Find out what stops a colleague completing the same task: access, know-how or ownership. Then hand over one repeatable result and see whether it comes back.
Isn’t it simpler to let the person who’s good at it keep doing it?
Until they have a queue. The useful question is not who is best at AI. It’s what prevents someone else from doing the same task.
1. Separate three different bottlenecks.
Access. Does the colleague have an approved account and the right permissions?
Know-how. Are the inputs, steps and checks understandable?
Ownership. Who handles exceptions and changes after the handover?
Practical check: Don’t share passwords to solve an access problem.
2. Hand over one repeatable result.
Show. Use a safe example and explain why the result is acceptable.
Let them run it. A colleague completes a second case with their own authorized access.
Check the handover. Compare the result and record where help was needed.
3. Count the work that no longer queues.
Check whether the colleague can complete and verify the task without sending it back. Keep specialist support available for exceptions.
Practical check: The aim is shared capability, not leaving people alone with a new tool.
Read more: Using AI to Pass On Expert Know-How to Newer Staff · AI Skills Beyond Prompting: What a Business Team Needs
This is general guidance, not advice for your situation.