Under the Radar

Most AI tools remember things inside their own walls. A conversation with Claude sits in Claude. Notes created in ChatGPT stay there. That sounds manageable until a team is using several tools and each one has a different version of the customer, project, policy, or decision history.

OzBrain is trying to solve that with a shared knowledge store that both Claude and ChatGPT can read from and write to through their connector menus. Rather than keeping loose chat memories, it stores structured articles with links, provenance, and freshness information. In practical terms, that means a team could keep a record of where an answer came from and whether it still needs checking, instead of relying on whichever AI tool was used last.

The project has drawn early attention on Hacker News, with 85 points and 50 comments, plus reposts on X and Instagram outside that audience. That does not prove the product will become standard infrastructure, but it points to a real problem: businesses are accumulating AI-generated knowledge faster than they can organise or trust it.

For owners, the question is not whether every employee needs another AI app. It is whether the information those apps use is controlled, current, and shared across the business. This is the kind of thing we build into an AI command centre, where the useful context belongs to the company rather than being scattered across individual chats.