Design it, build it.

A kitchen is designed three or four times before anyone cuts a panel. Once by the designer, in a modelling tool that knows nothing about how cabinets are made. Again in a drawing package, so the plans look right. Again in a spreadsheet, to work out what it costs. And once more at the factory, where somebody redraws the whole thing in a system that actually drives the machines.

Every one of those passes is a chance to lose information. A gap that was fine on screen becomes a filler nobody priced. A door style changes and the hardware list does not. The person paying for the room absorbs all of it — in delays, in change orders, in a finished kitchen that is subtly not what they approved.

The strange part is that none of this is a modelling problem. The tools draw beautifully. They just do not know what a cabinet is. To them a carcass is six rectangles, not a part with a thickness, a joinery method, a hinge that needs 35 mm of clearance, and a panel that has to come out of a standard board.

So we built the other way around. In Mixloom a cabinet is a real object: it knows its own construction, and it adapts when the space changes. Drag a run against a wall and every carcass, door, drawer, and filler resizes to the opening. Change a width and the joinery, the hardware, and the panel sizes follow — because they were never separate from the geometry in the first place.

That single decision is what makes the rest possible. If the model knows the parts, it can price them honestly. It can check them against real production rules — board sizes, edge banding, hardware, machine limits — and tell you at design time that something cannot be made, instead of at the factory three weeks later. And it can export what the workshop actually needs: a cut list, DXF, drilling, machine programs, and a hardware list that matches the design the client signed off.

This is also why we care about both ends of the job. Someone planning their own kitchen and a factory running hundreds of them are working on the same file, from opposite directions. The homeowner needs it to be obvious. The factory needs it to be exact. A tool that only serves one of them pushes the work — and the errors — onto the other.

We are not trying to replace craft. The people running the saws know things no software will encode. What we want to remove is the re-drawing, the re-typing, and the quiet loss of intent between a good idea and a built room.

If you can design it, the factory should be able to build it. That is the whole product.