Are AI Agencies Dying? Follow the Money and You'll See Who Wins

Short answer: No, not over the next few years. When AI automates work, the value from that work doesn't vanish. It moves. A meaningful share of it moves to whoever did the automating.

I get asked some version of this every week. Is the AI agency model dead? Should I pivot? Am I too late? A few people have told me my content leaves them feeling a bit doom-pilled. I understand why.

I don't know what the next 5 or 10 years look like, and I'm sceptical of anyone who claims to. Past a certain point, AI progress is an event horizon, and nobody can see beyond it.

I can reason about the next 2–3 years, though. The most useful question to ask is simple: when a job gets automated, where does the money go?

Where does the value go when work is automated?

Say a business spends $80k a year on a process, and you automate most of it. That $80k doesn't evaporate. Part of it stays with the business as savings. Part of it goes to the software. Some of it goes to the person who designed, built and maintains the system.

I don't love describing the economy like a scoreboard, and saying it out loud feels a little cold. I'd still rather be honest about it: as automation spreads, value is being redistributed. The implementer sits right where that redistribution happens.

Why not pick a "safer" career instead?

People often suggest nursing, trades, or something physical that AI can't touch. I respect that instinct, because those jobs matter.

But "safe from automation" and "positioned to benefit from automation" are different things. One protects you from the wave. The other puts you on it. If I'm thinking about expected value, I'd rather be next to the thing that's reshaping every industry than hiding from it.

Why don't more creators say this?

Most people in this space will tell you to keep selling the same offer the same way, and that skill alone will keep you relevant. I think that's comforting, and I think it's incomplete. What protects you is being where the value is moving.

So if you're wondering whether to keep going: keep building. I genuinely can't think of much with a higher expected value right now. If you want to learn to build automations yourself, our live training takes you from zero to your first working automation.

FAQ

Will AI replace AI agencies?

Some offers will get commoditised. But implementing AI inside real businesses (understanding their workflows, data and people) remains hard to automate in the near term.

Is it too late to start an AI automation agency in 2026?

Adoption is still early in most industries: as we covered in September's AI news, no more than 10% of organisations are scaling AI agents in any single business function. The supply of people who can actually implement is far smaller than the demand.