AI isn't making work disappear. It's moving the bottleneck.

A lot of the early fantasy made it sound like better tools would mean less labor. That's not how it feels from inside the work. It feels more like getting pushed up a layer of abstraction. The mechanical parts shrink. The orchestration expands.

I see this most clearly in coding. I haven't really been manually writing code line by line for a while now. Instead, I'm running agents across multiple codebases at once, giving them longer tasks, checking plans, deciding when to use a faster model versus a more capable one, and figuring out when to parallelize versus when to focus.

Those are still work problems. They're just different ones.

the economics don't stay still

The naive hope is that if I can now produce 10x more, maybe I get to work one tenth as hard for the same result.

That's almost never how technological shifts work.

If everyone suddenly gets more leverage, the baseline moves. The market doesn't keep paying yesterday's rates for yesterday's difficulty just because the output used to be expensive. Expectations rise. The new tools don't just change what is possible. They change what counts as normal.

That's the sobering part of this.

There's something a little brutal about how quickly new capacity turns into new demand. Once more output is possible, more output starts feeling required.

what gets unlocked

But I don't think that's the whole story, and it's definitely not the most interesting part.

The most interesting part is all the valuable stuff that used to sit below the priority threshold.

Not impossible. Not even unimportant. Just too annoying, too time-consuming, or too administratively heavy to justify.

Today I used an agent to look through bank statements and flag recurring subscriptions I should pay attention to. That's the kind of task that used to produce immediate resistance in me. Not because it was hard, but because it was just frictional enough to invite procrastination.

And that's what feels new: AI doesn't just save me time on the tasks I was already going to do. It gets me to do things I probably would've neglected.

The same thing shows up with taxes, paperwork, and all the little forms of financial hygiene I know are good for me but don't naturally call to me. I don't need a separate service, another login, another subscription, another funnel trying to sell me harder on the problem. I can just do the thing.

Sometimes the breakthrough isn't speed. It's lowering the activation energy enough that the task finally happens at all.

what finally clears the bar

I remember talking to an engineer at Trimble about this. He said their work had changed because they could now take on projects that simply would've been too prohibitive before.

That word matters: prohibitive.

A lot of valuable work used to lose not because it lacked value, but because it couldn't compete on cost. It would never float to the top of the backlog.

Now more of it does.

Things like:

• refactoring a codebase that technically works but has become unpleasant to touch
• adding tests that everyone knows would help but nobody could justify this sprint
• cleaning up copy and rough UI edges
• making internal tools understandable enough that designers and product managers can safely contribute
• shaping code so it is easier for both humans and agents to work with

This is what I keep noticing: AI doesn't just help us do the obvious work faster. It changes which work becomes worth doing.

the fun part

That's also why I'm more excited than alarmed most days.

There are projects I've wanted to build for years that were always sitting just beyond the edge of feasibility. One example is a devotional app that can pull together prayers and writings by theme, cite the original source, and let someone explore a faith tradition without manually digging through everything themselves.

That idea used to feel like one of those "someday" projects. Interesting enough to imagine, too expensive to prioritize.

Now it is buildable.

I get to play a little mad scientist with ideas that used to be too expensive to indulge.

And I think we're still very early in that curve. Most people haven't fully integrated AI into the surfaces of their lives yet. We still tend to talk about it as a chatbot or a coding assistant, when the bigger shift is that a single person can now push far more ideas across the line from interesting to real.

That's where the solo-company fantasy comes from. Not because one person suddenly becomes superhuman in every dimension, but because so much previously prohibitive work is no longer prohibitive.

the line that moved

So no, I don't think AI is going to make work go away.

I think it's going to keep changing the level at which we work. It will raise expectations. It will create new coordination problems. It will probably intensify certain economic pressures too.

But it will also make a huge amount of neglected, useful, meaningful work finally clear the bar.

And maybe that's the more important shift to notice.

The future isn't a world where nothing is left to do. It's a world where many more good ideas become worth doing.