Is Heidi just a vector database with a chat interface?
No. Vector search is one technique inside the Cortex, which is a typed knowledge graph where every fact keeps sources, permissions, and history. And the Cortex is only half the product; the other half is the runtime that acts on it.
Embedding similarity finds text that sounds related. It cannot tell you which of three conflicting statements is current, who is allowed to see it, or where it came from. The Cortex is built to answer exactly those questions.
On top of the memory sits the part a vector database does not attempt: agents with earned trust, durable scheduled jobs, receipts, and an API and MCP surface.