Mike Thomson shares a practical method that consistently produces better AI-built projects: write a project brief first, then use “planning mode” before you touch a single tool.
The idea is simple but most people skip it. Instead of dropping a task into an AI and letting it run, you chat with the AI first. You get it to ask you questions, surface assumptions, and write its own plan before any work starts. Then you review that plan, correct it, and only then do you execute.
The result is an AI that stays on track instead of self-directing into something you didn’t ask for. Mike explains why this matters, how to do it in practice, and what it looks like when you don’t - including what happens when you leave an AI agent running without a clear definition of done.
This is one of the most transferable habits in the video. It works regardless of which AI tool you’re using.