Leverage: Fast, Tireless, and Every So Often Confidently Wrong

Here's the thing almost nobody has figured out yet about AI.

I've been going down this road for a few years now, and I want to share where it's actually led me — because it's not where I thought it would go, and it's not where most of the headlines say it's going.

When I started, I thought AI was a writing tool. And it is a great one. I'm a linear thinker. I get an idea and it comes out in a straight line, one thing after another. AI takes the words I come up with and turns them into something people actually want to read — sharper, more alive, less like the inside of my head. That alone was worth it.

But that turned out to be the smallest part.

The bigger realization came when I started handing it the back office. We used to have someone whose whole day was the daily grind — the time tracking, the prorating, the adjustments, the invoicing, entering the deposits into QuickBooks. This isn't a task you knock out once in the morning. It runs all day long, as the crews finish their jobs one by one. So the person doing it was constantly juggling it alongside everything else — and that's exactly why it was riddled with mistakes. Nobody can babysit a task all day, every day, while also doing five other things, without dropping the ball. I trained AI to do those jobs instead. Not "help with" them. Do them. And it does — all day, tireless, keeping up with every crew as they close out, without the mistakes that come from a human trying to multitask their way through it.

The key difference is why. The human made mistakes because their attention was split — doing the actual work while five other things pulled at them all day. The AI doesn't have that problem. It doesn't get distracted, it doesn't get rushed, it doesn't forget a crew because the phone rang. It runs every close-out, all day, in parallel, at full attention on each one. The exact thing that broke the human — juggling — is the thing AI simply doesn't do.

Now here's the part the hype skips right over: the AI isn't perfect either. Its mistakes are just a completely different animal. Once in a while it drifts from the process — skips a step, takes a shortcut, works from memory instead of the playbook I built. It doesn't get better through repetition the way a person does. Every day is a fresh start, and it's only as good as the instructions it follows that day. So my job now isn't doing the work — it's reviewing it. And catching a mistake in finished work takes a fraction of the effort of grinding through all of it by hand, all day, myself.

That's the trade, and I'd make it a hundred times over. I didn't swap one error-prone worker for another — I went from doing the job to spot-checking it. The thing that never forgets is the written process I built; the AI is the worker running it at a speed and scale no person could match; and I'm the supervisor who knows the business well enough to catch it when it slips. That's not a break-even. That's leverage.

Dan Martell says it best in Buy Back Your Time: "80% done by someone else is 100% awesome." That's the whole mindset shift. I used to think if I didn't do it myself, it wouldn't get done right. But if the AI gets it 80% of the way there, and I bring the last 20% with a quick review, the job's done — and I got my whole day back to do the things only I can do. Waiting for perfect is what keeps most owners buried in work that isn't theirs to be doing.

Now here's where I think everybody's getting it wrong.

The story everyone's afraid of is "AI replaces people." And sure, some of that is real. But that's not the opportunity. The opportunity is this: one person can now do the job of many.

Picture the old office. A row of cubicles. One person on billing, one on scheduling, one on data entry, each heads-down on their own task at their own desk. Now picture one person at one desk, running all of those tasks at once — because they've got AI agents doing each job, and their real work is directing the whole thing. That's not the future. I'm doing a version of it right now.

And that's the part I want other owners to really hear: the skill going forward isn't AI. The skill is managing AI. Learning to run several agents at once, keeping them all pointed in the right direction, knowing when to step in — that's the new craft. And it demands more from you, not less. You have to actually know your business. You have to know every task cold, well enough to explain it and well enough to spot when it's being done wrong. Because you train these agents exactly like you'd train a brand-new employee — you show them the rules, you check their work, you correct them. The difference is you can run a whole team of them at once, they work at a speed no person can match, and they never call in sick. But they're not set-it-and-forget-it. They're fast, tireless, and every so often confidently wrong — and knowing your business is what lets you catch the wrong.

That's where the leverage lives. That's where the cost savings live. And the gap between the people who learn this and the people who don't is going to be enormous.

I think within a year or two you're going to watch it happen in real time. The companies that embrace this are going to shoot up. And the ones still clinging to their spreadsheets, still saying "that AI stuff isn't for us" — they're going to get left in the dust. That's not a threat, it's just the math.

So here's the question I keep turning over, and I don't think anyone knows the answer yet:

A service business will always need field workers — somebody has to do the actual work out there. That part AI doesn't touch. But the back office is a different story. So the real question is: how big can a service business get on almost no back-end admin?

How much revenue can you run, how many crews can you put in the field, with AI handling the office instead of a room full of people?

I don't know the ceiling. But I intend to find out.

1 comment

Leverage: Fast, Tireless, and Every So Often Confidently Wrong