Time First or Your Calendar Eats Everything

Just over 60 people showed up for the second Claude Users Group meetup in Pasadena.

That still feels a little wild to me. I started the group because I wanted a room where folks could compare notes, ask better questions, and talk honestly about what Claude and AI can actually do. Apparently, a lot of other people wanted that room too.

The event went well. I facilitated, asked presenters clarifying questions, and helped spark some real discussion. I could probably be less shitty at vetting talks beforehand, but that’s another essay.

During the meetup, I asked for a show of hands.

“How many of you are using AI to save serious time?”

Hands everywhere.

People talked about ChatGPT, Claude, Gemini, agents, and automations. They talked about saving hours. Some said hundreds of hours. This was different from our first meetup, where a room full of Claude users set me straight, but the same energy was there: people were doing the thing and getting real results.

Then I asked another question.

“How many of those hours have you given back to yourself, to live your life?”

Hands went down.

Silence.

Not fewer hands. None.

And honestly, my hand should have gone down too.

The $30,000 I never felt

A few years ago, I could look at my business numbers and see roughly $30,000 in profit.

On paper, anyway.

Did I feel $30,000 more profitable? Hell no. I never paid it to myself. The money sat in the business, got mixed in with everything else, and eventually disappeared into payroll, software, taxes, contractors, or whatever urgent expense showed up next.

I had made a profit without ever experiencing the profit.

The book that helped me understand what I was doing was Profit First by Mike Michalowicz. I plainly recommend it. Mike’s mechanism is simple: when revenue comes in, you route predetermined percentages into separate accounts. Profit gets taken first. The business learns to operate on what remains.

That changed how I thought about money because it made profit something I allocated, not something I hoped would survive.

Standing in that meetup room, I realized we’re doing the exact same thing with time.

We save 12 minutes drafting an email. We save 20 minutes summarizing notes. An automation saves another little chunk somewhere else. Then all of it gets dumped back into one giant checking account called “the workday.”

Saved time that isn’t allocated isn’t saved. It’s just absorbed.

I know the savings can be real. I’ve written about the automations that quietly save our agency a day a week. But if you don’t decide where that day goes, work will happily cash the check for you.

Of course work ate it

C. Northcote Parkinson named the problem in a 1955 essay in The Economist: work expands to fill the time available for its completion.

Nobody wakes up and says, “Claude saved me 45 minutes, so I’m going to invent enough extra work to consume exactly 45 minutes.”

It just happens.

You polish the document a little longer. You answer three more messages. You squeeze in a task from tomorrow. The empty space closes before you even recognize that it was there.

Michalowicz built Profit First around this same law. If all the money stays visible in one account, spending expands to consume it. His answer was to reduce the cash available for operating expenses and force the business to adapt.

“I’ll use my extra time better” fails for the same reason “I’ll save whatever money is left at the end of the month” fails.

There usually isn’t anything left.

Slack found that workers mostly reinvest AI-saved time into more work: 22% do more administrative tasks, 21% do more work on existing projects, and only 13% spend any of it on non-work tasks.

So the silence in that meetup room wasn’t especially strange. It was the default working exactly as designed.

About those “hundreds of hours”

I should push back on something before this gets too tidy.

When people say AI has saved them hundreds of hours, I’m not always sure the math is fully baked. Self-reporting gets fuzzy. We count the fast output, forget the time spent fixing it, and occasionally compare an AI-assisted task to some imaginary version where we carved the answer into stone.

The broader research is more modest. A Federal Reserve Bank of St. Louis analysis found that generative-AI users reported saving an average of 5.4% of their work hours, about 2.2 hours in a 40-hour week. A large Danish study found time savings of roughly 3% with no detectable change in earnings or recorded work hours.

That does not kill the idea. I think it strengthens it.

Those savings rarely arrive as a clean Friday afternoon with a bow on top. In my work, they show up as scattered little fragments: 10 minutes here, maybe 20 there. You don’t feel them because they never gather in one place.

Neither did my $30,000. It was spread across a year of transactions until Mike’s system forced me to aggregate it and name it.

Time needs the same move.

I’m calling it Time First

I’m openly adapting Michalowicz here. I did not invent allocation, calendars, or leaving the damn office.

First, audit one honest week. Tag the moments when AI actually saved you time. Use real numbers, not vibes. If something normally takes 40 minutes and took 25, record 15. If you spent another 10 correcting the output, you saved five.

Then add the fragments together.

Whatever the weekly number is, allocate a percentage before the next week begins. Michalowicz encourages businesses to start with a tiny profit allocation and increase it over time. For Time First, I think 20% of your saved time is a reasonable place to begin.

If you honestly saved several hours, maybe your first allocation is a single 90-minute block.

Put it on the calendar before the week starts.

And this matters: that block is for life. It is not for lower-priority work. It is not where you finally clean up your inbox or reconsider your project-management software. Go on a walk. Read. Have lunch with somebody. Take a class. Sit outside and do absolutely nothing useful!

I’ve been banging this drum about time away from work for years. I wrote about it a decade ago, and I still suck at consistently following my own advice. Knowing a thing and living it remain annoyingly different skills.

Every quarter, raise the allocation a little, just as Michalowicz raises financial allocation percentages. Let your work adapt to having less of the reclaimed time available.

What if you haven’t created much reclaimed time yet? Start with a few practical automations for a small business, measure what they really save, then claim a piece of it.

There’s still a hard part I haven’t solved: friction.

A separate bank account makes profit harder to raid. What stops a client fire from steamrolling a calendar block?

I’m testing appointments involving other humans. A standing dinner, a class, a training partner, something where another person expects you to arrive. An appointment with someone else is harder to raid than an appointment with yourself.

Is that enough? I’m not sure yet. This part needs more testing, and I’d genuinely like to hear what works for you.

I’m turning Time First into a talk, likely for a future meetup. Until then, try one small experiment tonight: put a single 90-minute “paid back” block on next week’s calendar.

Don’t keep grinding through it.

What would you do with the first hour you actually took back?