C R E A T I V E C H A L L E N G E
The AI Experiment
SneakerHues is a self-initiated project exploring AI as a collaborator in design and prototyping. Inspired by my past hobby of customising sneakers for TV and movies back in the day, I set out to create a simple desktop web app where users can colour sneaker layouts and download their creations.
The project began with vector sneaker designs I had made during a 'Vector a Day' Instagram challenge a few years back and how I could reimagine them with AI to unlock a new colourway app. While the design process came naturally, the real challenge and learning opportunity was making the experience interactive and functional with AI.
C R E A T I V E D E F I N E D
How Easy Can It Be?
My first step with ChatGPT was turning my sneaker layouts into interactive assets. I asked GPT to generate a script for illustrator to vectorise my designs and assign the files correctly for an interactive asset. The results initially looked promising but that was short lived.
With sections not correctly cut out something was wrong. I went back to GPT with my issue and after a series of frustrating tests, and back and forths I eventually discovered the root of the issue myself: the SVGs that worked properly had their sections grouped into their own folders, while the broken ones didn’t. By comparing the working files against the non-functioning ones, I was able to restructure the SVGs and solve the problem. One nil to me.
With the SVGs issue fixed, I moved on to asking AI for the full code that would make the colouring process live. I set up a project on GitHub to house the files, but found out quickly design and layout is not at the top of agenda for ChatGPT. Whilst some code worked partially, tweaks failed often broke the site or removed things entirely. I decided to get the code as close to a perfect interaction as possible with GPT but prompts for correcting layout were not getting me anywhere. New direction?...
W E G O A G A I N
Back To The Drawing Board
At this point, after far more hours than I'd care to mention talking/arguing with ChatGPT I shifted into Figma to design a mock up I wanted. This not only sped up the process but removed the nuances.
One positive out of the failures though was testing the concept. I expanded the idea by adding more sneaker options and refining the features. A restart button, which cleared all colours became an important usability improvement, and it quickly became clear that an eyedropper tool was needed to make the process more fluid and enjoyable. Each of these improvements was a direct response to testing and iterating with AI’s help.
S M A R T E R N O T H A R D E R
The Game Changer
It turns out. To make a great design, you need a designer! Once I had the Figma mock as a ref the process was instantly streamlined. The placements and alignments became clearer, and the things like spacing which before took ages to explain got corrected instantly.
The final result was SneakerHues: a simple, interactive web app that lets users pick a sneaker, colour in each section, restart if needed, and download their finished design. The interface is clear and intuitive, built on AI-assisted code shaped by my design direction and iterations. What started as static Illustrator layouts and a dream evolved into a working prototype.
W A S I T W O R T H I T
Takeaways
One big thing I took away from this experience is just how important designers and developers really are. You’d probably expect me to say that as a designer, but truthfully, I went into this thinking the process would be alot easier. With AI popping up in every corner of business, I have the same doubts a lot of people do—wondering if it might make our roles less relevant.
Turns out in my view, it’s the opposite. I realised AI works best with skilled people, not instead of them. It’s great for kicking off boring, repetitive stuff, cranking out a social layout in various dimensions, extending a background or removing one but that just clears the way for more creativity.
The thing is, AI forgets context, skips over things, and often spits out code or “help” that just doesn’t work. Sometimes it feels like you’re going in circles, re-explaining what you meant, troubleshooting its mistakes, or nudging it back on track. That’s a stark contrast to having a good designer or developer sitting next to you. Someone who not only understands what you’re asking, but also asks the right questions, catches the gaps you might not see, and thinks ahead with the bigger picture in mind.
Time this task took:
All together about 13 hours.
All together about 13 hours.
Time this would take with a good designer/dev:
2 hours
2 hours
Number of times I got angry at ChatGPT and felt guilty:
5 times
5 times
