Complete
Desk

This grew out of an earlier furniture exercise, filed here for years as FirstTable, where the point was simply to design a desk properly: fit a person, fit a room, carry equipment, and be practical to build. That project answered the engineering question and raised a commercial one. Desks get replaced not because the surface wore out but because the setup around them changed.

Complete Desk is the attempt to design that problem away. If the desk exposes a fixed physical interface, then upgrading your setup means buying a module, not a desk.

The interface came first and was drawn precisely. On the medium top, forty-five inches by twenty-two, three identical ports sit in a row along the back edge, each a stepped rebate: a wider shallow pocket over a smaller through-cut, with radiused corners so a router or CNC can actually cut it and a lip for a module to sit flush against. Underneath, a grid of mounting holes at roughly six-inch spacing gives anything that hangs below the surface somewhere to bolt to.

With the interface fixed, the rest becomes module design: a work lamp, a wireless charger, a tray, a two-tier shelf, single and dual monitor arms. All of it modeled in SolidWorks, with vendor CAD for a real monitor arm pulled in to fit-check against rather than trusting my own assumptions about clearance. Three desk sizes, each configurable by top finish, leg style, and leg color.

Then the parts of a product that are not the product. A brand and logo system. A cost model built from the bottom up, per desk, per order, and fixed startup cost, based on an MDF core with a laminate or veneer surface and edge banding, which is what the machines I was quoting could realistically produce. And a storefront: a custom Shopify theme with a desk configurator I wrote by hand in vanilla JavaScript, a multi-step wizard holding size, finish, and leg state, compositing a layered preview image, mapping the combination to a purchasable variant, and totalling the price with add-ons live.

A complete design and go-to-market package that has not yet been manufactured. The CAD exists for the desktops, three leg variants, and eight accessory modules. Two dimensioned drawings pin the port geometry and the underside hole grid. The brand identity is finished. The storefront is built, with a working configurator, a landing page, and preorder and waitlist capture, and the unit economics are modeled down to glue and electricity.

What does not exist is a physical desk. Every image of the product is a CAD view or a marketing render, the store is a preorder page rather than a shipping one, and the machine to cut the ports was still being sourced when the project paused. Some material and pricing figures in the pitch material are draft assumptions rather than quoted facts, and I am not going to repeat them here as though they were measured.

The engineering lesson is that in a modular system the interface is the product. The lamp can be redesigned next year, the legs can change, the finishes can rotate, but the pocket dimension is a permanent promise to every customer who ever buys a module. Getting a stepped rebate and a hole grid right matters more than any individual accessory.

The harder lesson was about sequence. I designed a full product line, a brand, and a storefront before cutting a single port in a single sheet of MDF. The correct order was the opposite: make one desktop with one working port and one module that drops into it, and let that tell me whether the tolerance, the finish, and the cut time were realistic. Renders cannot fail a fit check.

If it restarts, it restarts with a prototype and a fit test, not a pitch deck.

• Costs, margins, and market figures in the draft pitch material are unverified and deliberately not quoted here.
• Native CAD lives on another machine; what survives in the archive is drawings, exports, and views.