“We use ChatGPT” is not a lie. It is just a much smaller answer than the question deserves, and it usually means AI has reached one team and stopped there.
The useful version of the question is where AI sits in the work you are actually paying for. Drafting is the visible part and the easiest to adopt. The parts that change what you get are further back: how a campaign gets analysed, how reporting is assembled, how research is done before a strategy is written, how work moves between people without being re-explained each time.
An agency that has pushed AI into those places can describe them without being prompted. They will tell you what they tried that did not work, which is usually the strongest signal in the room, because it means there were enough attempts for some to fail.
Three questions that separate the two groups:
Where has AI changed a process rather than a deliverable? Ask for one specific example with a before and after.
What do you still do by hand on purpose? A team with no answer has not thought about where judgement belongs.
What happens to our data? Which tools see client material, whether those tools train on it, and whether that is written down anywhere. This is the one buyers skip most and the one that carries real exposure, particularly if you are in a regulated sector or handling customer records.
None of this requires you to understand the tools. It requires the agency to show that AI reached the parts of their work you are actually buying.
This clip is from my conversation on the 2Stallions Marketing Expert Series.
