Built this for the @Flowstep X @Contra Challenge.

Most people will probably look at the final UI.

I hope you look at the process.

Instead of spending days designing every screen manually, I wanted to see how far an AI-first workflow could go before I stepped in.

Here's how I built it:

• Researched the product, user flow, and screen prompts with ChatGPT. • Generated all illustration assets with Google Flow. • Uploaded my design system into Flowstep first. Typography, colors, spacing, border radius, component rules, everything. • Generated the onboarding flow first to test consistency. • Then generated every remaining screen using detailed prompts.

In around 10 minutes, Flowstep produced roughly 70–80% of the UI.

That saved hours of repetitive work.

Then came the designer's job.

I exported everything into Figma, cleaned the layouts, fixed inconsistencies, improved spacing, replaced placeholders with the illustration assets, and made the experience feel much more playful.

After polishing everything, I imported the designs back into Flowstep to build the prototype.

It wasn't all smooth.

I hit Figma rate limits multiple times while syncing files, and prototype generation failed more than once.

But that's part of working with new AI tools.

The biggest lesson from this project:

AI didn't replace the design process.

It removed a lot of the repetitive work so I could spend more time making the product feel thoughtful.

Good luck to everyone participating. Looking forward to seeing what everyone built. Flowstep protype link : https://app.flowstep.ai/prototype?activeFileId=71b1f607-3236-49cf-9551-3b26b5593dad

Flowstep File link : https://app.flowstep.ai/file?activeFileId=71b1f607-3236-49cf-9551-3b26b5593dad As Flowstep doesn't support imported screens prototype you can check still test it here : figma link