Agencies often see each other as competitors, but what happens when the real competition becomes the very tools powering their work?
The uptake in AI is gaining rapid momentum as 64% of marketers use generative AI weekly (Salesforce State of Marketing Report 2024) and 54% of CMOs are prepared to increase spend on AI-driven tools in the next year (Dentsu CMO Navigator, Q1 2025).
ChatGPT, Claude, Google Gemini, and the like may build time-saving practices and increase efficiency, but they can also deceive clients into believing that strategy, creativity, and brand-building can be automated.
AI is a tool, not a team, and there are many instances where agencies must step in—not just to defend their value but also to educate the client on what AI can't do.
Five Inconvenient Truths About AI That Clients Can't Ignore
1. AI Is Void of Strategy
AI may craft your email copy, generate limitless versions of a subject line, and spout keywords, but it can't comprehend the agency objectives that are prescriptive to each client. Only an agency understands "the why''; AI doesn't know the target market, customer relationship, previous workings or business goals.
Ultimately, an AI strategy that cannot go beyond assembling facts will not cut it. Innovation, making tradeoffs, and knowing what not to do is very much the job of humans.
2. A (Potentially) Dangerous Lack of Context
Fully 73% of consumers expect brands to take a stand on societal issues (Edelman's Trust Barometer, 2024). Even though AI is trained on large language models, it lacks not only emotional intelligence but also cultural literacy. That can easily lead to ill-informed and clumsy AI-generated ad and press release copy.
An often-silent role of an agency is to have its finger on the pulse of generational cultural movements, world events, and the subtext of what makes the likes of millennials and Gen Z tick. Finding what resonates is no job for algorithms.
3. Yes, No, Maybe? Sorry… AI Needs Clear Instructions
The notion that AI can fully replace the briefing process is a myth. As any account manager will testify, the draft of a brief is often very different from the final product. Clients often figure out what they need throughout the briefing process rather than knowing it at the very beginning.
AI is unable to read between the lines, ask questions, or consider internal client debates.
Agencies aren't just executors—they're interpreters.
4. AI Outputs Require Expert Oversight, Editing, and Human Judgment
AI cannot exercise editorial judgment, understand legal risks, or grasp the nuances of brand messaging. Although its output may be grammatically correct, it may not be legally safe, consistent with your brand, or emotionally engaging.
Someone needs to review, refine, and take responsibility for the content. That someone should be your agency.
5. AI Is and Will Always Be a Tool
No matter how powerful AI is, it's a tool. It may be a powerhouse, but it requires human operators with not only expertise but also experience.
It can accelerate workflow, but not deliver it from beginning to end.
Five Ways Agencies Can Reframe the Conversation
Instead of viewing AI as a threat, agencies should reframe the conversation to how it can enhance their value, not replace it.
AI can assist with execution and automate tasks to make things simpler, but it's what humans can bring—such as strategic thinking, cultural insight, and emotional nuance—that makes it a tool, rather than a competitor; agencies should shift the conversation this way to reinforce their relevance and strengthen client trust.
1. Your opinion is your currency; show your thinking, not just your deliverables
It's more important than ever for agencies to be able to articulate the 'why' behind their thoughts and ideas: show how you got to the end result, not just the result itself.
AI can generate endless creative output at the click of a button, but it doesn't have a point of view or original thoughts; agencies, though, have the insight, opinion, and ability to connect the dots that the client needs to see.
Show your thoughts behind the scenes to let clients know that they're not just getting content, they're getting experience and creativity that can't be replicated by an algorithm.
2. Don't be afraid to show how you use AI, and educate clients on your AI stack
Transparency is important: You don't need to keep your AI use a secret.
Some clients may be anxious about AI, but by demonstrating how you're using AI tools to do work, such as analyzing data and speeding up processes, you can build trust and reassure them while showing that your human insight is still required for the quality they're used to.
3. Reframe AI as a partner, not a producer
Agencies need to guide their clients to a realistic understanding of the role of AI, to ensure they know it can't be a substitute for the agency; it may look like a lower-cost solution, so agencies need to be clear that it supports their thinking, rather than replacing it.
The experience, creativity, judgment, and instinct of the agency is the difference between creating content that lands and content that totally misfires.
4. AI is not personally responsible: Highlight accountability
If content misfires, a campaign goes wrong, or even when a legal issue arises, AI doesn't take accountability—the agency does.
Highlight the importance of accountability and responsibility to your clients, stressing that AI can generate content, but it can't understand legal or ethical implications of what it's generating.
An agency doesn't just create ideas; it protects clients, their brand, and their credibility—and that kind of safety net is something that AI can't generate.
5. Create an AI policy to show that you use AI responsibly
An agency can build trust with clients by creating a policy that shows how you're using AI, how you ensure human oversight, and that you have thought about what AI should do, not just what it can do.
The policy should explain how AI tools fit into your processes, to demonstrate that you're applying them with control and consideration, and that you're using them because you know why they should be used and you know how to use them well.
