ADLC: What It Is & What It Ain't, or How to Tame AI Agents So You Don't Get Vibe-Fired

Title: ADLC: What It Is, What It Ain't Subtitle: How to Tame AI Agents So You Don't Get Vibe-Fired Audience: Product managers and product leaders

Abstract: When AI made building cheap, AI agents became easy to ship and hard to trust. They've invented refund policies, sold trucks for a dollar, and deleted production databases. And when an agent runs over Mrs. Smith's cat, the agent doesn't get deposed. The PM does.

The "Agentic Development Lifecycle" is quickly becoming how enterprises govern AI agents, and the term means two different things. This session cuts through the buzz to the part PMs own: deciding whether an agent is worth building, what it may and may not do, which evals catch what demos hide, and who has the authority to stop it.

We'll walk through the six-step loop the practical, tactical way: how mere mortals validate intent, watch what actually gets deployed, and make pivot, punt, or pursue calls.

You’ll learn when to pull the plug on what's expensive, dangerous, or just "meh” or scale what improves productivity, saves money, and keeps you safe and compliant.

You'll leave able to tell your boss, or your next interviewer, when an agent is ready to ship and when it isn't. You'll also get a sneak peek, and access to, a set of agent skills that work with your favorite chatbot and make you look like a pro, run live on an agent idea from the room.