AI Was Supposed to Save Me Time. So Why Am I Working More?

Artificial intelligence was supposed to help me finish work earlier. Automate the boring stuff, solve problems faster, maybe even close the laptop while there was still daylight outside. A lovely idea. Very sensible. Unfortunately, nobody explained the arrangement to my brain.

Because while AI has absolutely made me faster, it hasn’t made me work less. It has mostly helped me squeeze more work into the same day, then quietly expand the definition of “the same day” until it includes two in the morning.

I started using AI to save time. Somehow, I ended up with more projects, more browser tabs, and another plugin I apparently need to build.

The AI productivity paradox: faster work, longer to-do lists

Before AI became my unofficial coding partner, a new idea came with a fairly effective reality check. Building it meant researching, reading documentation, writing code, debugging, and occasionally staring at a PHP error as if prolonged eye contact might resolve it.

I’ve worked with WordPress, PHP, CSS, and HTML for years, and I enjoy figuring things out. But enjoying a challenge doesn’t mean every idea deserves an entire weekend. Some projects stayed on the “maybe someday” list because the time investment simply didn’t make sense.

AI has made that excuse considerably harder to use. Something that once took an evening might now take an hour. A feature outside my usual comfort zone suddenly feels possible. There’s less friction between “wouldn’t it be useful if…” and a working version I can actually test.

That’s a huge improvement. The problem is what I do with the time it saves.

Do I take the afternoon off? Go outside? Discover what people mean when they say they’re “caught up”?

Of course not. I start the next thing. Apparently, my reward for finishing work early is additional work.

Every finished feature comes with three new ideas

My automotive community project, DriveZone.cz, is a good example. What started as a small automotive blog in 2019 grew into a platform with thousands of members, community features, custom tools, and a list of improvements that has never shown any serious interest in getting shorter.

AI helps me build those improvements faster. It also makes the next three ideas look suspiciously reasonable. A custom plugin? That seems manageable. A smarter community tool? Worth trying. An entirely new application? Well, I already have the browser open.

The question used to be, “Can I justify spending several days on this?” Now it’s often, “Could I get a rough version working tonight?” Those are very different filters, especially when you enjoy building things and are dangerously optimistic about what counts as a rough version.

There’s a real benefit here: I can attempt things I would previously have postponed indefinitely. But removing the barrier to starting a project doesn’t remove the need to maintain it. Tomorrow, that exciting little experiment may need bug fixes, updates, and answers to questions I should probably have asked before midnight.

“I’ll just test one thing” is how the evening disappears

The most expensive sentence in my workflow is now, “I’ll just quickly test this.”

It sounds harmless. Responsible, even. Three hours later, I’m debugging a feature that didn’t exist when I sat down, my coffee has gone cold, and I’m explaining to an AI why its fifth proposed fix has recreated the original problem with slightly different formatting.

AI makes starting easy. Stopping remains a manual process.

When a task takes two hours instead of eight, the six hours saved are real. What isn’t automatic is turning them into free time. If there’s something else I want to build, those hours already have somewhere to go. The backlog used to wait for an available evening. Now it can consume one at remarkably short notice.

And not every idea that arrives at 11:47 PM deserves a development environment. Some deserve a note that looks considerably less exciting in the morning.

Faster code doesn’t mean unlimited mental energy

There’s another limit that hasn’t moved much: my ability to think clearly after a full day of making decisions.

AI can generate code, explain unfamiliar technologies, and suggest approaches I might not have considered. I still need to understand the solution, test it, decide whether it fits, and deal with the consequences when it breaks. A confident explanation is helpful. It is also sometimes attached to code that has absolutely no business being near a working WordPress installation.

My work has shifted. I spend less time typing repetitive code and more time reviewing solutions, designing features, testing assumptions, and deciding what to do next. I enjoy much of that work, but it still takes concentration.

Generating another version is cheap. Giving it proper attention is where I start running out of resources.

Maybe I wanted less frustration, not less work

The slightly uncomfortable realization is that I don’t necessarily want fewer projects. I want fewer evenings lost to repetitive setup, obscure errors, and problems somebody has already solved a thousand times.

I want to spend more time making useful things. AI helps me do that, which explains why I keep coming back to it. Having an idea in the morning and testing a working version that evening is satisfying in a way that clearing an inbox has never managed to be.

I wouldn’t want to return to my old workflow. I can learn faster, explore unfamiliar tools, and give ideas a chance that they wouldn’t have had before.

But enjoying the work doesn’t make rest optional. It just makes it easier to postpone.

The part AI can’t do for me

Saving time and having free time turn out to be separate achievements. AI has helped enormously with the first. I’m still responsible for the second.

For me, that means getting better at leaving things unfinished: writing down an idea without immediately building it, accepting that a useful improvement can wait until tomorrow, and closing the laptop even when I know exactly what I’d like to try next.

The ability to build something quickly is a good reason to consider it. It probably shouldn’t be the only reason I say yes.

I’m still working on this. In fact, I had another plugin idea while writing this article. I’ve put it on the list for tomorrow.

That feels like progress.