One of your clients asks you to confirm in writing that no AI is being used on their account. Another one is fine with AI but now asks for twice as much work…for the same fee. You suspect the team is using their own AI to stay on top of the increased volume of client demands. The junior account manager asks, half joking, whether they are training their replacement, referring to the new AI tool you rolled out.

Each of these questions lands on a different day. Each gets an answer, usually a sensible one, from whoever happened to be asked.

Add those answers up and you arrive at your AI policy. The problem is you did not write it. The immediate clock did.

There are two clocks for everyone. The immediate clock is the work in front of you. For an agency, it is answering the AI question client by client, day by day. Every agency runs this one.

The structural clock is what the agency needs to become: its value proposition, and how it charges when hours no longer hold the value. Few agencies run this clock.

This is a rational response to incentive structures. The immediate clock is where the account risk is visible. It is also self-defeating. Not repositioning the offer is what escalates the risk. Each AI disclosure ends up being a fee negotiation.

“Do you use AI?” is an old worry with a new name

I have been the agency pitching the work and the client buying it.

The unspoken arrangement is this: you see the most senior agency people at the pitch. Your account is usually run by a more junior team. This arrangement works as long as the junior team delivers good work and senior involvement is there when the work calls for it.

AI is now part of that account team. Clients want to know what it does, how it does it and when seniors get involved. It needs mindful introduction to the client, just like any new account member, and it will be accepted as far as it proves its value and credibility over time.

Change management is critical to account health: explaining AI’s place in how the work gets done now, the guardrails around its use and how any time reclaimed gets reallocated.

It does not eliminate the trust penalty of AI disclosure, the new bias where work is valued less once AI use is revealed, but it allows you to reduce it and pay it on your own terms.

Agencies face different risks when it comes to AI

At the agency, the AI spend and how the team allocates their time are watched closely to manage costs, for good reason. This creates three challenges: little room to experiment with AI, increased volume of AI-assisted work that needs to be overseen by humans whose bandwidth did not increase equally, and people resorting to unsanctioned AI, also known as shadow AI, to get work done.

Shadow AI is the by-product of the mismatch between what people actually need and what they have. That gap is how an embargoed announcement ends up pasted into a personal ChatGPT account when a client brings the deadline forward. It isn’t necessarily a budget problem. It’s a question of whether people have the right tools for the work they need to do.

An agency’s AI risk exposure is also a different shape from its clients’. It holds confidential material for dozens of organisations inside one set of AI systems. Keeping one client’s confidential material out of a competitor’s draft or keeping work going when you max out your AI budget or lose access to a tool, are real risks to manage.

Where to start this week

These are not a pile of separate challenges. Underneath them all, there is a single question: what are clients willing to pay an agency for when AI does the first pass? Below is how you can break this big question down and start tackling it.

1. What’s our position on AI, and how do we frame it across our portfolio where one answer does not fit all?
2. Which of our services still hold value, which ones need to evolve?
3. What problems does AI create for clients? This is where new client work emerges.
4. How can we create a safe environment for the team to experiment with AI?
5. What are the risks AI brings and how can we prepare for them?

These questions need answering on purpose, before a client or their procurement team answers them for you.

If you’d like support in thinking any of this through, email me at elif@thetwoclocks.com.

Elif Güvençer is a strategic advisor and founder of The Two Clocks. She works with agency and in-house communications leaders on AI strategy and governance: what AI changes, and often breaks, in the work, and how they should reposition before someone else’s logic decides it for them.