Stop reading every line of AI-written code. You can't keep up, and reading isn't understanding anyway. The Meta engineer who said it all feels like a black box skipped the step that matters, the design. I decide the interfaces and the plan, and the model fills in the bodies. Then I know what each part is for without reading it twice. Then I make it write the tests I'd have written, and nothing merges without a check from something that didn't write the code. In my own Claude Code setup that's a review gate: a second pass reads the diff cold and has to say yes, and a commit hook blocks secrets and client names before they leave my machine. The black box isn't the tool. It's what you get when the model writes the design as well as the code. If you let it do both, how did the code look three months in?

Messy. Now I write the interfaces myself and let it fill in the bodies. Way easier to come back to later.

I had the pleasure of working with 21st.dev to make small design fixes easy to see. AI can build a page, but the details don’t always match. Different button sizes. Inconsistent text. Selected states that aren’t clear. The design feels off. To explain Design Bug Bot, I built a UI component with those problems and animated the fixes one at a time. I brought each issue into focus, showed what changed, then pulled back to show the result. Why this approach? People won’t pause and replay your video to figure out what changed. That’s why I make each feature obvious, one at a time. If a feature still needs explaining after you’ve shown it, the video hasn’t done its job.