TwojeMeble: a wardrobe configurator that feels native to the shop.
TwojeMeble is a furniture manufacturer, freshly rebranded and building out a new online store — a site we were designing with them at the same time, so I worked directly against their design system, their tokens and their UX behaviours. The configurator lets a customer build a wardrobe from ready-made modules: pick the wardrobe type, choose door styles, adjust height and width, add accessories like handles and knobs, edit the whole configuration, and drop it straight into the cart. The real challenge wasn’t the 3D wow — it was making something immersive that still feels like a native part of the shop, reuses what’s already there, and doesn’t fracture a purchase flow customers already know.
Native to the shop, not a detour.
A configurator easily becomes a strange island inside a store. I designed this one to belong: it reuses the established ecommerce patterns, the same packshots and product imagery, and components built the same way as the rest of the shop. That did two things at once — it kept development lean by not reinventing parts that already existed, and it kept the experience familiar, so a customer configuring a wardrobe never feels like they’ve left the store. The whole thing lives inside the existing purchase flow and ends where every other product does: in the cart.
Immersion without the trade-off.
I owned the sidebar, the full configuration process and the UI; the Omnishelf team handled the 3D implementation of the model that rotates live in the centre. Because the configurator is embedded from an external server, I could push for a full-bleed layout — maximum immersion around the 3D model — without paying for it in the shop’s performance. So the usual tension between a rich experience and a fast one mostly dissolved: the customer gets the spatial, hands-on feel of building a real piece of furniture, and the store stays quick.
Extending someone else’s system, carefully.
The design system I was handed was genuinely good — detailed and considered — so my job was to extend it, not overwrite it. I reused its tokens and patterns wherever they fit and changed things only where the configurator truly needed something new, keeping faith with the decisions the original designer had already made. Where gaps existed I designed the missing pieces — toggle switches, tabs for moving between configuration steps, containers — in the same language. I ran the project solo, from flow research through UX and UI to client meetings, and stayed on through implementation. It’s in development now, with success metrics to follow. Delivered with Omnishelf, who also built the live 3D.