“Stop writing prompts. Start building the loop that writes them for you.” You’ve probably heard some version of that line by now. In this workshop we’ll actually explain what it means, and build one with you for your React Native app and your use case. This hands-on workshop teaches you to build a harness of focused agents that plan, implement, and verify each other’s work, backed by: Reusable tools, Project memory, Checkpoints. You’ll build it one piece at a time, starting from a single agent and ending with a working pipeline you run against your own app. We use TV as the live target because it’s a genuinely non-trivial adaptation, focus navigation, 10-foot layouts, remote input, many different operating systems, but the target is just the example. As a final layer, we’ll add a context-aware agent powered by Bee. It records the conversations you have about the app you want to build, saved whenever you want and available anytime, then feeds them back into the harness when it runs. You’ll leave with a forkable repo and a clear mental model for building development systems instead of babysitting prompts.