Everything leaving Digital Goliath now has AI in it somewhere. Monthly reports, campaign recommendations, account reviews, proposals…

AI drafts it, the team shapes it, and Jeremy signs off on it.

It makes sense. The reports are better for it. The recommendations are more thorough. It’s a super net positive.

Here’s something that’s starting to happen more.

When a client or prospect gets the deliverable, they’re now running it through ChatGPT and coming back questioning the validity of the doc. Presenting me reasons why the recommended approach could be even better if we did it “this way”.

I guess if the positions were switched, I’d probably do the same. I can’t and shouldn’t get offended by that. If my accountant handed me a strategy document and I had a tool that could pressure-test it in five seconds, I’d use it. No offence to my accountant.

But now I’m on the receiving end of that, and realistically there are two ways I can handle it.

I can fight it. Spend a few hours building a proper response, addressing every objection, explaining where the AI missed context on this specific account.

But the client might read that as me being defensive. Or they’re close enough to a keyboard to type one follow-up prompt, and within seconds their AI doubles down — more forceful this time, better sourced, having scoured the internet for anything that proves their point. All to please the master.

The other option is to take the easier road. Use my AI to meet theirs where it wants to land, agree where it needs to agree, and move on.

Neither sits right.

Let me play devil’s advocate on myself for a second.

17,000+ hours in online ads is real. But I can’t dismiss the counter-argument.

The models being used to scrutinise my work have ingested millions of hours worth of campaigns, post-mortems, and perspectives from practitioners across every market and budget level imaginable. More than any single person could accumulate in a lifetime.

So who actually knows more?

Here’s the distinction I keep coming back to. The leg I have to stand on.

The AI has processed every outcome. What it hasn’t done is apply judgment in situations where the data pointed one way and experience said another — and experience turned out to be right.

Knowing which recommendation fits this business, in this market, at this stage of growth, with this specific offer — that’s not in the training data. That’s what a specialist is actually for. Isn’t it?

So when a client runs our report through their AI and it comes back with objections, the question isn’t really whether the AI is right or wrong.

The real question is whether 17,000 hours of experience still counts for something.

You tell me. Because whether you’re the client running reports through ChatGPT, or the consultant watching your recommendation get picked apart — someone in that loop is losing something.

And it might not be obvious who. If you’d rather work with someone who’s seen this situation before and knows how to navigate it, let’s talk.