The build threshold just moved

A year ago, if I had asked myself, “How do I get 20 LinkedIn posts into WordPress?”, the answer would have been painfully boring: copy and paste. Twenty posts would hardly have justified building software for the job, and spending time finding and learning some specialised tool probably wouldn’t have been worth it either.

This time, I didn’t even look for a plugin. I knew exactly what I needed, so I built it. “There’s no suitable solution out of the box? Fine. Give me 30 minutes and I’ll build it.” That sentence has become something of a recurring experience for me lately.

I’ve always liked making things fit. I used to be ridiculously happy when I stumbled across tiny tools like Espanso, Thaw or KeepingYouAwake: someone, somewhere, had exactly the same oddly specific problem and built the little thing I needed. My former colleague Alex recently did exactly that with Pegel, a tiny Mac tool that lets me dictate into pretty much any text field. Admittedly, that was AI-assisted coding too — which rather proves the point. I’ve also always built, adapted or repurposed things myself when the standard solution wasn’t good enough. AI hasn’t created that instinct. It has massively expanded the territory where I can act on it.

That’s the change I find interesting. For small tools, plugins, integrations and automations, the old choice often wasn’t make or buy at all. It was do it manually, accept an awkward workaround, or simply don’t bother. Now, if I understand the problem well enough to describe exactly what I need, building something purpose-made can be the pragmatic option.

That doesn’t make software engineering trivial. Large systems still need people who understand architecture, security, testing, operations and all the other specialist disciplines. AI can make that work better and faster, but it doesn’t magically make those skills disappear. At the small end, though, the boundaries are becoming much more fluid.

And that makes good APIs, clean interfaces and extensible products even more valuable. When building the missing piece becomes cheap, I care much more about whether I can connect, adapt and extend the products I already use.

The biggest change may not be that software became cheaper to build. It’s that “I’ll just build the missing bit” has become a perfectly reasonable answer to a small problem.