Getting started with AI dev #3 | The engine

This is part three, the last one in a series on developing with agents. Part one gave your coding agent a real process for building a feature, and part two gave it the backlog to work from. This one ties them into a single flow you run left to right, from an idea to a shipped feature. A flow is just data, an ordered list of steps the engine walks through, so each of the last two videos becomes one step and the rest of the work sits on either side, from deciding what to build to running QA, deploying, and recording what changed. I teach it in two passes, the theory first and then a live build of a collection log feature on Balance, my walking-gated RPG, in raw Claude Code. It ends with a prompt on the site that writes you your own engine and a starter flow, then interviews you to fit it to your project. The full write-up is on merc.camp for members. ▸ Full write-up on merc.camp: https://merc.camp/experiments/the-dev-suite ▸ The whole series (playlist): https://www.youtube.com/playlist?list=PLZaVKZdeMNqo ▸ Part one (start here): https://youtu.be/gxYYrzst1so ▸ Part two: https://youtu.be/qZPFxi9fmSE ▸ Twitch (I stream the dev): https://www.twitch.tv/n3rkmind ▸ Discord: https://discord.gg/zFZE7mfGke ▸ merc.camp: https://merc.camp Chapters 0:00 Intro 0:36 Your two videos are two steps 2:09 What is a step 3:05 A prompt to build your own 5:49 Creating substeps 7:27 Building live on Balance 9:16 Kickstarting the engine 12:09 Spec and tickets 14:37 QA on the emulator 20:13 Deploy and post-deploy 21:58 What I believe If you're building something and want a team on it, that's what we do at Warden Labs: https://wardenlabs.ca/en