You Can Just Build It Yourself

The Final Frontier Was a Grid

For as long as I have been building UI, the DataGrid has been the control I measured everything else against.

It is the hardest thing in the toolbox. It has to be fast. It has to handle 100,000 rows without choking. It needs sorting, filtering, grouping, frozen columns, some bullshit display option because the Product Manager said so, inline editing, and keyboard navigation, and all of that has to feel right with a mouse and on a touch screen. Every UI framework I have ever used had one, and every one of them was expensive, limited, or both.

I spent years taking runs at my own. Different architectures, different rendering approaches, a few that got far enough to be embarrassing when I abandoned them. Slow progress, measured in weekends.

Then, basically overnight, it was done. I sat down with an LLM and jammed out KumikoUI, a fully canvas-drawn DataGrid for .NET MAUI on SkiaSharp. Sorting, filtering, grouping, summaries, editing, drag and drop, a 100K-row stress test in the sample app. The pieces that had stalled me for years, text input and keyboard handling on a canvas, clicked in a few sessions because the feedback loop finally ran in minutes instead of weekends. I wrote about the release back in March, and it is MIT licensed and free.

It is not perfect. There are rough edges and APIs that will change, but the control I assumed I would be chipping at for the rest of my career took days.

A Vendor’s Whole Catalog in a Week

If I can do that with one grid on one platform, Syncfusion or Telerik can do it with the whole catalog. Syncfusion advertises more than 1,600 components across every platform Microsoft has ever shipped. Telerik’s DevCraft bundle starts at $1,149 per developer per year. That business, selling a suite license by the developer by the year, was built on the fact that a DataGrid or a scheduler or a rich text editor was hard enough that paying for it beat building it.

That math just flipped. If you are one of those vendors, you can stand up a version of your own library for whatever platform you want in a few days. Uno, Avalonia, Blazor, pick one. And if you are cool, you open source it and hand it to the world, because the thing you were charging for is no longer the thing that is scarce.

The part I like most is the part nobody talks about. You do not have to buy into the ecosystem anymore. When I bought a control suite I was paying for 1,600 things to get the handful I needed, and dragging the rest around in my dependencies forever. Now you build the handful and design them exactly the way you want, and nothing else.

A Journalist Wrote a Word Processor

The first great example I saw of a non-developer doing this was Harry McCracken at Fast Company.

He has been writing about technology for decades, and he apparently has used more word processors than most of us have heard of. In April he wrote about building his own with Claude Code. He calls it Doolee Write. It is web-based, it is stripped down, and it does not pretend to compete with Word or Google Docs on features. By his own account the essentials came together in a couple of nights, and he kept going until his wish list ran out.

That is the part that matters. It is not as full-featured as the alternatives, and he does not care, because it is the tool he wanted. Nobody was ever going to sell him that. A market of one is not a market, right up until the cost of building for a market of one drops to a couple of evenings.

A Decade of Snagit, Replaced by a Side Project

My favorite recent example, though, is Tiny Clips from James Montemagno.

I have paid for Snagit for probably a decade. It is an alright app. Not great, but serviceable. It captures the screen, it has an editor, it does the job. TechSmith rarely updates it in any substantial way, but they are glad to bill you $39 a year for it, and somewhere along the way it became Camtasia Snagit, a door into a bigger suite of things you could also be paying for.

Tiny Clips lives in the menu bar on a Mac or the system tray on Windows. Region, window, or screen capture. PNG, MP4, GIF. Scrolling capture, OCR, a webcam overlay, a trimmer, global hotkeys, auto-updates. No account, no telemetry, no paid tier. It is MIT licensed on GitHub, it is in the Mac App Store and the Microsoft Store, and I think it is better and more modern than Snagit in every way that matters to me.

This was not a company. It was not a team, and it was not anybody’s full-time job. It was one good developer working with an LLM to build the software he wanted, and the first commit landed in February. Think about that for a second. One person, a few months, and a fully functional alternative to a decades-old paid app.

Now Scale That Up

So imagine you are a company with a team of full-time developers.

You are paying monthly for some piece of software you need to run the business. A ticketing system, inventory, the thing the warehouse scans into. You hate it. Or you do not hate it, but it does not do what you need, and every feature request goes into a portal and dies there. You pay anyway, because the alternative was building it, and building it was a multi-year project with a team you did not have.

You can just write it yourself now.

Take the parts you actually like and emulate those. Cut every bit of fat you have been paying for and never once clicked on. Build the workflow the way your people actually work instead of the way a product manager in another state guessed they might. When you need it to change, you change it that afternoon.

I honestly do not think most companies have absorbed this yet. The instinct to buy is deeply ingrained and years deep in the corporate DNA. The thing that made buying the right call, that building was slow and expensive and risky, is the exact thing that just evaporated.

Yes, Somebody Has to Host It

I can hear the objection, and in fairness it is a real one.

You have to host it. You have to maintain it. You have to be the one who gets paged when it falls over at two in the morning, and for some categories of software that is a serious argument. I am not going to tell anyone to rewrite their payroll system over a long weekend. Compliance, uptime guarantees, data you cannot afford to lose, those are legitimate reasons to keep paying someone else.

But even then, you do not have to go all in on day one. Most of this software has an API. Start there. Build the piece you wish it had on top of what it already exposes, and keep paying for the parts underneath. When that goes well, snap off one more feature. Then another. At some point you look up and the subscription is doing very little, and turning it off is an afternoon decision instead of a migration project.

The hosting problem also gets smaller the more of this you do. The first internal tool you own feels like a risk. By the fifth one it is just part of how the team works.

This Is Only the Beginning

I am starting to see this pick up with the companies I work with. Renewals get a harder look than they used to. People are asking what the thing actually does for them, and more often than not the honest answer is a version of it they could build themselves.

One of my clients has already replaced JIRA, which is software that everyone hates and wants a better solution for, with a custom-built tool that fits their exact workflow and needs. I feel like is the prime candidate for being rewritten internally. It is something you use every day, know it doesn’t fit your needs and is a constant agitation. What they have generated over the matter of weeks is a fully functional, tailored solution that meets their needs far better than the off-the-shelf alternative ever could. They have ideas in an afternoon and the implementation is done and usable that same day.

This is the same shift I wrote about in Software’s Flat-Pack Era and Lights Out on Offshoring, seen from the buyer’s side this time. When software is cheap to make, the companies whose entire business rests on it being expensive have a problem, and they have it now.

I do not know how fast it goes. Probably slower than I think, because habits are sticky and most companies are not run by the kind of person who builds a word processor for fun. But a grid I chased for years took me days. A journalist built a word processor in a couple of nights. One developer replaced an app I paid for for a decade with a side project.

If you are paying for software you don’t love, you have more options than you did a year ago, and it is just a prompt away.