AI UX Audit
Drop your screens or flows in. Four rival AIs audit the experience against a 38-point pre-launch standard and hand back a prioritised fix list with a ship gate.
A prioritised list of UX issues (P0–P3) with severity, owner and concrete fixes.
Signed off by an independent Final QA audit — with a “verify before you rely on this” list, never a black box.
Why this team
A UX audit fails when everyone in the room already knows the product — so this council seats the End-User Advocate and the UX Researcher, who read your flows the way a first-time user does, cold. The Onboarding Specialist owns the minutes before the “aha” moment, where most launches quietly die, and the Accessibility Advocate catches the states and users the happy path forgot. The Devil’s Advocate exists to say the uncomfortable thing: that a flow everyone internally loves is confusing.
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 two-person team is a week from launching a habit-tracking app; onboarding has five screens and sign-up comes before the user has logged a single habit.
Five onboarding screens before the first habit is logged is four too many. The user came to track a habit — let them do it on screen one and ask for the account after the first win, not before it.
Deferring sign-up sounds clean until you look at what it costs here: streaks are the retention loop, and an anonymous streak that vanishes on reinstall is a support fire waiting to happen. If sign-up moves, streak recovery must ship with it — that is launch scope, not polish.
I walked the flow as written. Nothing tells me why I am granting notification permission on screen two — I have seen no value yet, so I tap “Don’t allow”, and the product’s main re-engagement channel is dead before day one. That prompt has to move to after the first logged habit.
Questions
What do I actually paste in — I don’t have a formal spec?
Screens and honesty. Screenshots or a screen recording of the real flows, plus a sentence on who the user is and the one job they hire the app for. The council reads what you give it — the more real the material, the more specific the P0 list. A polished spec is not required; a napkin description of the onboarding is enough to start.
How is this different from just asking one AI to review my app?
One model tends to produce a polite, generic checklist — and it agrees with however you framed the question. Here five personas on different frontier models each read the flows through a different lens, then have to answer each other’s objections by name. The disagreement is what surfaces the non-obvious problems, like a retention loop that breaks when sign-up moves.
We launch in days — is a P0–P3 list going to be actionable that fast?
Yes — every issue in the final synthesis carries a severity, the affected screen, why it hurts the user and a concrete fix — plus the three things to fix before launch and, just as important, a list of what is fine and should not be touched. It is built to be worked through in an afternoon, not admired.
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 →