Agentic advertising: Ad agencies will need to give up some of their creativity myths

I’ve been in enough rooms to know this isn’t a stray provocation. It arrives after the quarterly numbers, in that pause before someone suggests coffee. The chief marketing officer grows careful. The agency side avoids eye contact. The CFO, once a courteous participant in creative conversations, has stopped pretending she doesn’t have an opinion.

The industry calls it AI anxiety. That’s flattering. Anxiety is what you feel before an exam you might still pass. What is underway is an architectural reckoning. We are arriving at it about 18 months late.

The line that aged badly: Not long ago, I was speaking on Generative AI to a room full of senior leaders. Midway through, a celebrated creative director leaned back and declared: “AI can do the dishes. I will continue to create.” The room laughed. It was a good line. It also aged rather fast. Technology doesn’t have to listen to dismissals.

What has emerged since aren’t just smarter content-making tools, but Agentic AI systems that observe, decide, optimize and iterate—continuously, often in near real time—without waiting to be asked. Artificial intelligence (AI) is not an instrument; it is now the operator. The old comfort was that the machine would always need a human to hand it a brief. That comfort is evaporating.

What’s breaking: For decades, the agency model rested on scarcity: good creative judgement was rare and agencies packaged it. A client brought a brief. The agency retreated. Weeks later, a campaign emerged, wrapped in craft and presented with theatre. Effectiveness was measured after the fact. Inefficiency was tolerated because the output was presumed to be human magic.

Agentic systems don’t just improve a step. They reject that sequence. They watch behaviour, test variants and learn what lands for which audience at which moment in the journey.

When creative generation, media optimization, audience intelligence and performance feedback sit inside the same self-improving loop, effectiveness stops being a post-campaign PowerPoint presentation. It becomes a built-in feature of the system. That is not “a better agency.” That is a new definition of marketing.

The arithmetic nobody wants to do: Then comes the arithmetic. Agentic systems are faster than agency workflow, cheaper than retainer structures and in a growing number of tasks, not obviously worse than human-led work. In some areas, they are better.

This isn’t because machines have developed taste, but because they run on what the industry long treated as supporting evidence rather than primary fuel: continuous, granular data intelligence. Human strategists spot a pattern once. Agentic systems find thousands at once and act on them before an insights deck can be formatted.

The CFO is not conducting a philosophical inquiry into the soul of creativity. She is doing arithmetic, which is getting hard to argue with.

What survives and what exits: So, is human judgement dead? No. Taste will matter. Cultural intelligence will matter more as content becomes cheap and what becomes scarce is the discernment to know what should exist at all. The ability to understand what a brand should mean in a particular society, at a particular moment, remains stubbornly human.

But the comfortable middle is under pressure: work produced slowly; elegant decks that rarely meet the market; reverence for process over outcome; the old drawbridge line—’You wouldn’t understand, it’s a creative thing.’

These are losing protective cover because no serious brand owner is happy with the idea that a ‘brand’ is a mystical element that depends on the occasional brainwave of one maverick marketer or an inspired agency flourish delivered with an obnoxious production budget. The new expectation is deconstruction: marketing broken into measurable, improving units—creative, media, audience and commercial intelligence all connected and all accountable.

Re-imagine the agency: The future agency will not survive by bolting AI tools onto an unchanged operating model. That is cosmetic surgery on a structural problem. The agency has to become something more demanding: an intelligence operating system that orchestrates multiple learning loops at once. Less a supplier of campaigns and more a platform that magnifies and coordinates distributed intelligence. The conductor, not soloist. More importantly, a platform, not performer.

This is not a smaller role. It is a harder one. It requires agencies to relinquish some comfortable myths about creativity while doubling down on what machines still struggle to replicate: taste, context, cultural reading, narrative coherence, ethical calibration. But it begins with a simple act of honesty. Stop defending a model because it has been there. Start designing one because the writing on the wall is now in legible font.

The agencies that respond with structural clarity will orchestrate an intelligence layer for their clients. The rest will continue to produce occasionally beautiful work for a steadily shrinking audience.

The author is a senior advisory professional.