Skip to content
Back to selected work

OTTO Sport AI · 2026

An AI-driven development lifecycle

Orchestration repositories and organization plugins that make spec-driven development the default and let AIDLC frameworks plug in the way gems port into Rails.

SDLC → AIDLC
Published org plugins
Pluggable frameworks

The problem

As OTTO leaned into AI, the interesting question was not whether a model could write code. It was how a whole team changes the way it works so that AI becomes a dependable part of delivery rather than a novelty that quietly creates rework.

The approach

I lead the shift from a traditional development lifecycle (SDLC) to an AI-driven one (AIDLC), with spec-driven development at its center: a clear, reviewable specification comes first and becomes the contract that engineers and agents both build against. To make that repeatable I built orchestration repositories that let agents pull in the right knowledge and work against the right code, and I developed and published organization plugins so the team shares one setup. The orchestration is deliberately opinionated, much like Rails and its MVC conventions about separation of concerns and where things belong, which lets other tools plug in the way gems port into Rails.

The outcome

Because the orchestration sets the conventions, established frameworks plug straight in rather than competing for the same space. BMAD-Method, GitHub Spec Kit, and AWS Labs AIDLC workflows all run on top of the same structure, so the team can reach for whichever fits a task without rebuilding how agents find knowledge or where their work lands. Spec-driven development becomes the normal way to work rather than a special case.

Built with