AX Field Notes #3
From one requirement to release: what the AI agent does
Plenty of teams write code with AI now. But if the documents, tests and reporting around it stay the same, overall speed barely changes. So we gave an AI agent a whole requirement, from start to finish.
The problem: when only the code gets faster
Writing code takes less time, but requirement analysis, document updates, testing and progress reports are all still there. The slow part just moves to before and after the code.
The approach: five steps, with roles split at each
On the global automotive parts project, a single requirement flowed like this.
- Requirement — a person registers it, and AI analyses it and drafts a PRD. A person approves the scope.
- Design — AI writes the development spec and test scenarios and breaks them into work items. A person reviews the spec.
- Build — AI writes code and tests, verifies locally and opens a merge request. A person reviews and approves the merge.
- Release — merged changes ship at fixed times and AI writes release notes for each release. Up to the reference date (October 8, 2026), releases ran three times a day on a schedule.
- Record — AI logs the start and end of every step in the requirements management system. Progress is calculated from those records; a person checks them on screen and approves completion.
The result: people moved to a different seat
What changed most wasn't speed but where people sit in the process. They moved from writing things themselves to requesting, approving and checking.
Because a step can't start until the previous one is done, and scope grows only through approval, decisions stay with people even as the pace picks up.
Related case: Global automotive parts case — Vibe-Ops