AI UI Design Review
Drop your screens in. Four rival AIs hold the visual design to an Apple-grade bar — hierarchy, type, spacing, states — and hand back the exact fixes.
A prioritised list of UI/visual issues (P0–P3) with the exact fix for each.
Signed off by an independent Final QA audit — with a “verify before you rely on this” list, never a black box.
Why this team
Visual quality dies by a thousand small allowances, so this council pairs the Visual Designer and the Creative Director — one auditing the craft detail by detail, the other judging whether the whole reads as one intentional product. Steve Jobs sits on the panel for a single reason: the refusal to accept “good enough”. The QA Auditor sweeps the states polish reviews skip — dark mode, error, loading — and the Accessibility Advocate stops beautiful-but-unreadable from shipping.
Each seat runs on a different frontier model — GPT, Claude, Gemini, Grok — so it’s a genuine cross-model review, not one AI in costumes. How a run works →
What the debate sounds like
An illustrative excerpt — A solo founder is about to launch a finance dashboard with a custom dark theme, three accent colours and typography assembled from two different component libraries.
Three accents is two too many for a data product. The green means “profit” on the chart but “success” on the toasts and “primary action” on the buttons — the same colour is doing three jobs, so it does none. Pick one semantic per colour and rebuild the palette from that rule.
Nobody will list the fonts, but everyone will feel them. The numbers table is set in a different face to the cards beside it, and the whole screen whispers “assembled, not designed”. Choose one type family, one scale, and delete the second library this week — that is the launch decision.
I checked the states, not the screenshots: the loading skeleton flashes white in dark mode, the error toast is unthemed system default, and the empty portfolio state still has placeholder copy. None of this appears in the happy path the founder has been staring at for months.
Questions
Can the council really judge visuals — do I paste screenshots?
Yes — attach screenshots or exports of the actual screens, ideally including dark mode and one error or empty state. The critique cites specific elements (“the toast”, “the numbers table”), not abstractions. Text descriptions work too, but the review is only as concrete as what you show it.
What if my design is fine and I just need confidence?
Then the verdict says so — the brief explicitly asks for a “what is already excellent” list and a yes/no on launch-quality craft, and the house rules forbid inventing issues to pad a list. A clean pass with two P3 nits is a legitimate, and common, outcome.
UI or UX audit — which one do I run first?
UX first if users get lost, UI first if the product works but feels cheap. The UX audit interrogates flows, states and friction; this one interrogates hierarchy, type, colour, spacing and polish. Teams close to launch often run both and merge the two P0 lists into one punch list.
Your material is used only to run your review — never to train public models. Encrypted in transit and at rest. Security & privacy →
Want full control — pick your own minds, set the depth? Open the full council →