You knew how to do the work. Then you started giving it to an AI agent, and a different set of problems appeared. Instructions that seemed clear produced the wrong result. A lesson had to be taught again. Getting impatient did little to improve the next attempt.

These are familiar problems to anyone who has managed people or run a business. This talk explores the management skills that become part of everyday work with agents: delegating an outcome, explaining the local way of working, making lessons persist and responding usefully when mistakes happen.

We also examine the difficult part of supervision: deciding whether a result is good when you have little detail about how it was produced. That calls for clear expectations and useful evidence.

The title is the invitation. The session is about learning the job that came with it, so your agents get a manager who helps them do better work.

For developers, technical leads and professionals starting to delegate work to AI agents. The management perspective connects clear instructions, patient feedback, retained learning and validation with limited implementation detail. Attendees can recognise familiar frustrations and turn them into skills to practise. A fit for engineering leadership and AI adoption programmes, including audiences without previous management experience.