More work please

My first job was marketing, and I spent it trying to automate myself out of it. I built the pipelines, finished them, and the only payoff was room to do harder work I'd never reached. I thought 2026 would be the year we all did less. I had it backwards: it's the year we do more, just higher up.

This isn't a marketing quirk. Every rung you automate raises the demand on the rung above it. Automate the bottom task and you don't empty the building; you free a floor.

The more work that gets done, the more work there is to do.

The plumber who works himself out of a job

Take the hardest case: an exceptional plumber. No callbacks, no rework, done by noon. He's worked himself out of the job, nothing left to bill. Except the customer catches him at the door (while you're here, the kitchen?), refers him to friends, and stops tolerating tiny leaks, because now they'll just call him. Do the work well enough and you manufacture more of it.

Two centuries of receipts

This isn't particularly new. Economists explain:

• The Jevons paradox, 1865: "[machinery] throws labourers out of employment for the moment. But such is the increased demand for the cheapened products, that eventually the sphere of employment is greatly widened." Cheaper work induces more work.
• The lump-of-labour fallacy, 1891: The false idea there's a fixed amount of work to go around.

Advancement after advancement has proven it:

• 1800s — the power loom automated 98% of the labor in weaving a yard of cloth, and the number of weavers grew anyway. Cheaper cloth meant the world bought far more of it.
• 1980s — the spreadsheet arrived and 400,000 bookkeeping and accounting-clerk jobs vanished. Then 600,000 higher-paid accounting jobs were added. (NPR's Planet Money, citing BLS.)
• 1988 to 2004 — ATMs meant banks could cut tellers per urban branch from 20 to 13. So the banks opened 43% more branches, and total teller numbers rose. (James Bessen, BU Law.)
• 1990s — e-discovery software grew into a billion-dollar business doing work paralegals used to do. But paralegal jobs still grew 1.1% a year faster than the workforce. (Bessen.)
• 2020 — a super elite builder squad at Capgemini Invent finished their part of a much larger project weeks ahead... we got given new, bigger work integrating and optimizing the system as a whole.

The more work that gets done, the more work there is to do.

The ladder isn't free

The rung is real (even if it takes me a few gos to learn it). But the ladder isn't free. Turning a bookkeeper into an accountant is hard, slow, and expensive; turning a support rep into an AI agents orchestrator is harder still. For whoever's asked to climb, it's genuine turmoil, and no long-run optimism pays this month's rent.

The climb is steep. Asymmetric. Unfair.

But, for better or worse; work is here to stay.

Maybe this time is different

AI is climbing the rungs behind us. There may be no safe floor above. Right?

Of the 246,000 tech jobs cut in 2025 (layoffs.fyi), the companies leaning in hardest, shipping in days not years, are hiring more engineers, not fewer. There's likely (I'm guessing here) a backlog of ideas at Anthropic they can't get to. They're shipping faster than ever and still can't keep up with what's worth building. The bottleneck moved up, past the hands, to deciding what to build next. And yet, all hires, no fires.

To the executives and leaders

AI is coming for your industry whether you're ready or not. Stock up. Not on savings, on work. On every idea you ever shelved for lack of hands.

When you give the word and agents descend, two doors open. Walk through one and you keep the same work, done by fewer hands, cheaper and faster. But standing still.

The other turns every hand you have toward every idea of value you shelved.

One door saves money once.
The other never stops.

To the builders, the tinkerers, the doers

The tasks you do today are the most automatable, the most boring thing about your work. Let them go.

Your value is in finding the next problem, the itch to fix it, the hunger for the one after. Automation doesn't replace the person who finds the work worth doing.

Be the one thinking three problems ahead while your agents solve the first. That person doesn't get automated. That person gets an army.

So when the work is done by lunch, and someone turns to you and asks "what's next?"... don't flinch. Have an answer. Have a hundred.

More work is coming.
Let the agents have the grind.
Keep the part you'd stay late for anyway.