AI didn't kill your business. You did.

Every agency owner blaming AI for their shrinking margins is looking in the wrong place.

"I'm paying more on ads!"

"It's made it so that content creation is commoditised."

"My clients are all telling me that ai is capable of replacing me!"

Look, I get it... I’m jaded as fuck over the blistering speed that ai has changed the agency space.

But here's what I actually think:

AI didn't break your business. It only exposed the actual value your business provides.

This newsletter is going to outline how stop your self getting replaced by AI in a matter of months.

The Moat You Thought You Had 

The old way of doing business in services was simple. Know more than the customer.

Take SEO. You knew how to do it. They didn't. Every client you worked with widened that gap between what you knew and what they knew. That gap was the moat.

Then AI came in. At first, it was cute.

Now someone can run an SEO audit on something like Claude or Co-Work, or use a tool like Open.Claw or Claude Code and let it operate as an SEO itself. What used to be the moat, the knowledge gap, has completely collapsed.

It exposed the fragility of the service-based business model.

The New State of Affairs

We're living in a new reality where it's not about knowing more.

It's about owning the means of distribution. Owning the right product or service. And owning niche-specific IP that others simply won't have.

If you're still delivering off the fact that you know something your clients don't, "hey, we know how to build you a linkedin funnel," you're on a shrinking island. As the LLM models get better, that gap keeps narrowing. Margins keep shrinking.

Knowing things is no longer an advantage.

The alternative? How do you turn that knowledge into a system, package it, and make it into assets your client can actually own, bolstered by the real-world experience you've built up working with people like them?

Proprietary Data Is the New Moat

Let me use myself as an example.

Lead generation is completely commoditised. Setting up the right positioning on a profile, writing the right posts, there are 10,000 people out there doing exactly the same thing.

So why would somebody pay a premium to work with me over someone else?

One thing: proprietary data.

What evidence do I have that I've done the thing my client wants, and how many times have I done it?

What stats and patterns have I taken from previous engagements that I can bring into a new one?

That means the systems I build will get them closer to the finish line faster than starting from scratch.

With all tools being equal, the only thing that differentiates you is your previous experience.

That's the true moat. Proprietary data and how you package your offer.

What AI Can't Replicate (Yet) 

Anyone can write an outreach sequence. Any LLM can sketch a generic workflow.

What it can't do is this: if I've run 100+ LinkedIn campaigns with specific messaging, AI can't tell you how that messaging performed across all 100, what patterns worked, what got burned out in four weeks, which segments ghosted after promising CTAs, or which cadences actually created pipeline versus wasted everyone's time.

You'd only know that by actually doing it.

"We've run this specific campaign for 100 people like you. Here's how it usually turns out."

That's the insight. And AI can't prompt its way into it, not without the time and experience to collect the data in the first place.

AI can build the LinkedIn outreach flow. What it can't tell you is which messages burned out an audience in four weeks, which ones had the highest likelihood of doing that, which segments ghosted after the CTA, or which cadences actually created a pipeline.

That's three years of specific results across dozens of clients.

Where This Is Actually Going

This is why I think agencies are becoming something closer to systems and software businesses.

They'll build the infrastructure. Or they'll plug clients into their infrastructure, with all the proprietary data and understanding baked in. Then the top-level strategy runs across every function of the business.

The question to ask yourself right now:

What's the best client you could clone? If they could take just one thing from working with you that helped them most, what would it be? Would it be how you see the market? Your hands-on execution?

And how much of that is actually documented in a system that's theirs versus stuck in your head or buried in your team's Slack?

If a client fired you tomorrow and kept everything you'd built for them, what would continue to compound without you there?

The Packaging Problem

People can now get to the how very easily.

Premium clients are paying for speed. 

They're paying for your packaged playbooks, in-house assets, dashboards, scorecards, and how those get enabled and used by the people inside their business.

So it comes down to a couple of things:

  • How are you packaging your agency services so you're not just a pair of hands? You're selling systems and repeatable workflows based on what you've actually seen work.

  • Do you have actual decision rules for when things go wrong (and they will go wrong) so clients know how to get back on track?

If I Were Starting Out Now 

I'd be focused on building some kind of top-of-funnel tool aimed specifically at getting data from clients, then feeding that back into building models I could use and sell.

More data = more advantage in the market.

This isn't about doing stuff for the sake of it because we can now build with Claude Code.

It's about reporting on what's actually happening in the market, based on what you've seen, what's worked, and what's actually gotten people results.

That's what always serves customers at the highest level.

The New Game

Whilst AI shrank the information gap, the next five years (until we all get replaced by AGI) will probably be defined by who turns their experience into systems, assets, and insight that sit above what those models can do on their own.

If all you sell is "I know things you don't," you're competing in an ever-shrinking window with everyone else screaming "they took our jerbbss" (screaming hill billy from southpark voice).

If you sell "I've turned what I know into something you can own and compound," you're on completely different ground.

So if you’re reading this and quietly thinking:

“Okay, but what do I actually do with AI to grow this thing in 2026… without becoming replaceable?”

That’s exactly what I broke down in a new YouTube deep-dive: how to use AI to grow your agency this year in a way that compounds your moat instead of erasing it.

In it, I walk through:

  • The biggest mistakes agency owners make when trying to scale their business with AI (these can literally destroy your business)

  • How to plug AI into your existing systems so it gives you leverage to grow your business.

  • Where to use AI tools like Claude, ChatGPT, and LinkedIn AI automation for unfair speed (without compromising your brand)

If you want to see how this looks in practice, go watch the video and treat it as a working session...

Because AI didn’t kill your advantage yet.

But if you don’t start building the systems and IP that it can’t copy, it eventually will.