Software Worth Building for One Person

There is a thin touchscreen strip under my main monitor, a Corsair Xeneon Edge. I tap a tile and four applications rearrange themselves into the exact layout I use for focused work. Another starts a timer. A third shows what is playing and skips a track without my reaching for the mouse. None of this software should exist. No company would fund it, because the total addressable market is one person, and it runs on a piece of hardware most people have never heard of. I built it anyway, and why I could is the actual subject here.

For most of the history of software, build cost set a floor under what was worth making. Writing an application took enough skilled hours that it had to serve a lot of people to earn those hours back, which meant almost nothing specific to one person’s desk ever cleared the bar. A capable coding agent moves that floor. When the hours collapse, the arithmetic changes, and it becomes rational to build software for an audience of one: one desk, one workflow, one odd device you happen to own.

There was a real problem to solve, which is the only honest reason to build anything. I run three monitors, a wide Samsung ultrawide with two Dell displays stacked above it, and across that much glass I am forever hunting for one window or dragging an app two screens over. The work itself changes constantly on top of that: on a given day I switch context every half hour, up and down in altitude between executive, operator, and builder, and sideways between meetings, building agents for different client teams, working on tooling, and writing. Each mode wants its own set of applications arranged its own way. A tile that tears down one context and rebuilds another in a single touch is worth real money in time and attention. But wanting a tool has never been the constraint. Being able to afford to build it was.

So how affordable is it, really. The honest answer is that this took a week rather than a weekend, and parts of it were genuinely hard. The work did not disappear. What changed is which work a single person can now finish.

The panel exposes its touchscreen through no documented interface at all. Corsair’s published vendor ID turns out to be only the monitor’s control channel; the actual touch stream comes from a separate, undocumented controller baked into the display, reporting on a different USB identity entirely, in a raw seven-byte format you have to decode by hand. A few years ago that is where a project like this quietly dies, or where it acquires a contractor who has done HID work before and a week on the calendar.

What happened instead is that two people had published open-source drivers for this exact panel, and the agent read both, cross-referenced them against what the device actually emitted, and reconstructed the working sequence in an afternoon. I am not an HID specialist and did not become one. The specialist knowledge existed in the world, the agent could absorb and apply it, and terrain that used to require a person with that background in the room was simply crossable by me at my desk.

That is the whole economic shift, in miniature. The work stayed hard; what changed is the size of the problem a single motivated person can carry end to end. Reverse-engineering an undocumented device, wiring low-level system input, debugging a native app’s render loop: work that used to sit behind a team or a specialist now sits behind an afternoon of directing an agent that has read more driver code than I ever will. The floor did not vanish. It dropped far enough that a whole basement of one-off tools came within reach.

So I built a control center for a touchscreen almost nobody owns, to solve a problem only I have, and it was worth it. That sentence would have been absurd for most of my career. The software many people depend on still has to earn a team and a business case, and it still will. Underneath it, there is now room to build for exactly one user, on exactly one desk, because the cost of crossing the hard parts finally fell below the value of having the thing. Importance was never the bar. For the first time, a tool this small cleared the only bar that used to matter, whether it was cheap enough to be worth building at all, so I built it.

If you want to see the thing up close, the widgets it ended up with, the war stories, and the dead ends I hit, I wrote about the build itself separately in the next article in this series.