Here’s something that sounds obvious once you say it out loud: the people who build software are almost never the people who use it.

A team of engineers at some company in San Francisco built the case management system your law firm uses. Those engineers have never practiced law. They’ve probably never spent a Tuesday afternoon trying to find a specific client note buried in a filing system that makes perfect sense to a product manager and zero sense to a paralegal. But that paralegal uses the software anyway, because what choice does she have?

That’s the premise buried in The Verge’s piece today on what they’re calling the personal software revolution, and it’s worth sitting with for a minute. We’ve accepted this arrangement for fifty years. Software is made by one tribe and used by another. The using tribe adapts. Always.

That’s over now. Or it’s starting to be over, which amounts to the same thing.

What Vibe Coding Actually Is

“Vibe coding” is the slightly goofy name that’s stuck to the practice of building software by describing what you want in plain language to an AI, then iterating. No syntax. No Stack Overflow. No bootcamp. You say “I want a tool that tracks my clients, sends them reminders three days before their appointments, and shows me who hasn’t paid,” and the AI builds it.

The results aren’t always pretty. They’re often brittle. A real developer will look at the output and wince a little. But here’s the thing: it works. For a specific person, with specific needs, well enough to actually use.

That’s a genuinely new thing. Not a better version of something we already had. A new thing.

The Monopoly Nobody Noticed

Think about what software actually is, at its core. It’s a set of decisions made by someone else about how you should do your work. What fields exist in this form. What the workflow looks like. What counts as “done.” These decisions get baked in, and then millions of people shape their behavior around them.

This is fine when the decisions are good. It’s a slow tax on productivity when they’re not. And they’re often not, because the people making the decisions are optimizing for the median user, not for you.

Enterprise software is the most extreme version of this. The bigger the company selling it, the more the product has been sanded down to fit everyone and therefore fits no one perfectly. You end up with Salesforce fields your company has repurposed into something unrecognizable, or Excel spreadsheets held together with macros because the actual software doesn’t do quite the right thing.

Small businesses and individual professionals have it worse, because the enterprise tools are too expensive and the consumer tools are too simple. The paralegal I mentioned earlier? She’s probably using a combination of Google Sheets, a generic calendar app, and a notes folder that she has spent years organizing in a way that only makes sense to her. That’s her “software.” It works. It’s also deeply inefficient.

Who This Is Actually For

The tech coverage of vibe coding tends to focus on builders. People making fun little apps for themselves, productivity tools, hobby projects. That’s real, but it’s not the interesting part.

The interesting part is professionals with highly specific needs and no budget for custom development. The therapist who wants a HIPAA-compliant note-taking system that matches exactly how she thinks about her sessions. The restaurateur who wants an inventory tracker that accounts for the specific way his kitchen works. The high school teacher who wants a gradebook that weights things differently by semester and automatically flags students who missed more than two assignments in a row.

None of these people are going to hire a developer. The price doesn’t make sense. But all of them would use a tool built exactly for them if the barrier to creating it was “describe what you want.”

That’s where we’re heading. And it matters more than most of the AI features getting announced this week.

The Android Thing Is the Opposite Impulse

Speaking of AI features getting announced this week: Google is rolling out something called “contextual suggestions” on Pixel 10 devices. It watches your habits and predicts what you’ll do next. Arrive at the gym, and it suggests your usual playlist. Open Maps in the morning, and it guesses you’re going to work.

It’s not a bad feature. It’ll probably be useful. But notice what it’s doing: it’s a company using AI to predict how you’ll interact with software they built. The AI adapts to you, but the software itself is still fixed. You’re still living in their world. They’ve just gotten better at guessing how you move around in it.

That’s a fundamentally different impulse than what vibe coding enables. One is AI as a smarter interface layer over the same old locked box. The other is AI as a tool for building your own box.

Both will exist. But only one of them breaks the fifty-year arrangement.

The Real Question

The honest version of the pushback here is: most people won’t actually do this. Building software, even with AI help, still requires a certain tolerance for iteration and debugging. You have to be willing to describe what you want precisely, notice when it’s wrong, and go back and fix it. That’s not trivial.

Fair. But look at who’s doing it already. Teachers. Small business owners. Researchers. Writers building their own editorial tools. These aren’t developers. They’re people with a specific pain and a low enough barrier now to do something about it.

The question isn’t whether everyone becomes a software builder. They won’t. The question is whether the population of people who can solve their own software problems grows from a few hundred thousand to a few million. It will. It already is.

And when it does, the strangest monopoly in modern life quietly ends. The people who use software and the people who build it won’t be quite so separate anymore.

That’s worth paying attention to.