Where AI helps, and where the team must stay

AI can take a large share of delivery busywork off your plate, but it cannot own the parts of agile that only work when humans do them. Getting that line right is the whole game.

Where AI helps. Drafting first-pass user stories from rough notes, summarizing long threads into decisions and action items, reformatting a messy backlog item into a clear one, preparing agendas, and turning a meeting transcript into follow-ups. These are the low-judgment, high-friction chores that slow a team down.

Where the team must stay. Estimates, commitments, and the actual conversations. When AI suggests a story point range, that is an input, not a verdict — the people who will do the work still decide together. A sprint commitment is a promise the team makes, not a number a model produces. And the refinement conversation itself, where a group builds shared understanding, is the point of the ceremony; automating it away removes the value.

A simple test: if a task is about producing an artifact, AI can draft it. If a task is about building agreement or accountability, keep it human and use AI only to prepare.

Integrity angle. AI drafts; the team owns. A story a model wrote is not refined until the people who will build it have read it, questioned it, and agreed on it.

Example prompt: "Turn these rough notes into three draft user stories in the format As a [role], I want [goal], so that [benefit]. Mark anything you had to assume so we can check it in refinement."