Cross-meeting memory layer via MCP integration

[pintomatic]

Hey everyone,

Love what is happening with Meetily — the privacy-first, Rust-based approach to capture is exactly where the stack should start.

I have been thinking about what happens after the meeting. Right now, Meetily excels at Layer 1 (Capture). But to get true utility from meeting data, we need a Layer 3 (Memory) that lives across sessions.

I would like to propose an integration with Kernal via MCP (Model Context Protocol). Kernal builds a persistent graph of your professional relationships and history.

The Workflow: Meetily captures the transcript → Kernal extracts entities/intent → Data persists in a local-first graph.

Concrete Example: After 10 meetings with a specific client, Kernal has mapped out that the Lead Engineer cares about latency, the PM is worried about the Q3 deadline, and the last call mentioned a blocker on the API side. Instead of searching through 10 files, your context is pre-loaded with the actual state of the relationship.

Since Kernal is MCP-compatible, it plugs right into the modern AI stack — local-first, privacy-respecting, same philosophy as Meetily.

How do people here feel about adding a persistent memory layer to the Meetily ecosystem?