There’s a certain kind of confidence that comes from eating your own cooking. When we decided to rebuild a major subsystem of Maestro using Maestro itself, we weren’t trying to make a marketing point. We had a real deadline, a real feature spec, and a real question: could our own tool handle a non-trivial project from start to finish?
The Setup
We wrote a specification for a new story management system — the part of Maestro that tracks which Coder is working on what, which stories are blocked, and how the Architect dispatches work. It was roughly 2,500 words of requirements written in plain English.
We fed it to the PM agent, which asked three clarifying questions. After we answered, it produced a finalized spec. The Architect broke it into eleven stories. Three Coders started pulling from the queue.
What Worked
The separation of concerns held up. The Architect caught a schema design issue in PR review that none of the Coders would have caught individually — because they were each working on their own story. The Coders ran tests before every PR submission. The Architect rejected two PRs and sent them back with clear feedback.
What We Learned
The biggest insight wasn’t about the code quality — it was about the specification quality. Every issue we hit traced back to ambiguity in the original spec. Maestro built exactly what we asked for. The challenge was asking for the right thing.
That’s the same challenge any engineering team faces. The tool doesn’t change the fundamental discipline required. It just compresses the timeline.
The Result
Eleven stories. Twenty-three pull requests (including rejections and resubmissions). Working, tested code merged to main. Total wall-clock time: about four hours. Total cost in API tokens: $14.72.
We shipped it the next day.