Module 1 recap

You started this module looking at AI pages the way everyone does: "fine, I guess." You are ending it with a diagnostic kit.

You can name why an undirected page looks generated: the model converging on the average of everything it has read, down to the same gradient, the same three cards, the same polite typeface — and the newer, tastier defaults that replaced them. You can decompose any page into the four levers — typography, color, motion, background — and pull them one at a time. You can replace adjectives with named directions that carry a whole world of decisions in a few words. And you can write a critique that doubles as a prompt: element, measurable observation, instruction.

One artifact came with you through all of it — undirected slop in lesson 1, one lever pulled in lesson 2, a full direction in lesson 3. That page is proof of the module's real claim: the model was never the constraint.

Where this goes next

Seeing is the free half of the skill. The rest of the course is producing on demand:

• Module 2 — Direct the aesthetic. The prompt patterns that get a point of view on the first try: the four-part design prompt, banning defaults explicitly, tokens before pixels, and why three variations beat ten iterations.
• Module 3 turns your taste into a file — a design system Claude follows every time, so you stop re-explaining yourself.
• Modules 4–8 go from the Claude Design canvas through self-reviewing browser loops to a client-ready build you can charge for.