AI-Augmented Delivery Sprint

AI coding assistants change one variable dramatically: volume. A Developer who can generate three times as much code in the same Sprint has not necessarily gained three times the judgement needed to review it, and this module starts from that gap. The core argument is that the Definition of Done, not the backlog, is where AI-augmented delivery either holds together or quietly collapses, because a green pipeline and a passing test suite say nothing about whether a human actually understood and stands behind the code that shipped.

The module works through concrete failure patterns: a pull request approved in under a minute because the volume of AI-generated code makes genuine review impractical at the team's current pace, tests that were also written by the same assistant that wrote the implementation, which means a passing suite proves internal consistency rather than correctness, and a sudden throughput spike that looks like a win until defect rates a few Sprints later reveal it was borrowed velocity. Scenarios put you inside a Sprint Review where leadership wants to know why the team is not shipping even faster, and a retrospective where the team has to name, out loud, that the review process has not kept pace with the code-generation process.

The back half of the module covers what a resilient Definition of Done actually requires under AI assistance: a human who can explain the change in their own words, tests that were reviewed independently of their generation, and explicit ownership of every merged change, because accountability does not transfer to a tool any more than it transfers to a contractor. The module also addresses capacity conversations directly, since leadership frequently mistakes AI-assisted throughput for headcount that can be cut, and gives Scrum Masters language for correcting that assumption before it becomes a staffing decision.