Most of the discussion about AI and brands is still happening one layer too low.
First, the conversation was about content. AI would make ads faster, cheaper, more personalized. Then it moved to discovery. Brands now have to be found by machines, cited by answer engines, optimized for GEO, structured for AI search, and legible to agents.
That is all real. It is also still too narrow.
The deeper shift is not only that AI changes media. It is not even only that AI changes discovery. The deeper shift is that AI is moving into the mechanism that connects demand and supply. It is beginning to interpret what people want, frame the options available, narrow the choice set, and guide action. That is not just a media shift. It is a market shift.
I have argued this before in my writing on the evolution from the attention economy to the intent economy.
The Stanford AI Index 2026 also helps explain why this matters. The report does not directly argue that AI is becoming the coordination layer between demand and supply. But it does show the conditions for such a shift: generative AI reached roughly 53% population adoption within three years, organizations reported 88% AI adoption, and estimated consumer value from generative AI in the U.S. reached $172 billion. In other words, AI is no longer a marginal tool. It is becoming part of the infrastructure through which markets are searched, interpreted, and acted on.
Standford University 2026 AI Index Report
In my earlier article, I argued that brand performance is a layered 5 — part system. It starts with category meaning, then demand logic, then brand strategy and positioning, then execution, and finally market performance and economics. What has changed since then is not the model. What has changed is that AI is now making the consequences of that model much harder to ignore.
What AI Teaches Us About Strategy and Brands
The market is reacting mainly at the execution layer, while the real strategic effects are working across all these five layers. Stanford’s 2026 AI Index reinforces that point. It shows both rapid diffusion and rapid capability gains, especially in areas tied to reasoning, professional tasks, and agentic performance. That matters because AI is moving beyond content generation into areas closer to judgment, task completion, and market action, even if reliability remains uneven.
You can see it in the practical discussion now shaping the market.
Advertisers are using generative AI to create video ads at scale. Google is pushing marketers toward AI-powered campaign systems. Meta is moving toward more automated advertising, from creative to targeting to budget allocation.
This is what most people mean when they say AI is disrupting marketing: faster content, automated media, better targeting, more efficient execution.
Then comes the discovery layer.
BCG now talks openly about the future of discoverability, arguing that brands need to be selected, summarized, and surfaced by AI systems rather than simply rank in traditional search. OpenAI’s own product direction points the same way. ChatGPT shopping is built to help people discover, compare, and decide on products. Atlas, shopping updates, and agentic commerce all point toward the same structural move: discovery, recommendation, and action are being compressed into fewer steps and more conversational flows.
This is where the current wave of advice comes from.
Make your content machine-readable. Optimize for answer engines. Structure your product data. Prepare for conversational commerce. Make sure your brand can be found by AI.
None of that is wrong. It is still too downstream.
Because it assumes the strategic task is to optimize execution inside a market whose deeper logic is already understood. That is the mistake.
The more important question is not whether the machine can find you. The more important question is what market logic the machine is learning, reinforcing, or rewriting.
That question starts upstream. It starts with category meaning.
A machine cannot help a consumer decide unless it has a working model of what the category is for. That sounds abstract until you see it in practice. Instacart’s Ask Instacart was not introduced as a better ad format or a smarter search box. It was introduced around a consumer problem: what is for dinner, what fits my dietary needs, what can I make on a budget, what ingredients do I need, and how do I get them fast. The system turns that moment of demand into a shoppable basket.
That is not media optimization. It is category meaning and demand logic in action.
The consumer is not entering through “sauce” or “snacks” or “brand X.” The consumer is entering through a demand moment: dinner, budget, dietary restriction, convenience, time. The system then organizes supply around that moment.
The same pattern is visible at Lowe’s. Lowe’s describes Mylow as an AI shopping assistant and virtual advisor that helps customers plan projects and find the right solutions, whether that means a refrigerator with installation or a washer-dryer set that saves on utility bills. That is not just search. It reflects a different understanding of the category itself. Lowe’s is not simply helping shoppers find products. It is helping people solve home projects and household problems.
That distinction matters.
Once AI moves into the match between demand and supply, the strategic questions move upstream very quickly.
First, category meaning matters more. If the machine is working with the wrong understanding of what the category is for, it will organize the choice set the wrong way. A firm may be easy to find and still lose, because the machine is playing the game using someone else’s frame.
Second, demand logic matters more. If the demand map is wrong, AI will simply optimize the wrong map faster. It will look more efficient while deepening the mistake.
Third, positioning matters more. Positioning is no longer just what the brand says about itself in communications. It is what role the brand can credibly occupy in recommendation, comparison, and action flows. A brand can show up in AI results and still have no clear reason to win.
Fourth, portfolio and architecture matter more. In a machine-mediated environment, overlap gets exposed faster. Gaps get exposed faster. Weak role clarity across brands becomes more visible, not less.
Fifth, brands must be repositioned not as a communications function but as the central engine of economic value creation — measured through what I call compound economic profit (my favorite CEP): the mechanism by which strong brands generate returns above cost of capital, reinvest that surplus into meaning and demand logic, and compound the advantage over time. This is the loop that distinguishes brands that build lasting value from those that merely sustain volume. Layers 1 through 4 feed it. Layer 5 is where you find out whether they did their job.
WARC – Reinventing brand strategy: The holistic brand model
This is also why I am cautious about the current fashion for talking about a Brand OS, a brand operating system.
There is something useful in the phrase. It recognizes that brand must be more systematic and more embedded in the organization. But most Brand OS thinking today still lives too far downstream. It is usually about one source of truth, prompt-safe guidelines, brand-consistent AI output, content governance, and executional coherence. In other words, it is often an execution system dressed up as strategy. Even where Brand OS is framed more broadly, it still tends to systematize expression before it systematizes the market logic the brand should run on.
That is why Brand OS, as it is often used, still misses the point.
The issue is not whether the brand has a better operating system for content. The issue is whether the business has a better operating system for category meaning, demand logic, positioning, portfolio role, execution, and value capture. If those upstream layers are wrong, a more efficient Brand OS will only produce more coherent irrelevance.
I have a similar problem with the phrase AI strategy.
Most companies do not need an AI strategy in the narrow sense (see our AI report from Vivaldi). They do not need a separate deck about use cases, pilots, agents, or tooling. What they need is an understanding of how AI changes the system they are in.
This is why the broader system lens is more useful (Reshuffle, Choudary 2025). The real question is not how to deploy AI. The real question is how AI changes coordination, value creation, and competitive structure. Recent work on agentic organizations makes the same point from a different direction: AI is not simply another technology layer, but a force that changes workflows, team structure, budget logic, and the operating model itself.
So no, most firms do not need an AI strategy.
They need a view of how AI is changing the category, the decision process, the way demand is formed, the way supply is organized, and the way value is captured.
That is a much harder task.
But it is also the real one.
The first wave of AI strategy treated AI as a content engine. The second treats it as a discovery engine. The larger shift is that AI is becoming a coordination layer between demand and supply.
Stanford’s 2026 AI Index is especially useful here. One of its clearest findings is that AI agents advanced in 2025 from answering questions to completing tasks. On OSWorld, performance rose from roughly 12% to 66.3%, within six percentage points of human performance. That does not mean AI is already dependable everywhere. It does mean that the economic role of AI is expanding from information toward coordination and action.
That is the more useful framing.
It explains why the media story is visible but incomplete. Brands such as Coca-Cola and Nestlé are already using AI in practical ways. Coca-Cola has continued to use generative AI in major campaign work (an example is the Christmas campaign done by our creative partner, Silverside.ai). Nestlé has launched an AI-powered content service using digital twins of products to generate ecommerce and digital media assets faster and at lower cost. Those are serious moves. But they are still mainly execution moves. They improve execution. They do not answer the harder questions upstream.
So what does it look like when a company gets the upstream layers right?
A strong example is La Roche-Posay.
Not because it has the flashiest AI story. But because it shows what happens when a brand builds coherence across all five layers and then uses technology to strengthen that system.
Start with category meaning. La Roche-Posay does not simply play in skincare. It sits inside a much sharper meaning space: dermatological beauty, skin health, science-backed care, and solutions for real skin conditions. L’Oréal’s Dermatological Beauty division describes itself through healthcare partnerships and dermatological expertise, and reported €7 billion in sales in 2024.
Then demand logic. La Roche-Posay is not organized around generic beauty language. It is built around clear conditions and moments: acne, hyperpigmentation, sun care, sensitive skin, compromised skin barriers. The Mela B3 launch is a good example. L’Oréal says the brand conducted a global epidemiological study on pigmentary disorders covering 48,000 people across 34 countries before launch. That is not a content move. That is demand structure work.
Then positioning. The brand’s role is unusually clear: dermatologist-backed efficacy with broad accessibility and high tolerance. That role is stable across categories and geographies. It is not just advertising copy. It gives the brand a distinct right to win in specific demand spaces. L’Oréal’s own materials emphasize that La Roche-Posay remains a major growth engine within the division.
Then execution. Here AI becomes useful, but only because the upstream layers are already coherent. La Roche-Posay’s SPOTSCAN+ is an AI-based acne skin analysis tool developed with dermatologists. It analyzes selfies, scores acne severity, and connects the diagnosis to product recommendations. That is a much better example of AI than generic ad automation because it links a real consumer problem, a clear demand space, the brand’s positioning, and the service layer in one flow.
Finally, economics. The system is working. L’Oréal’s 2025 annual report said La Roche-Posay remained the number one growth contributor in Dermatological Beauty. In first-quarter 2026 results, L’Oréal said the division continued to outperform its market, with La Roche-Posay’s momentum supported by Cicaplast and Hyalu B5.
It did not win because it optimized for AI first. It won because it built a coherent system across category meaning, demand logic, positioning, execution, and economics. AI then became an amplifier of that system, not a substitute for it.
That, in the end, is the point.
Brands do need to be found by machines. But being found is not strategy.
Brands do need more systematic operating discipline. But a Brand OS that lives only at the execution layer is not enough.
Companies do need to respond to AI. But that response should not begin with a narrow AI strategy.
Stanford’s report points to the same conclusion from another angle. AI is spreading at market scale, delivering real consumer value, improving in task execution, and becoming more central even as transparency and institutional preparedness lag. That is why the real question is not how to optimize for AI, but what kind of market logic AI is beginning to run.
It should begin with a harder question:
What system of meaning, demand, positioning, portfolio, execution, and economics do we want the market to run on as AI increasingly mediates the match between demand and supply?
The brands that win in this next phase will not be the ones that merely optimize for machine visibility. They will be the ones that shape the category logic, demand structure, and brand roles that the machine is learning to interpret, recommend, and act on.
That is no longer a media issue.
It is the next chapter of strategy.