70-90% profit margins. That's what AI-native agencies are producing right now, and it isn't a rounding error on a traditional model. It's a different model.
This week I sat down with Jordan Doughty on his podcast and broke down exactly how. We covered four things that matter more than any single tool you could buy.
- To become an AI-native company, you first have to become a data-first one. Before agents, before automation, your processes and client data have to be organised in a way AI can actually use. Skip this and the agent has nothing to think with.
- Why agencies get stuck between $1M and $5M. It's almost always structural, not a sales problem. The founder is still the delivery model, and you can't out-sell that.
- Stop thinking of AI as something you bolt on top of existing processes. The leverage isn't in making the old workflow 30% faster. It's in rebuilding the workflow around what AI can do.
- Hiring. View yourself as a recruitment agency. Character matters more than technical skill. And we use AI to check our own biases when evaluating candidates.
Then Jordan asked me about my backstory, and I went somewhere I don't normally go.
The part I don't usually talk about
Factory floors in south England at 18. English as my fourth language. Sharing a house where one of my housemates told me he wanted to end his life. Sitting on a bus at 5am full of migrants heading to a shift, thinking, what am I even doing here?
I talked about my dyslexic tendencies. How I get distracted by a hundred ideas a day. How I used to think something was wrong with me because I couldn't stick to one thing.
The thing I thought was broken in me turned out to be the thing I build with now.
What would an AI-native operating model change in your agency first?
The Founder Dependency Audit diagnoses where you're still the delivery model and where a data-first, AI-native structure would change your margins first. We work through it together.
See where your agency depends on you- Romans