mechanism both

The model-swap mechanism

The brain is the asset. The model behind it is a slot, not a bet. Year one, you can swap the model that reads and writes the graph without re-ingesting anything — the same brain, a different engine.

This node The model-swap mechanism depends-on Standard deployment — your environment

Ownership is the point of the whole system, and ownership means the model can never be the lock. The graph, its receipts, and its classification are yours and stay put. The model is the layer that reads and writes them — and that layer is replaceable.

Year one, you can swap the model behind the same brain with no re-ingest. Nothing about the asset moves; only the inference engine changes.

One honest boundary: swapping the chat client you talk to, or a copilot embedded inside a workflow, is a design cherry we may build — not a year-one promise this page makes. This page claims the model swap and nothing more.

Claims

What this page stands behind.

#claim-model-swap

Year one, you can swap the model behind the same brain — no re-ingest.

Applies to
Standard and Greenfield
Evidence
This is the whole claim. Nothing about timing, downtime, or a drill cadence — those numbers are not measured, so they are not stated.
Since
2026-08-27
#claim-same-graph

The graph, receipts, and classification do not move when the model does. Only the inference layer changes.

Applies to
Standard and Greenfield
Evidence
The asset was never the model — it lives in your environment. See deployment-standard #claim-your-env.
Since
2026-08-27
#claim-not-chat-ui

Swapping the chat client or a copilot inside a workflow is a design cherry, not a year-one promise.

Applies to
Standard and Greenfield
Evidence
States what this page does not claim — the honest-boundary pattern, same as what Greenfield does not include.
Since
2026-08-27

Machine-readable → index.json · /ai-info · llms.txt

browse The rest of the brain