Back

How I Rebuilt a 5-Day UX Project in 2 Hours Using AI

From Pixels to Production

Introduction: The 4-Year Time Capsule

Four years ago, as a Senior UX Designer, I spent at least an entire work week (5 full days) building out a patient/clinician design system and a prototype of a homescreen. It was a standard, high-quality workflow: building a design system from scratch, wireframing, and eventually creating high-fidelity static mockups in Figma.

Recently, I decided to run an experiment. Could I replicate that same project using the 2026 AI stack? As a designer with limited coding experience, I wanted to see if Claude Design and Claude Code could actually replace my "comfort blanket" (Figma) and turn my designs into a living product.

The results were honestly frightening. I'd like to point out that I have minimal coding experience, I can understand the basics off CSS and HTML but that's it. I'd never used GitHub, and i had no idea how I would share the prototype.

The Workflow: 2022 vs. 2026

The difference isn't just in the speed; it's in the output.

The old way
Figma Figma
The new way
Claude Claude Stack
Time
5 Days
2 Hours
Starting Point
Blank canvas: everything built manually
Style injection from existing Figma file
Design System
Hand-crafted tokens, components, variants
AI-generated in minutes
Output
Static mockup
Live, deployed, interactive app
Prototyping
'Spaghetti strings' between static frames
Real browser interactions out of the box
Dev hand-off
Upload design to Zeplin
Direct export to Claude Code

The "Living App" Advantage

The biggest epiphany wasn't the 10x speed increase; it was the tangibility.

In 2022, I shared a Figma link. In 2026, I shared a Vercel URL. Seeing the app live in a mobile browser, responding to actual touch and behaving like a final product, changes the stakes. It removes the "imagination gap" for stakeholders. You aren't asking them to imagine how it works; you're showing them.

The Reality Check: Pros, Cons, and the "Prompting Tax" using Claude

It wasn't a perfect "magic wand" experience. There are significant trade-offs every designer should know before diving in.

The upside
Pros
The downside
Cons
Frightening Speed: 2 hours vs. 40 hours. The math speaks for itself.
The Prompting Tax: In Figma I just click and drag. In Claude I write a sentence and wait. It feels like painting through a letterbox.
Bridge the Gap: As a non-coder, I felt empowered to actually ship rather than just draw.
Token Hunger: I used 54% of my weekly Claude Design limit in one session. Run out mid-flow and your momentum is dead.
The Panic Factor: When the code breaks and you aren't a developer, the Figma comfort blanket feels very far away.

Conclusion: Orchestrator vs. Pixel-Pusher

This experiment proved to me that the role of the Senior UX Designer is changing. We are moving away from being pixel-pushers and toward being Product Orchestrators.

I'll always love Figma for the "thinking" phase; it's still the best place for total creative control. But for moving from a validated idea to a sharable, living prototype? Claude has officially taken the crown.

Is the static mockup dead? Maybe not yet. But it's definitely on life support.

View Claude Prototype