The Brands Winning With AI Aren't Using It to Replace Expertise — They're Using It to Amplify It
There's a version of AI adoption happening at a lot of brands right now that looks something like this: a leadership team gets nervous about falling behind, someone buys a suite of tools, a few people get trained and suddenly "AI-first" shows up in the strategy deck. Boxes get checked, but not much changes.
Then there's a different version. A marketing team identifies a specific, expensive problem — creative testing takes too long, personalization at scale is impossible, campaign variants require too many production hours and uses AI to solve that problem faster than they could before. The thinking stays human. The execution accelerates.
These two approaches produce completely different outcomes, and the gap between them is widening. Brands in the first group are spending money on tools. Brands in the second group are building real competitive advantages. The difference isn't access to technology, it’s the expertise and strategic clarity behind how it gets used.
The Pressure to Adopt AI for Its Own Sake
The "AI-first" imperative is real, but it's also producing a lot of noise that's easy to mistake for signal.
Why "We Need to Be Using AI" Is the Wrong Starting Point
When AI adoption starts with the technology rather than the problem, the results tend to be mediocre at best. Teams end up using generative tools to produce content faster without asking whether that content is any better (or more strategically sound) than what they were producing before. Speed without direction just produces more of the wrong things faster.
The pressure is understandable. Everyone is watching competitors, reading industry reports and fielding questions from leadership about what the team is doing with AI. In that environment, adoption becomes the goal rather than a means to one.
The tool gets purchased before the use case is defined. The workflow gets redesigned before anyone has confirmed what problem it's solving.
What Most AI Implementations Miss
The implementations that underperform tend to share a common trait: they treat AI as a replacement for the hardest parts of the work rather than a multiplier of the best parts. They use AI to generate ideas instead of refining them, to produce first drafts instead of pressure-testing them, to simulate audience responses instead of supplementing them with real human judgment.
The result is output that's fast and mediocre rather than thoughtful and good. AI can generate twenty variations of a headline in seconds. It takes a skilled copywriter to know which one is right, why it's right and how it fits into a larger brand narrative.
Removing that expertise from the process doesn't make AI more valuable; it removes the thing that makes the output matter.
What "Using AI Well" Looks Like
The brands getting real results from AI are starting from a completely different place. They begin with a specific business problem, identify where AI can meaningfully accelerate or improve the work and keep human expertise firmly in charge of the decisions that require judgment.
Starting With the Problem, Not the Tool
KIND snack bars is a useful example of this approach done right. Facing pressure from private label competition and struggling to reach Gen Z, the Mars-owned brand's CMO Osher Hoberman identified a specific, quantifiable bottleneck: creative testing used to take three to four months and $10 million in spend. With AI-powered creative tools and synthetic audiences built from past campaign data, that turnaround shrank to a week or matter of days.
The insight that makes this worth paying attention to is what didn't change. Agency partners still account for 80% of Kind's marketing work. The AI didn't replace the expertise, it shifted what that expertise was needed for. As Hoberman put it: "Now I can use my creative agencies more as brand stewards; more strategic thought partners and less like creative and production suppliers." The problem was solved, but human judgment stayed in the room.
Amplifying Expertise at Scale
Coca-Cola’s approach to AI in their World Cup campaign tells a similar story from a different angle. Their team has embedded AI across the creative and media process; but the Coca-Cola CMO's position at POSSIBLE 2026 was unambiguous about what AI can and can't do: "AI will always be a tool for a human storyteller. If you don't have phenomenal human storytellers crafting the right level of creativity... AI is allowing all of them to do that at a much faster pace, at a scale that was completely unprecedented."
The World Cup campaign that resulted from this approach produced over 160 different combinations of the original creative, adapted across more than 100 market clusters with AI-powered variation and targeting. The creative foundation (the idea, the emotion, the story) was human. AI handled the scale, variation and delivery at a level of customization that would have been logistically impossible before.
This is what amplification looks like in practice. One strong creative idea, multiplied across markets, formats and audience segments by AI, while the human expertise that made the idea strong in the first place stays intact.
Where Human Expertise Is Still Irreplaceable
Understanding where AI genuinely adds value requires being equally clear about where it doesn't (and where assuming otherwise creates real risk).
Strategy, Judgment and Brand Instinct
AI can generate options, but can’t make good decisions about which option is right. That distinction matters enormously in marketing, where the right answer isn't always the one that tests best or optimizes most efficiently, but the one that's true to what the brand stands for and where it needs to go.
Strategic positioning, brand voice, campaign architecture and the judgment calls that determine whether something feels right — none of these are safely delegable to a model. They require people who understand the brand deeply, have context that no prompt can fully capture and are accountable for the outcomes in ways that tools aren't.
The brands that maintain this distinction perform better with AI than the ones that blur it. Walmart, for instance, has AI-enabled over 73% of its marketing investment across targeting, bidding, media placement and dynamic creative. But Walmart's own marketing leaders are explicit that the workflow coordination between those systems is still "clunky" and manual in ways it shouldn't be — a signal that even heavy AI integration requires ongoing human judgment to function well.
Taste, Cultural Nuance and Authentic Voice
Creative quality, cultural relevance and authentic brand voice all require a kind of understanding that AI approaches but doesn't reliably reach. This shows up most visibly when brands use AI to produce consumer-facing creative without sufficient human editorial judgment in the loop. The output can hit all the expected notes while still feeling hollow or missing a cultural nuance that a person would have caught immediately.
Kind's own agency partners flagged this directly. Synthetic audiences, while useful for rapid feedback, "aren't the voice of God," as VML's chief innovation officer put it — they're a model's prediction of how an audience might respond, not a substitute for genuine human understanding of who that audience is and what actually moves them.
Relationships and Trust
The relationships that make marketing work — between brands and their agencies, between brands and their creators, between brands and their customers — are built on trust that AI can support but not generate. A model can draft a brief, but it can't build the working relationship that makes a brief worth acting on. An algorithm can optimize a campaign, but it can't develop the client-agency partnership where the agency understands the brand well enough to push back on a bad idea.
This is partly why Kind's AI adoption didn't reduce its reliance on agency partners. It changed the nature of the relationship. The agencies became more valuable for what only they could do (strategic counsel, brand stewardship, creative judgment) rather than for production and execution that AI could now accelerate.
The Framework: Problem First, Tool Second
The question that separates productive AI adoption from performative AI adoption is simple: what specific problem are you trying to solve?
Identifying Where AI Genuinely Helps
Not every marketing problem is an AI problem. The ones that are tend to share certain characteristics: they involve repetitive work at high volume, they require generating many variations of something to find what performs best, they benefit from faster feedback loops or they involve processing large amounts of data to surface patterns a human would miss.
Creative testing and iteration, audience segmentation, media optimization, dynamic content personalization and rapid asset variation all fit this profile.
These are places where AI can demonstrably compress timelines, expand scale and improve efficiency without requiring it to make the judgment calls that determine whether the underlying strategy is sound.
Designing for Human Oversight
The implementations that work maintain clear human ownership of the decisions that matter. AI generates options; people choose. AI surfaces insights; people interpret. AI produces variations; people set the creative direction those variations need to express.
This isn't a limitation on AI's capabilities, it's a deliberate design choice about where accountability should sit. The teams seeing the best results aren't the ones who've removed human judgment from the process. They're the ones who've gotten better at identifying exactly where human judgment is irreplaceable, and structuring AI to handle everything else.
What This Means for How Brands Work With Agencies
The shift in how AI gets used is also shifting what brands need from their agency partners — in ways that increase the value of expertise rather than reduce it.
The agencies most valuable in an AI-accelerated environment are the ones who bring the deepest strategic thinking, the strongest creative judgment and the most genuine understanding of a brand's positioning and audience. Those things compound. They're also the things AI is worst at replicating.
For brands, this is a reason to invest in fewer, deeper agency relationships rather than more transactional ones. An agency that knows a brand well enough to be a genuine strategic partner is more valuable when AI is handling execution at scale, not less. The expertise becomes the scarce, differentiating resource which is exactly what it should have always been.
How Breef Helps Brands Find the Right Expertise
The brands winning with AI aren't just adopting tools, but finding partners who understand how to deploy them well. That means agencies with genuine expertise in their craft who know how to use AI to go further, not to cut corners.
Breef connects brands with vetted agency partners across creative, media, content and strategy; agencies that bring the human expertise that makes AI implementation actually work.
Whether you're looking for a creative partner who knows how to build a strong foundation that AI can scale, or a performance team that uses AI to optimize without losing strategic clarity, Breef matches you with partners who know the difference between using AI well and using it for its own sake.
Ready to find agency partners who amplify expertise instead of replace it? Book a demo call with Breef and build the team that makes your AI investment count.
