I could tell the same story about chefs and the microwave, graphic designers and desktop publishing, travel agents and online booking, and any tool or technology that promises to democratize specialized jobs.

Notice what actually happened in that story, though: every time someone was able to do something themselves, a new problem opened up right behind the old one. The business was never really about horses or cars. It was about standing a few steps ahead of the customer.

The gap is the business

Design agencies are in the same bind right now.

There was a magical moment when websites were both very important to have and very hard to make. That’s the sweet spot for a service business. Demand was high and supply was low, so agencies who could build websites had it good.

Then came the tools: FrontPage, Dreamweaver, WordPress, Squarespace, Webflow, Framer. Every one of them promised some version of “now you can build a website yourself.”

The agencies that fought that message are gone. The ones that leaned in noticed something: these tools made building a website easier, but they didn’t make it easy. So they worked with the messaging instead of against it: “Now you can build a website yourself. And we can help you do that.” That’s where WordPress agencies and Webflow agencies and Framer agencies came from. None of them would exist if they’d fought the current instead of swimming with it.

What those agencies understood is that there are a bunch of steps between a regular person and a person who can build a website. The more steps away someone is, the more it’s worth hiring help. Every new tool moves them a step or two closer. And the closer they get, the less your help is worth.

So here’s the bet at the center of every long-lasting service business:

How big of a gap can you open between yourself and everyone else, and how long can you keep it open?

Measure the gap in years

How long would it take for your clients to do what you do? If you stopped today and they started trying, how long before they wouldn’t need you?

Three years is the number I’d aim for.

Three years is roughly what it takes to establish a real niche. You’ll spend that long rebuilding your offers, your pitch, your team, and your reputation around the new thing. So a gap your client could close in six months isn’t worth going all in on. You’d build a business that had no market.

A gap that takes them three years to close is a different animal. That’s long enough to plant a flag, become known for it, and get paid well the whole time.

Run the same test on your competition. How long would it take the next agency your client would call to do what you do? Being three years ahead of your client doesn’t help much if six other shops can deliver the same results you can.

When you find something where both answers are three years, double down. Build your reputation there. Only take this kind of work so you can turn three years into five.

Many agency owners have never done this exercise.

When I ran my agency SuperFriendly, we did a lot of things for clients: web design, app design, design systems, CMS builds, branding, video, and more. I didn’t have a vision for how we could be three years ahead of everyone in web design or CMS builds, but I did for design systems. We had more unfair advantages there than anywhere else, and the further we went, the more our clients confirmed it. They were constantly surprised by how much we brought that they couldn’t do themselves, and they weren’t seeing anything close from our competitors. That’s why we picked that niche. Three years before I shut the agency down, if you’d asked me how long it would take one of those clients to do design systems work as well as we did, I’d have said at least three years without blinking.

What AI just did to the gap

Now we have AI.

The scary part for agencies is how many steps it closes at once. A regular person is much closer to “I can build my own website” than they were two years ago. Maybe they’re already there. Either way, they’re close enough that hiring someone to do it isn’t worth much.

What’s even scarier is that I suggested finding a three-year gap, and almost every agency owner I talk to can barely forecast three months out.

That’s new. For most of my career, you could squint and make out the shape of the next few years. Now an AI model ships on a Tuesday and can wipe out a whole branch of business. The future is foggier than it’s ever been.

But I’d rather guess at three years and be wrong than plan for three months and be right. A three-month plan is just reacting. If you can’t picture three years out for what you sell today, taking action now will get you further ahead than waiting it out.

If I had to suggest a place to start, I’d start with your expertise.

We trust people who have done something for a long time to know the difference between a good version and a bad one. We trust that a professional chef knows good scrambled eggs from bad ones. We trust that a pro basketball player knows a good jump shot from a bad one. More important, we trust that all those reps taught them how to produce the good version on purpose.

Someone who has been making websites for fifteen years knows what makes a website good. Anyone can make a website with AI now. But the site an experienced designer makes with AI will still beat the site a regular person makes with AI. Everyone can drive a car today, but back when most people couldn’t drive well, driving for them was a lucrative service. If I ran a web design agency right now, I’d test selling websites made with AI.

And then keep going. Any time someone can finish a task on their own, it unlocks the next task they need help with. Clients can make their own websites now, which means they now have websites to maintain and update. So I’d also test selling maintenance and updates for sites made with AI.

Once they can handle that themselves, the problem after that opens up. Then the one after that. Solving the next problem before your client gets there will keep you in business.

Trigger it yourself

One day, the hard, specialized thing you do will be democratized. It’s inevitable. So don’t spend your energy avoiding it. Anticipate it.

The hard part with AI is speed. It might take you longer to build the next service than it takes your client to arrive there without you. That’s always been the real game of business: putting off your own obsolescence as long as you can. The best way I’ve found to do that isn’t to just anticipate it. It’s to trigger the obsolescence yourself, and to keep triggering it.

Here’s what I mean. Choose one thing you sell. How long before a client could do that themselves or before every other agency offers it? If the answer is short, don’t defend it. Attack it. Find the way to make that work faster, cheaper, and easier for yourself even though that’s exactly what makes you replaceable. You’re going to get replaced on that thing either way. It’s better for it to be on your timeline.

Doing it yourself buys you something your competitors don’t get: you find the gaps earlier.

If you haven’t been doing this, start today. Write down everything you sell. Put an honest number of years or months next to each one. Whatever has the smallest number is the thing you need to end first, on your terms.

Most agencies do the opposite. They hold onto the thing they’re good at for as long as it works—which makes sense—but then they hold on well past that—which is agony. Hand-crafted websites are going the way of graphic design paste-ups. Selling them is only going to get harder.

Instead of violently holding on, put the same amount of effort into letting go.