Calling all control freaks

Jonathan Nolen, Chief Product Officer, LaunchDarkly
Tim Cook, Dir. Product Management, LaunchDarkly

Everyone's talking about AI software factories. Far fewer have actually built one. In this panel, engineering leaders, partners, and teams putting agents into production go past the hype to the real work: where agents accelerate delivery, where they break, and what it takes to keep humans in control as autonomy scales. Expect specifics, not theory—what worked, what didn't, and what they'd do differently.

Whether you're experimenting with coding agents or scaling toward autonomous delivery, you'll leave with practical patterns for building a factory that's faster and safer. Join us for an honest conversation about engineering at agent speed.

Kevin Cochran, Sr. Manager, Systems Engineering, LaunchDarkly
Abram Thau, Sr. Solutions Architect, LaunchDarkly

Welcome to the only factory builder where the thing you're automating is the software delivery pipeline itself. Just like your favorite base-building games, you'll start with nothing but a skeleton and a plan, then construct the machine piece by piece: Wire up a LaunchDarkly Factory agent graph powered by Amazon Bedrock, lay down a CI process, and watch AI-generated code flow down the line and get built automatically.

The factory must grow. Add a second Agent Graph to validate inputs and judge gameplay, bolt on new features, and optimize your throughput. And because no factory run is complete without a stress test, you'll throw garbage inputs at your creation to see whether your agents catch the chaos or the whole line jams.

No conveyor belts required—just AgentControl, Agent Graphs, and the factory doing the heavy lifting.