Why Agency Founders Still Pay for Advice in an AI World

Something shifted when the models got good.

Agency founders who used to spend fifteen minutes explaining their situation before asking a question now start with the AI. Get a framework. Check it against what their advisor said. Come to the call with a sharper question.

That's healthy. That's what tools are for.

But it raises a harder question about what advisory is actually for, when the frameworks are free.

We've been thinking about this carefully, because our clients have started asking it out loud. And the answer has changed what we think we're selling.

The gap between information and judgment

There is more information available to agency founders than at any point in the last twenty years. Benchmarks, pricing models, org design templates, exit strategies, AI integration playbooks. All of it. Publicly accessible, or one prompt away.

The gap isn't information any more. It's judgment.

Specifically, the judgment to know which piece of information matters for this agency, this founder, this moment.

That judgment doesn't come from a model. It comes from someone who knows your context, and has seen enough agencies to know the difference between a problem that looks structural and one that actually is.

What the transcripts show.

We read back twenty months of coaching calls recently. What we were looking for was evidence of what was actually happening beneath the agenda.

The technical content shows up, obviously. Gross margin discussions. Propositions that have drifted from who the founder actually is now. Referral dependency that looks healthy until it isn't. These conversations are real and they move things.

But they're rarely where the breakthrough lives.

The breakthrough tends to arrive in the middle of something else. A founder mentions, almost in passing, that her three best account managers are waiting to see whether she's serious about the changes she announced last quarter. That's not a pricing problem or a proposition problem. That's a credibility gap that touches everything else on the agenda.

You can't surface that in a prompt. It shows up because someone has been in the room long enough to notice what wasn't said directly.

Three things that don't commoditise.

When we look at what actually creates value in these conversations, three things keep showing up.

First is specialist knowledge. Not general business frameworks, but what the agency model specifically requires at each stage of scale. What gross margins look like for a retained-services business at twelve people versus twenty-five. What client mix creates fragility. What hiring decisions tend to precede the problems founders notice eighteen months later. This is pattern recognition built from working inside this model, not alongside it.

Second is synthesis across problems that look separate. The reason a single hour can move from pricing to a people issue to an AI workflow to the structure of a board pack isn't inefficiency. It's because those things are connected, and pretending otherwise wastes everyone's time. An advisor who can hold the whole picture without needing to compartmentalise it is genuinely rare.

Third is continuity. Not just knowing the numbers, but knowing what those numbers mean to this founder, in this context, with this history. Knowing that the conversation about the leadership restructure has been circling for three months, and that the way the founder talked about it last week was different from the week before. That context doesn't transfer. It can't be reconstructed from notes.

What AI is good for, and what it isn't.

None of this is an argument against using AI. The founders who get the most from coaching are often the ones who use AI well between sessions, to stress-test their thinking, build first drafts, run scenarios.

The distinction worth drawing is between questions that have a right answer, and questions that have a right answer for you.

For the first kind, the model is fast, cheap, and often correct.

For the second kind, you need someone who has been paying attention long enough to know the difference between what you're asking and what you actually mean.

That kind of attention compounds over time. The model resets every session.

The question worth sitting with.

If your advisory relationships are mostly delivering frameworks, you're right to wonder whether AI can replace them.

If they're delivering something harder to name, which shows up as better decisions, fewer regrets, and a clearer sense of what you're actually trying to build, they probably can't.

The agencies growing most clearly right now aren't the ones with the best information. They're the ones with the best judgment about which information to act on.

If you're asking yourself whether you have that, it's worth a conversation. You can find out more about how we work at gyda.co.