UX audit of the app
Every inconsistency and UX leak, ranked by impact.
AI-built apps
From AI-built app
From AI-built app
to real product.
Lovable, Bolt, v0 and Cursor get you to a working app fast. Then every prompt adds its own buttons, spacing and flows, and users stop trusting it. I clean up the UX and UI and put a design system under it, so you and your AI tools keep shipping in one style.
Sound familiar?
Fast to build, hard to trust
Every screen looks different
Each prompt brought its own buttons, colors and spacing. The same action looks five different ways across the app.
Users get lost after sign up
Features were generated one at a time, so nobody designed the path from sign up to the first moment of value.
Every new prompt makes it worse
With no design system and no rules for your AI tools, each new screen invents new UI and the drift keeps growing.
Deliverables
What you get
A cleanup that fixes what users see today and keeps the next hundred prompts on track.
Core flows redesigned
Sign up, onboarding and the main workflow, done properly.
A design system
Tokens and core components in Figma and in code.
Rules for your AI tools
A rules file Cursor, Lovable and Claude follow for every new screen.
Cleaned-up UI
Fixes shipped in your code or handed over as dev-ready Figma.
Process
The 4-step cleanup
Audit what the AI built
I go through every screen and flow, list the inconsistencies and rank the UX leaks by their impact on activation and trust.
Set the system
Colors, type, spacing and the core components become tokens and a small component set, in Figma and in your code.
Redesign the core flows
Sign up, onboarding and the main workflow get designed properly and rebuilt on the system, screen by screen.
Hand it back AI-ready
You get a rules file and component docs, so Cursor, Lovable or Claude keep new screens in the same style after I leave.
Proof
Design systems built for AI tools
GliaKit is my design system for AI-built products: one token set across Figma and a Next.js boilerplate, with rules your AI editor follows.
Investment
One fixed-scope cleanup
AI App Cleanup
On request
fixed fee
Scoped after a free review
From a drifting AI-built app to a consistent product your team and your AI tools can extend.
[ INCLUDES ]
- UX audit of the current app
- Design tokens and core components
- Sign up, onboarding and core flow redesign
- Rules file for your AI tools
- Fixes in code or dev-ready Figma
FAQ
Common Questions
What founders ask before handing over an app built with AI tools.
Or email me dejan@veljanoski.com
Lovable, Bolt, v0, and apps built in Cursor or Claude Code. If the app is React or Next.js with Tailwind, I work directly in the code. Otherwise I deliver dev-ready Figma and the rules your tools need.
Usually not. Most AI-built apps have working logic under a messy interface. I keep what works and replace the UI layer screen by screen on top of a design system.
Both. I design in Figma and build in Next.js and Tailwind, so for React apps I can ship the fixes myself or hand them to your developer.
That is what the rules file is for. It tells Cursor, Lovable or Claude which tokens and components to use, so new screens follow the system instead of inventing new styles.
It depends on the number of screens and flows. You get a fixed scope and a fixed quote after a free review call, before any work starts.