Skip to content

A design critic that does not need a meeting

Client
UX Buddy
Sector
Developer tools
Services
Product Design, Plugin Development, UX Research
Duration
9 weekends
Year
2025
UX Buddy · UX Buddy

Results

122

designers using it, live in the Figma Community

6

critic personas, each with a distinct domain and voice

9

weekends from idea to published plugin

The brief

Finishing a screen means waiting for a lead, a stakeholder, an accessibility review. Most of that feedback is mechanical: spacing is off, contrast fails AA, the loading state is missing. It does not need a human, it needs a checklist with eyes.

Design

The problem with design feedback

Every designer knows the cycle: finish a screen, wait for someone to look at it, receive feedback that is mostly mechanical. Spacing, wrong component, contrast failure, missing loading state. None of it required a person. It required a checklist with eyes.

The UX Buddy analysis result inside Figma, with an overall score and category ratings for accessibility, usability, visual design, information architecture and interaction design.
Six critics, six domains

Six critics, each with a point of view

The mistake most design checkers make is generic feedback. Each critic here has a distinct domain and voice: an accessibility auditor covering contrast, touch targets, focus order and screen reader labels; a visual design critic on colour harmony, type scale, the 8px grid and whitespace; a usability tester; and three more.

A before-and-after contrast comparison with a recommendation to darken text from rgba(153,153,153) to rgba(117,117,117) to meet WCAG AA.
Scan to report: a fix you can make before the next coffee

Build

What we shipped

A Figma plugin: select a frame, scan, receive a structured report. Built and published over nine weekends in 2025.

The UX Buddy Figma Community listing showing 122 users.
The plugin, live in the Figma Community

Outcome

122 designers use it, live in the Figma Community.

Front end

  • Figma Plugin API
  • TypeScript
  • React
  • Figma UI kit

AI

  • Multi-persona prompting
  • Structured critique output
  • Scan-to-report

Ship

  • Figma Community
  • Versioned releases
  • Usage analytics

Decisions that mattered

  1. 01

    Gave each critic a domain and a voice instead of one generic reviewer. 'Improve visual hierarchy' is unusable; 'your CTA at 44px should be 48px to meet WCAG touch targets' is fixable before the next coffee.

  2. 02

    Made the output a report rather than a chat. Designers want a list they can work through, not a conversation to maintain.

  3. 03

    Scoped it to what a checklist genuinely does better than a person (contrast, targets, focus order, grid, missing states) and left judgement to the humans.

Next project Sabi Trace Product Design, Interaction Design, UX Research, Design System · Commodity trade · 2024