← Carlos Cardona

The End of the Semicolon Era

Every developer of my generation has the same scar. You spend three hours debugging, questioning your career, your stack, sometimes your life choices, and the answer turns out to be a missing semicolon. One character. Three hours.

That semicolon is a good symbol for something bigger: the enormous share of a builder's time, effort and cognitive capacity that used to go into mechanical correctness. Syntax, boilerplate, plumbing, the discipline of not making small mistakes in large quantities. We called it programming, but a lot of it was just groundwork. AI has absorbed that layer almost completely, and it is not stopping there.

Because it is not only the mechanical work. Layers we used to consider craft are dissolving too. A few years ago, choosing a frontend framework was a serious debate. React or Vue, then Svelte came out, and knowing how each worked and what it implied for your project was part of your value. Today I run a frontend built in React and I do not know the first thing about React. I built it, I modify it, I maintain it. It works. The decision that used to be a craft debate has become trivial, and more decisions are heading the same way.

So the honest question is the one a CTO asked at a meetup I attended recently: what will software developers actually be doing in three years?

Here is the thing. Two hours later, the same person described his own workweek to me: recruiting, building the team, choosing tools, setting strategic direction, designing the system. He leads seventeen developers and barely touches code. He had already answered his own question without noticing. What happened to him through promotion is now happening to the whole profession through tooling: the role moves up one level of abstraction. From the semicolon to the roadmap. And it will keep moving up.

This is where I think the real shift is, and why I do not buy the pessimism. The time and cognitive capacity that used to be consumed by groundwork does not disappear. It gets released. You can spend it thinking about the product, talking to users, designing the system, automating your own process. The developers who are thriving right now are simply the ones redirecting the freed capacity upward instead of defending the old altitude.

Which changes what is scarce. When everyone has the same tools, the ability to build fast stops being an advantage and becomes table stakes. It reminds me of English in Spanish-speaking countries: two generations ago speaking English was a real edge, today not speaking it is the problem. Nobody gets hired for speaking English anymore. Nobody will be hired for making code run anymore either.

What remains scarce is what the tools do not provide. Appetite: actually wanting to go get the outcome, not waiting for a spec. Taste: knowing what good looks like when a machine can produce twenty versions of mediocre. Judgment: seeing what a choice will cost you six months out. Agency: moving without being pushed. These were always the traits of the best people. The difference is that before, they sat on top of years of groundwork, and now they are the job itself.

The semicolon era is over. Nobody is going to mourn it. The question worth your attention is not whether AI will take the mechanical work, it already has. The question is what you do with the hours it gave back to you.