UX audit

Where people get stuck, and what to fix first

Your AI reviews an app or flow the way a usability consultant would: task walkthroughs, findings tagged with heuristics and rated 0 to 4 for severity, an accessibility check, and a prioritized fix list ready for the backlog.

By Vander Labs. Works with Claude, ChatGPT, Cursor and Claude Code.

What your AI learns

Walks each key task as a first-time user on a phone
Tags every finding with a heuristic and a 0 to 4 severity
Checks contrast, targets, labels and errors, and prioritizes fixes

How it works

Set the scope: Flows, numbered screens, user profile, device, and what is out of scope.
Walk each task: Step by step as the user, counting taps and recording hesitations, wrong turns and dead ends.
Score against heuristics: Nielsen's ten as the shared vocabulary, one tag per finding so patterns are visible.
Rate severity: 0 to 4 from frequency, impact and persistence; catastrophic items go first.
Check accessibility: Contrast, touch targets, focus states, labels, error messages, color-only information, motion.
Review forms and errors: Field count, input types, validation timing, error placement and recovery.
Review content: Button labels, consistent terms, empty and loading states, destructive confirmations and undo.
Prioritize and recommend: Severity, then reach, then effort; the smallest change that removes each problem.

Limits and safety

States that it is an expert review and recommends testing with real users for severity 3 and 4 items.
Accessibility items that cannot be verified from screenshots are marked unverified, not passing.

Topics: ux, usability, heuristics, accessibility, audit