Managed AI in practice / FIELD NOTES
Application Maintenance Handoff: Make Everyday Ownership Practical
Prepare an asset register, access handover, operational checks and recovery instructions, with a form-notification incident rehearsal.
What does a maintenance team need beyond source code and an administrator account?
A maintenance handoff explains dependencies, fault detection, recovery and decision ownership. Microsoft's operational guidance emphasizes observability, defined responsibilities and exercised incident procedures. The handoff should make those responsibilities executable.
Inventory assets and owners
Record the production URL, code and deployment locations, environments, domain, database, storage, external services, renewal items and owners. Distinguish production from testing and identify business contacts.
Transfer access through appropriate accounts. Document where credentials are managed rather than copying secrets into the handbook. Verify that authorized maintainers can access essential resources without relying on a departing person's account.
Write operational checks as steps
For each critical business journey, specify the entry point, expected result and exception procedure. Identify relevant logs, failure records or business counts, and explain how to trace one request across steps.
Record backup scope, location and recovery procedure, then test restoration. For AI features, include configuration, prompt versions, evaluation examples and post-change checks.
Worked example: submission succeeds, alert fails
Suppose an enquiry form reports success but no internal notification arrives. The runbook first checks whether the submission exists, then inspects notification processing to isolate the failed stage.
If storage succeeded, recover the missing notification rather than asking for another submission. Record impact, temporary handling, owner and recovery evidence. Add the cause and follow-up improvement after the incident.
Rehearse the handoff
Have the receiving maintainer perform a check and controlled recovery using their own authorized access. Document any step they cannot complete independently.
- Assets, dependencies and owners.
- Working maintainer access.
- Business checks and diagnostic locations.
- Recovery, rollback and pause procedures tested.
- Change history, open issues and review date.
Before you hand over the work
- Transfer executable procedures and responsibility.
- An independent rehearsal exposes documentation gaps.
Demonstration projects and examples illustrate working methods. Follow the receiving party’s specifications and the scope agreed for your project.