UX checklist for AI-built websites

AI-generated hero sections default to confident-sounding but vague copy — check whether a first-time visitor with no context could say what the product does and who it's for. If the headline could plausibly sit on ten other landing pages, it needs a rewrite, not a restyle.

Confirm there is exactly one primary call to action per screen. AI builders often add secondary and tertiary buttons with equal visual weight, which forces a decision the visitor didn't come to make.

Check that headings actually decrease in size and weight down the page, spacing is consistent rather than left at framework defaults, and there's a clear focal point in every section instead of several competing elements. Generic gradients, stock icon sets, and centered-everything layouts are the easiest tell of unreviewed AI output — swap or restyle anything that looks default.

Look at the page on mobile specifically. Coding-agent output is usually functionally responsive, but visual hierarchy that worked in a desktop-width prompt preview often collapses on a narrow viewport.

Verify there's real social proof (not placeholder logos or leftover lorem-ipsum testimonials), that form fields are labeled and forgiving of input errors, and that the page doesn't overload the visitor with choices before they've understood the offer. These are the details an AI tool won't flag on its own because it isn't judging its own output — you have to.

Run the page through a structured check rather than relying on a read-through: it's easy to miss what you already know how to interpret. Sensei scores exactly these categories — functional UX, aesthetic quality, and design practices — free, on any public URL.