How should an agency answer client questions about AI?

Agree one honest answer before clients ask: which tools you use, what data goes in, who checks the work and what you will not do. Put it on a page.

Why the answer needs agreeing in advance

Clients now ask about AI in pitches, procurement forms and ordinary catch-up calls. If the agency has not agreed its position, the answer depends on who picks up the question. An account manager says one thing, a creative says another, and the client is left less confident than before they asked.

A clear, consistent answer is a sign of a well-run agency. It can help you win work.

What clients usually ask

  • Do you use AI on our work? And if so, for which parts.

  • Which tools? Whether they are business versions with proper terms or personal accounts.

  • Does our information go into them? And whether it could be used to train a model.

  • Who checks the output? Whether a person reviews everything before it reaches them.

  • Who owns what you produce? A question with real legal detail, covered in who owns the copyright in AI-generated work in the UK.

  • Will you tell us when AI is used? See whether agencies should tell clients when they use AI.

  • Does this cover your freelancers? Clients know agencies rely on them.

Put your position on one page

Write a short statement the whole team can use. It should cover:

  • Where you use AI: For example research, first drafts and summarising, and where you do not.

  • Approved tools: The tools the agency has checked and pays for.

  • Data rules: What client information may and may not be entered.

  • Human review: That a named person is responsible for everything delivered.

  • Disclosure: When and how you tell clients AI was involved.

  • Freelancers and suppliers: That the same rules apply to them.

This statement should match your internal AI acceptable use policy. If the two differ, fix the policy first.

Be honest about what is still being worked out

Do not claim more than is true. If some of the team are still experimenting, or you have not finished checking a tool, say what you have in place today and what you are doing next. Clients respond better to a straight account than to a polished one that falls apart at the second question.

Check what your contracts say

Some client contracts and procurement terms already restrict AI use or require consent. Read them before you answer, and make sure account teams know which clients have specific terms. Take legal advice where a contract is unclear.

Raise it before they do

The strongest position is to bring the subject up yourself. Share your one-page statement at the start of a relationship or at the next review. It turns a defensive conversation into evidence that you have thought the matter through.

If your agency does not yet have an agreed position, that is the first thing to resolve. Go Wisely helps agencies get there, as set out on the AI for Agencies page.

Want to talk this through?