You're Not Getting Your Time Back
In 1865, William Stanley Jevons published a counterintuitive observation about coal.
As steam engines became more efficient, British coal consumption went up — not down. The efficiency gain made coal cheaper to use. So they used more of it. Every unit of efficiency gained was immediately converted into expanded usage.
We’ve been calling this the Jevons Paradox for 160 years. It’s a foundational principle of resource economics.
Nobody applied it to cognitive labor. Until now.
The Berkeley data
Last February, HBR published findings from a UC Berkeley field study — 40 knowledge workers, eight months embedded inside a 200-person tech company. Engineering, product, design, research, operations. Real workflows.
The question: what actually happens when knowledge workers integrate AI?
67% of workers who adopted AI tools were, by year’s end, working more hours. Not fewer. Not the same. More.
The researchers called it workload creep.
Here’s the mechanism: AI reduces friction per task. Lower friction means you start more tasks. Starting more tasks feels achievable because each one is easier now. Easier tasks widen the scope of what you’ll commit to. A wider commitment footprint creates more coordination overhead. More coordination overhead demands more hours.
You got faster. The work got bigger.
Why discipline won’t fix this
The obvious response: be more selective. Say no. Don’t let AI trick you into overcommitting.
That framing treats this as a willpower problem.
It isn’t.
The Jevons Paradox is structural. When you reduce friction for a unit of work, you change the calculation that happens before the work starts. The cost/benefit of committing to something shifts. What used to feel unreasonable now feels achievable.
Your brain updates. So does your boss’s brain. So do your client’s expectations. So does your team’s planning horizon.
This is not solved by personal discipline. It’s solved by changing the constraint.
The layer you don’t have
Here’s what the research is actually pointing at.
Productivity tools — every single one — operate below the level of commitment. They help you do the things you’ve already agreed to do. They reduce friction in execution.
None of them manage what you’ve committed to in the first place.
There’s no tool that sees the full picture of your obligations — across email, Slack, calendar, tasks, AI-initiated workflows, verbal agreements made in passing — and tells you: you’re at capacity. Stop.
So every time you save an hour with AI, you fill it. Sometimes intentionally. Usually by osmosis. The social, professional, and psychological pressure to fill available capacity is constant and ambient.
Efficiency gain → expanded commitments. Jevons, in a button-down.
What coordination actually solves
Most of the conversation about AI coordination focuses on execution — getting agents to talk to each other, reducing handoff errors, orchestrating multi-step workflows.
That’s not the core problem.
The core problem is commitment sprawl. The accumulation of open loops, active obligations, semi-started things, and half-remembered promises sitting somewhere in your stack, drawing down attention you’re not aware you’re spending.
You can’t move faster out of this. Faster is what got you here.
What you need is something that knows what you’ve committed to. That tracks obligations — not just tasks — across channels and time. That knows when you’re at capacity before you do. That closes loops you’ve already opened before you open more.
Not a productivity tool. A coordination layer.
A productivity tool helps you execute faster.
A coordination layer manages what you’ve agreed to execute at all.
The difference matters. Especially now.
The 67% are not failing
They’re responding rationally to a system designed without a commitment constraint.
When AI lowers the cost of starting something, you start more things. When you start more things, you own more open loops. When you own more open loops, your cognitive overhead increases. When your cognitive overhead increases, you work more hours — not because you’re undisciplined, but because the overhead follows you home.
The Jevons Paradox isn’t a warning about waste. It’s a warning about efficiency without a governing constraint.
In knowledge work, that constraint has a name: coordination infrastructure.
Until you build it, the faster you get, the busier you become.
