// Use case / Edge AI assurance
AI decisions need receipts.
Bind AI requests, responses, workload identity, timestamp, and execution context into evidence that can be reviewed after the mission.
// Mission problem
Output without provenance is just output.
As models move closer to mission execution, operators need more than output. They need to prove which workload handled the request and what response was returned.
- Workload attributionQHx records the workload identity, pod metadata, and image information where available.
- Request linkageRequests and responses can be logged and linked by a request identifier.
- Signed verificationsignRequest mode creates signed evidence for offline review and non-repudiation.
- Edge reviewReceipts preserve evidence even when the model ran in a disconnected or constrained environment.
The model returned an answer. The receipt says which model, on which node, with which inputs, at which time.