Visible & editable
Every memory
shows its work.
Inspect who wrote a memory, when it was observed, and what it connects to. Nothing in your organization's brain arrives anonymous, and nothing is stuck wrong.
A source chain, not a guess: the agent that wrote the memory, when it was observed, and the entities it links to.
A source chain, not a guess
Written by. Observed. Linked entities.
Every memory carries the agent that produced it, the moment it was observed, and the entities it connects to. When an answer surfaces, you can follow it back to the thing that caused it instead of taking the model's word for it.
- Written by — the agent that produced the memory
- Observed — when it entered the shared memory
- Linked entities — what it connects to in the graph
- Full history — every edit, kept
One correction, everywhere
Correct the memory. Keep the context.
Open what an agent got wrong and fix it in place. One correction. Every connected agent reads the same sourced memory on its next call — no re-prompting, no stale copy sitting in someone's context file.
Keep exploring