For when
“We deployed something; is it still worth running?”
Usually commissioned by: Service owner, operations leader.
What you get
- Monthly or quarterly review of usage, quality, cost and incidents
- Evaluation re-run against the agreed test set after model or vendor changes
- Improvement backlog and agreed maintenance
- Retire / keep / extend recommendation at each review
Acceptance condition
The named service owner receives an actionable review and the agreed maintenance is done.
How it runs
- 1
Agree the metrics, the review cadence and the service owner
- 2
Collect usage, quality and cost data from the running workflow
- 3
Review session and written findings
- 4
Carry out the agreed maintenance and update the evaluation set
It may conclude…
Retire the workflow. When it no longer earns its keep, saying so is part of the service.
Not included
- New workflow builds (scoped separately)
- 24/7 incident response
Read before you decide
Free guides that cover the same ground. Plenty of teams get what they need from these alone.
Why does our internal knowledge assistant give bad answers?
Diagnose a poor internal knowledge assistant: classify failures, build an evaluation set from real questions, and apply the cheapest fixes before rebuilding.
How to measure AI adoption — and when to retire an AI workflow
A workflow health review for AI in production: usage, quality, review effort, incidents, cost and maintenance, plus clear criteria to keep, extend or retire.
Discuss a Adoption & Assurance Retainer
Send a short project brief. A person reads it and replies with whether this engagement fits, a different one would, or none is needed.