I wrote the whole configurator system myself, with no hired developers. The first engine took 2 months, and the product shipped in 3.
Furniture, jewelry and sportswear brands kept losing sales in the same place. A shopper picks a fabric, a metal, a color, a name on the back of a jersey, and then has to imagine the rest from a flat photo and a dropdown. Most of them do not buy. A good share of the ones who do send the thing back.
The tools that fix this already existed. They were priced for enterprise and sold as a monthly subscription, and a small or mid-size brand could not justify that number against the revenue it would move. So they stayed with the flat photo.
I got my first client for a 3D configurator in May 2024, built around their product. By 2025 the same requests kept coming back, so I turned that work into a product.
That product is ConfiguraThor, built for that gap. Enterprise-level customization at a fraction of the cost, no monthly fees, full ownership of what you pay for, and it had to plug into the stores these brands already run on.
I built the first version on PlayCanvas and WebGL, on my own, from the end of August to October 2025. Real-time 3D in the browser, no plugins, with native Shopify and WooCommerce integrations so the configured product lands in the cart like any other.
I used Claude and Gemini to move faster, and the first production version shipped in October 2025, against a 3-month roadmap. The first sale came 2 weeks after launch.
None of the iterations came from me deciding what would be cool to have. They came from one-on-one calls with the people who would use it.
I listened to what was actually breaking in their store, built exactly that, and showed it to them again. Features I had guessed at and nobody asked for did not survive the first call.
Every client configurator started from the same reusable architecture instead of a blank file. That is what brought one build down to 4 to 6 weeks.
Freelance 3D artists made the models. The code was mine.
A promise like 4 to 6 weeks only holds if nothing is improvised. So I wrote a structured asset handoff checklist, agreed with the client and with the artists before anyone opened a file.
It cut iteration time by around 40%, because most iterations were never design problems. They were missing or wrong assets.
PlayCanvas got me to market and I kept improving it with every client. In 2026 I rebuilt the system on Three.js so I could deliver more configurators with the same hands, using AI tools to keep shipping.
PlayCanvas and WebGL for the first engine, Three.js for the rebuild, JavaScript throughout. Shopify and WooCommerce APIs on the integration side, WordPress for the site, Stripe for payments, Claude and Gemini to speed up the work.
Building it was half the job. The other half was getting it in front of the people who needed it, with no audience and no ad budget. That is the Eyedex growth case study.
Their result: 80% less time on manual order management, and 2x revenue on the product in the 30 days after launch compared with the 30 before.
We entrusted Antonio with building our 3D configurator. What struck us was not just the technical expertise, but his presence from day zero: concept, experience architecture, UX details, all the way to go-live. You can tell Antonio has an eye for both design and how a user actually moves through a tool like this. Available, reliable, and, rare, he understood our business immediately without us having to explain everything twice.
Their result: 3 deals closed in the first week after launch.
The configurator completely changed how customers interact with our products. They can finally see exactly what they're ordering before they commit.
Their result: 350 jerseys sold since coming back with the configurator.
Our customers love building their own jersey designs. The visualization removes all the back-and-forth we used to have about 'what will this look like?'
I build from calls with the people who will use the thing. A feature nobody asked for on a call does not get built.
I make the first version reusable. The second client should cost a fraction of the first one, and that only happens if the core is modular from day one.
I fix the process before the code. Most of my iterations were never engineering problems, they were missing or wrong assets.
I rebuild when the foundation stops scaling, not before. PlayCanvas got me to market, Three.js is what let me deliver more of them.
I use AI tools to ship faster, and I still own every line that goes to a client.