AI Is for Upsizing

So much of the conversation about AI is negative. Companies are downsizing. Jobs are being cut.

Online, you'll find no shortage of videos promising to show you how to fire your entire team and replace them with AI agents.

I've built the AI team. I've handed it most of the daily work in my business.

And I have a different take.

Early in 2026, I was working late one night on my own setup, the system of agents that now runs much of my company.

A thought hit me. Once this system really started firing on all cylinders, I was going to have to hire some people to help me manage it.

That's a strange thought to have, if you listen to how most people talk about AI.

I'm apparently supposed to be thinking, "Ha! I don't need humans anymore!" That's not what I was thinking at all.

What I was thinking was this: if I had another person who understood this system even half as well as I did, the two of us could probably do three or four times more with it than I can do alone.

AI can absolutely let you do more with fewer people, and plenty of businesses will use it that way. I think there's a better use for it.

Below, I'll show you how to take the time and money AI frees up and use them to grow: by fixing the one thing that's holding your business back, by turning what you know into something customers can use, and by hiring the kind of people AI can't replace.

Hire People With Vision

By my own estimate, AI has multiplied my productivity about six times over. I'm trying not to exaggerate there.

But even that number changes how I think about hiring.

With few exceptions, I no longer think about hiring technicians, people whose whole job is doing one repeatable task. That's what everyone is worried about, and in many cases the worry is fair.

The repeatable tasks are what the robots are taking.

What I want to hire now is people with vision. People who see a little differently.

People who understand complex systems, or know how to manage them, or could learn how. Those people are harder to find, and they're more expensive.

But AI is what makes them affordable for a business like mine. My AI team can help me get past the early ceilings in a small business and build up the money to hire, without any outside investors.

It's possible to have the output of a ten-person team by yourself right now, grow the business with that leverage, and then hire.

There's a lot of talk online about the billion-dollar company run by one person and a hundred AI agents.

Maybe it's possible. It's fun to think about. But should you build a company like that? Probably not.

A business with a hundred agents and one human is fragile. When the one human gets sick, burns out, or simply runs out of ideas, the whole thing stalls.

And you lose something more important than efficiency: the people.

I don't manage people. People are for relationships. Systems are for management.

I manage systems, and then to some degree, the system manages the people. That idea is older than my AI team, and it turns out to be the best argument I know for hiring humans in an AI business.

If systems handle the management, then people are free to do what people do best: build relationships with customers, notice what the data doesn't show, come up with the next idea, and care about the work.

Robots are great at the reminder email about a late payment. A person is who your best customer wants to talk to when something goes wrong.

So reach for the rewards of AI, and guard against the risks. Look for ways to make other humans part of what you're doing instead of pushing them out, and you'll be a positive contributor to the people around you.

I believe this could be one of the best changes the world has seen. It could get more people living into their full potential and using creative thought to change the world instead of working for the man.

Every Business Has a Bottleneck

If you're going to grow, you have to know where to put the time and money AI gives back. For that, I use the idea of the bottleneck.

Jim Schleckser, in his book Great CEOs Are Lazy, describes it with a picture I love: a garden hose. When water barely dribbles out the end of a hose, the problem is almost never the whole hose.

It's one kink.

Schleckser writes that a great CEO "will notice the dribble of water, find the kink in the hose, untangle it, promptly hand the hose back to someone in the organization, make sure the system won't allow another kink at that place, and eventually head back to their lounge chair."

He borrows the idea from Eliyahu Goldratt, whose book The Goal taught a generation of business owners to think in terms of constraints.

Every business has one point that limits how much the whole thing can produce. Improving anything else barely helps.

If your sales team brings in more leads than your crew can install, more leads just make a longer waiting list.

If your crew could do twice the jobs but the phone isn't ringing, a faster crew doesn't make you a dime.

Writer and entrepreneur Dickie Bush calls this "bottleneck thinking," and he credits Goldratt as its source. I like it because it's flexible and simple: run the numbers, find the bottleneck, fix it.

Then do it again, because once you fix one, something else becomes the slowest part.

Schleckser sums up the whole game plan in one sentence: "Find the point of constraint that has the highest economic impact, engage in fixing the constraint in a high-leverage way that builds organizational ability, and repeat."

Notice the words "builds organizational ability." Schleckser points out that a hardworking CEO, once he finds the kink, will dive into the problem up to his elbows and fix it himself.

The better approach is to fix it in a way that leaves the business stronger, so the same kink doesn't come back the moment you walk away.

Find It, Fix It, Find the Next One

Here's how I apply all of this in a business with an AI team. I call it Find It, Fix It, Find the Next One.

Find it. You can't fix a bottleneck you can't see.

In 2025, I walked a coaching client through a simple spreadsheet I use for tracking marketing funnels.

Each stage of the funnel gets a target number, called a key performance indicator (KPI), and each number is color-coded red, yellow, or green depending on how it's doing.

Red cells jump off the page, and that's where you look first. The rule I gave was to focus on one channel at a time and fix its bottleneck before moving on to anything else.

If you don't have numbers yet, Schleckser offers a great question to start with: "If I came upon a genie and he granted me a single wish to make my business better, what would I change?"

Whatever you'd wish for is probably a good place to start looking for your constraint.

Fix it, with AI or a person. This is where the new tools change the game.

Once you know your bottleneck, ask whether it's the kind of problem AI can fix, or the kind that needs a human.

• If the bottleneck is follow-up (estimates that never get a second call, leads that go cold), that's a job an AI agent can own today.
• If it's capacity on repeatable work, like reports, scheduling, or content production, AI can probably take most of it.
• If it's judgment, relationships, or vision (the big accounts need a seller, the crew needs a leader, the strategy needs an owner), that's usually a person. And the money AI saved you on everything else is how you afford them.

Find the next one. When you fix the kink, water flows, and then the next kink shows up somewhere further down the hose.

That's good news. It means the business grew. So go find the next one.

Be ready for an uncomfortable discovery, though: the bottleneck is often you. When we set our company vision for 2025, one of the questions we put on the table was whether I was personally bottlenecking project delivery.

One of our big goals that year was to hire an integrator, someone to install the systems and manage operations so the work didn't have to pass through me.

I've now got AI agents doing much of that job. But the question is the same one every owner has to ask.

Two books I've read put this well. Jason Fried and David Heinemeier Hansson, in It Doesn't Have to Be Crazy at Work, include "not being a bottleneck" in their description of a great work ethic.

Michael Bungay Stanier comes at the same problem from another angle in The Coaching Habit.

His premise is that the best way to lead people is to coach them, and coaching mostly means saying less and asking good questions, so people work out the answer for themselves.

He points out a trap most owners know well: "The more you help your people, the more they seem to need your help."

Every time you jump in and solve the problem yourself, you teach your team to bring the next one back to you.

Building a coaching habit breaks that cycle. In his words, it helps your team "be more self-sufficient by increasing their autonomy and sense of mastery, and by reducing your need to jump in, take over, and become the bottleneck."

The same trap shows up with AI agents. When an agent gets something wrong, the quickest move is to grab the task and do it yourself.

It feels efficient, but the agent learns nothing, and the task comes back to you next time.

The coaching move is slower the first time and faster every time after: ask what the agent was missing, then fix the instructions, the access, or the context so it can handle the task on its own.

One of Bungay Stanier's questions works almost word for word with an agent: "How can I help?" Usually the answer is a missing piece of information, a missing connection, or a rule that never got written down.

So start by getting yourself out of the hose. Once you're out of it, you can see where the kinks are.

Productize What You Know

There's one more way AI helps you grow, and business owners rarely see it coming. You can turn what you know into a product.

Early in 2026, I put together a new offer around one of my courses. Along with the course, it includes an AI coach trained on all of my course materials.

That AI coach took me about an hour to spin up, and it works insanely well.

A student can bring it a question about their own business, and it answers from what I've actually taught instead of from the average of the internet.

Think about what that means for a business owner. A course on its own often leaves people alone with the material.

A coach that's available around the clock changes that. And me answering every question personally would never scale.

You don't have to sell courses to use this idea. Almost every business has methods its customers would pay to get at more easily:

• A contractor could give customers an AI assistant that answers questions about their project, their warranty, and how to care for the work, using the company's own documents.
• An accountant could offer clients an AI helper trained on the firm's checklists, explaining which documents to gather and why.
• A coach or consultant could give clients an AI version of their frameworks to use between sessions.
• A shop or service business could put its product knowledge, sizing guides, and care instructions into an assistant on its website.

Here's the simple way to build one:

1. Gather your materials. Put everything you've written or recorded about how you do the work into one place: trainings, SOPs, scripts, FAQs, price sheets, and the best answers you've given customers over the years.
2. Write its job description. Give it a purpose, what it may do without asking, what it brings to a human, and what it never does. Add one daily responsibility: a short report on what customers asked it.
3. Test it with your customers' questions. Ask it the twenty questions your customers ask most. Fix the materials wherever it answers badly.
4. Watch what people ask. The questions your customers bring to the AI coach are a gold mine. They'll show you what to teach next, what's confusing on your website, and what new product people would pay for.

The same rules apply here as with any AI that talks to your customers. Disclose that it's an AI.

Build verification into it, and give it your actual prices and policies so it doesn't guess.

And give it a written list of what it hands off to a human: a refund, a complaint, a question it can't answer from your material.

Follow those rules and get started. Your methods are already written down in your trainings, your proposals, your emails, and your head.

An AI coach just puts them where your customers can reach them.

Try this: Name your bottleneck and your next hire

• Make the wish. Write down your answer to Schleckser's genie question: If I could change one thing about my business, what would it be?
• Run the numbers. Pick your main way of getting and serving customers. List each stage (leads, calls, proposals, sales, delivery, follow-up) with one number for each. Mark each one red, yellow, or green.
• Name your biggest bottleneck. One sentence: "Right now, the thing that most limits our growth is ______."
• Ask whether it's you. Does that bottleneck run through your desk? If so, which part of it could your AI team take this month?
• Decide: AI or a person. If it's follow-up, capacity, or repeatable work, hand it to your AI team to design a fix. If it's judgment, relationships, or vision, it probably needs a human.
• Name the one human role. Write down the one person you would hire if AI freed up their salary. Describe what they'd own, and why a person needs to own it. Look for vision, systems thinking, or relationships.
• Do the math. Add up the money AI saves you and the hours it gives back each week. How far does that get you toward that salary?
• Pick one thing to productize. Choose one set of methods your customers ask about all the time. Gather the materials into one folder: your frameworks, scripts, FAQs, and price sheet.
• Set a date to look again. In 30 days, check the numbers. If the kink is gone, find the next one.