AURA · SAMPLE DELIVERABLE
Before the pilot: make the decision explicit.
This worksheet shows how AURA structures an AI question around one workflow. It illustrates a possible excerpt from SHIFT, or an entry in a Mission 2032 portfolio.
Fictional example · no client or measured results
AGARTHA · AURA
Decision note · internal request triage
01
The proposed workflow
- Receive an internal request
- Read it and suggest a category
- Obtain the responsible person’s approval
- Assign the request
- The question
- Could a category suggestion help with triage while keeping the decision with the responsible person?
- The hypothesis to test
- Reduce qualification time without increasing routing errors. No benefit has been established yet.
- The proposed scope
- Synthetic requests containing no personal data. No connection to the production system.
- Measurements to establish
- Measure qualification time and errors on the same set of requests, with and without assistance. The baseline has yet to be established.
- Human control
- Every suggestion requires approval before assignment. The owner and acceptance criteria still need confirmation.
- The risk to monitor
- Plausible suggestions that require correction or route a request to the wrong team.
- The stop rule
- Suspend the trial if an assignment occurs without approval or data leaves the agreed scope.
At D+30: record actions taken, available evidence and unresolved questions. Decide on the next step.
This worksheet illustrates the reasoning structure. It does not describe a pilot that has been built or deployed.