The full essay · on X
More ideas than time. Until AI. →

I've always had more ideas than time to build them. Most people who make things do. The list keeps growing and the hours don't, and for years I filed that under personal failing: weak discipline, bad planning, not enough sleep. Took me too long to notice it isn't a discipline problem at all. It's an exchange rate.

The budget nobody prints more of

You get a few decades at full power, the years when skill, energy, and nerve actually overlap. That window is the whole budget. Everything you'll ever make, you make inside it.

Most of it gets spent on things that were never the point: formatting, admin, the same lookup done for the fourth time, the small loops between having an idea and holding the finished version of it. You can borrow money, save it, move it around in time. Hours don't work like that. Time is the asset sitting under all the others, and the only one you can't make more of.

A bad exchange rate dressed up as virtue

The usual answer to "I want to do more" is "work more hours." But work is the one thing that never runs out. Finish today's list and tomorrow prints a fresh one. Your life does run out. So the standard ambition move is to trade the thing you can't get back for more of the thing you already have too much of.

Bad deal. "Locked in" culture just sells it back to you as a personality.

The move isn't to find more hours. It's to change what an hour buys.

What the agents are actually for

This is where the agents earn their keep. They run on energy and money: compute, model bills, a stack of subscriptions. What they hand back is cognitive labor. Scout researches. Builder implements. Critic tears into weak reasoning before the market does. A tiny crew running the loops that used to tax a whole human life, so the human can spend that life on the part only they can do.

Pointed the right way, that turns money and energy back into time. Pointed the wrong way, it just makes you a faster employee of your own calendar.

The hard half

Getting the rate is the easy part. Spending what it gives back is harder.

When you can build almost anything, taste turns into the bottleneck. Volume stops being the constraint, so the only question left is whether the thing deserved to exist at all. And the busywork you handed off doesn't disappear. It comes back as a different job: reviewing, correcting, keeping five half-built things balanced in your head at once.

I don't have clean answers for that yet. Pretty sure that's the actual work now, the part no agent gets to do for you.

Humans got good at turning energy into work, then got good at pricing that work across time. AI adds one more trick: turning a little of it back into hours. What you do with those hours is the only part that stays yours. I want two things out of mine: work that outlasts the afternoon, and a life I was actually present for. The agents are just the weird machinery that makes both possible at once.


Time is the only budget you can't refinance.