Idle tenancies are a silent infrastructure drain — but defining and detecting them at scale is harder than it sounds, especially when cost isn’t the primary signal. In this session, we’ll share how we approached this problem on OCI using native APIs and AI. We’ll cover how we defined “idle,” how AI helped evaluate ambiguous signals, and what a reusable detection framework looks like in practice. You’ll leave with an approach you can adapt — and an honest view of where AI helped and where it didn’t.