The problem
Once you have multiple AI projects running at once, inconsistency is the default. Docs drift, release steps get skipped, research quality varies.
What I built
A coordination layer. Four parallel research subagents vet product ideas against cost, effort, and monetization filters. Fleet management audits all repos for stale docs, hardcoded dates, and missing error handling. Release tooling enforces a consistent ship sequence.
Why it exists
It is less a product and more an operating system for the rest of the products. Without it, each project would drift independently.