Last week I recorded a video and gave it a deliberately blunt title: AI Will Kill Your Agency. I spent 13 years running agencies, and I do not actually believe AI kills the good ones. I do believe it is quietly coming for their margins, and most owners cannot see it yet. Karl O’Brien can. He spent eight years running a Dublin agency before building Storehero, the software that now sees inside hundreds of agencies across 15 countries, and on the Exit Ready podcast he put it in one old Jeff Bezos line: your margin is my opportunity.
The short version: AI is making it cheaper to deliver agency work, and the saving will not stay in your pocket for long.
The margin expansion you are enjoying is temporary
A lot of agencies are quietly more profitable this year, and it feels great. Karl named exactly why, and why it will not last.
“Our costs are starting to compress in order to deliver a service, but we haven’t seen the margin compression yet. A lot of agencies are experiencing margin expansion, where they’re in contract with clients and the clients haven’t reset those agency relationships and expectations. There is a short-term gain, but that’s the nature of competition.”
Your delivery costs have dropped because AI does in 40 minutes what used to take days. Your prices have not dropped, because your clients have not renegotiated yet. That gap is the margin expansion. The moment a competitor offers the same output for less, or the client works out what the work now costs to produce, the gap closes. Your margin becomes someone else’s opportunity.
Where the squeeze comes from
Two forces are pushing on the same margin. AI is lowering the cost of delivery, which invites price competition. And in some sectors the platforms are coming for the management fee directly. Karl sees it clearly in ecommerce.
“The likes of Meta and Google, to be blunt about it, are coming for that management cost. They would rather a brand invest that money directly with them than split it with an agency.”
You do not have to be an ecommerce agency for the lesson to land. When the thing you charge for gets cheaper to produce or easier to do without you, the fee attached to it is under threat. Pretending otherwise is how you get caught.
The two ways through
Karl framed the survivable positions as a fork, and I agree with both prongs.
The first is to go lower cost and higher volume on a specialised component, using AI to deliver it efficiently at scale. You accept the thinner margin and make it up in throughput, in a lane you can genuinely own.
The second is to justify your margin by being demonstrably better at something AI cannot do on its own.
“If you want to maintain that margin, you either need to make up for it in scale, or have a clear justification as to why you’re better than the next person using the same tools.”
Most good agencies should aim at the second. The question becomes simple and uncomfortable: what is your clear justification?
Own the outcome, not the deliverable
The strongest justification Karl pointed to is owning the commercial result, not the task. In ecommerce that means talking profit, not just ad performance.
“ROAS doesn’t pay the bills and doesn’t pay the tax man. We’ve seen ecommerce agencies really own that profit narrative, not only to deliver better results, but as a massive opportunity to win new business.”
This is the same way I have always sold. I am not a designer or a developer when I sit across from a client. I talk revenue, profit and commercial ambition, because that is the language the person paying actually thinks in. An agency that connects its work to the client’s margin and profit is having a different conversation from one arguing about which attribution metric to trust. One is a cost to be managed. The other is a partner to be kept. AI does not threaten that conversation, because it depends on judgement and trust, not output.
It is also why value-based pricing matters more, not less, in an AI world. When the hours collapse, pricing on the value of the outcome is the only model that holds.
The deeper play: build the IP
Karl’s whole business is the logical end of this. He saw a painful problem his agency clients could not solve, and instead of delivering the fix by hand forever, he built it into software and now sells it many times over.
“A SaaS became a delivery mechanism for a problem we wanted to solve. A lot of SaaS businesses are venture-backed agencies in the background. The future of agencies is probably some mould between SaaS and agency.”
You do not need to become a software company to use this. Build the proprietary stack, the framework, the tool that codifies how you deliver, and you own something AI cannot commoditise, because it is yours. As I said to Karl, you can build a small piece of software for a single client, get paid well for it, keep the IP, and reuse it across every client after that. That is a productised service taken one step further, into an asset.
Do not let AI hollow out your bench
One warning from the conversation worth repeating. The danger of AI goes beyond margin. Juniors learn to drop a screenshot into a chatbot and ship the generic answer, with no understanding of why. The average output rises and the actual expertise underneath it thins out. An agency with no senior development path is an agency with no justification for its margin in five years.
The agencies that win do the opposite. They take the days AI gives them back and spend them going deeper. The four-day strategy deck that now takes 40 minutes becomes four days of genuinely better thinking, which is exactly how you become the strategy-led agency worth paying for.
What to take from this
Assume your current margin is borrowed. Then decide which side of the fork you are on. If it is scale, pick the specialised lane you can dominate and use AI to deliver it cheaply. If it is value, build the clear justification now: own the client’s commercial outcome, price on that value, and start codifying your delivery into IP you own. And protect the bench that makes you better than a chatbot, because that expertise is the whole point.
Your margin is someone’s opportunity either way. The only question is whether you give them a reason not to take it.
Go deeper: Hear the full conversation with Karl on Exit Ready episode 010, then read value-based pricing for agencies and how to improve creative agency margins.