Vibe coding is not the mistake. Failing to decide that you are doing it is.

Sometimes I want to explore an idea without supervising every line. Sometimes I need to work alongside the agent. Sometimes I can delegate the work and inspect the result at specific points. The difficult part is choosing deliberately, and noticing when the task calls for a different kind of attention.

Through development examples and recorded demonstrations, we explore these choices using an accessible reading of Martin Fowler’s human and agent loops. The task, its consequences, and the evidence available to check the result matter more than loyalty to a particular tool or working style.

You leave with questions for deciding how much autonomy to give an agent and when to change your involvement. Each mode has a place. The skill is knowing why you chose it.

For developers already using coding assistants and agents. A practical session on choosing how closely to supervise different kinds of work. It gives teams a vocabulary for discussing autonomy without treating one working style as universally right. Recorded examples connect the choice to planning, execution and verification. Particularly suited to AI-assisted development, software engineering and DevOps tracks.