AI guesses what a good page looks like. Proof learned it from 400+ real product pages. One short brief → the right sections, in the right order, in your brand. It self-checks and never invents facts.
Hey Product Hunt 👋 I'm Sen.
AI can build a landing page in seconds now, but you can spot it at a glance: the same centered hero, three identical cards, a purple gradient. The structure is fine. The problem is that the model is guessing from its own taste.
So I built Proof the other way round. I studied 400+ shipped product pages and wrote down what they do, why it works, and when not to do it. When you ask for a page, Proof starts from who's visiting and what might stop them, decides which sections go on the page and in what order, and only then picks the look. The goal is a first draft that's close to final, so your edits take minutes instead of a rebuild.
What's inside:
• 8 page types (landing, pricing, product, about, contact, blog, downloads, 404) and 10 sections you can ask for on their own
• 4 styles (Minimal, Business, Colorful, Editorial), each measured from a real product site with my other tool, Taste Lab, all in light and dark
• Your brand: add a color, a font and your own content
• A self-check before handover: layout rhythm, style fidelity, real logos only, no invented facts
Every page on proofdesign.dev was generated by the skill, so open any of them and judge for yourself. It's free and open source:
npx skills add senlindesign/proof
I'd love your honest feedback. What page would you build first, and what usually gives an AI-made page away for you?
Replies
this is pretty cool, actually.