Should your agency tell clients when you use AI?

A version of this conversation is now happening inside most agencies. A client asks — sometimes directly, sometimes obliquely — whether the work you handed over was made with AI. And the room goes quiet for a beat while everyone decides how honest to be.

That pause is the real problem. Not the AI. The pause.

The quiet strategy, and why it ages badly

The default instinct is understandable: don't raise it, and hope it doesn't come up. If a client isn't asking, why introduce a question that might unsettle them?

The trouble is that “don't ask, don't tell” only holds until the first time it's tested — and it will be tested. A client hears about AI in their own board meeting. Someone on their side notices a tell. A competitor makes disclosure part of their pitch. The day that happens, an agency with no stated position is caught flat-footed, and a practical question — “how do you use AI?” — quietly becomes a trust question: “what else haven't you told us?”

You've converted a routine disclosure into a credibility problem, at the worst possible moment, on someone else's timing. Silence feels safe because its cost is invisible. It isn't zero. It's just deferred.

The reframe: disclosure is a trust asset, not a liability

Here's what agencies underrate: handled well, telling clients how you use AI is one of the easiest trust wins available to you.

Clients aren't actually afraid of AI. They're afraid of being quietly shortchanged — paying senior-strategist rates for something a tool produced in ten seconds, or having their brand voice handed to a model with no judgment applied. What reassures them isn't the absence of AI. It's evidence that a capable human is deciding where it's used, checking what it produces, and standing behind the result.

A confident, specific answer — “we use AI to move faster on X, we never use it for Y, and everything is reviewed by the people you hired” — does far more for a relationship than pretending the last two years didn't happen. It signals the exact thing the client is paying for: judgment.

How to disclose without making it awkward

Disclosure goes wrong when it's reactive, vague, or apologetic. It goes right when it's proactive, specific, and matter-of-fact. Three moves:

1. 1

   Raise it first. Put a short, plain line about your AI approach into how you contract and onboard — before anyone has to ask. Getting there first turns a potential ambush into a signal of confidence.

2. 2

   Be specific about the line. Name where AI helps (speed, research breadth, first drafts) and where it doesn't (final judgment, strategy, anything you'd stake the relationship on). A clear boundary reassures far more than a vague reassurance.

3. 3

   Anchor it to outcomes, not tools. Clients don't care which model you use. They care that the work is good, the thinking is yours, and their trust is safe. Frame disclosure around what stays true regardless of the tools.

Done this way, disclosure stops being a confession and becomes part of your positioning — a reason to choose you over the agency still hoping nobody asks.

The bigger picture

Whether and how you disclose AI use isn't really a communications question. It's a governance question — and governance is one of the clearest signals of how mature an agency's AI practice actually is. Agencies that have thought disclosure through have usually thought other things through too: where AI touches billable work, whether it's changed their margins, whether their use is consistent or improvised. Agencies that haven't, usually haven't.

Which is the genuinely useful thing to know about your own agency: where do you actually sit?