Why single-assistant memory is not enough
ChatGPT memory lives inside ChatGPT. Claude Projects live inside Claude. Each is useful, but neither can read the other, so the moment you switch tools you are re-explaining yourself again. If you use both, what you actually want is one knowledge base they both read from and write to.
How one shared knowledge base works
Hjarni stores your notes as plain Markdown and exposes them through a built-in MCP server. You connect ChatGPT once and Claude once. From then on both read the same notes, and either can write back. Update a note in Claude and it is what ChatGPT reads next. There is one source of truth, not two memories drifting apart.