AI is no doubt changing the path from brands to consumers. This blog post is not about AI for efficiency or how AI is used for creative production. I want to address what I consider to be one of the biggest blind spots right now — what AI means for brands themselves.
For hundreds of years, a brand controlled every word and visual on the path to the consumer. It would take months to agree on the exact phrase, format and angle before brand communication would become published. This brand-to-consumer path now runs through AI. Everyone is still trying to solve for search — and it's a good place to start — but the shift is so, so much more than search. Every channel is becoming AI-powered. So what does that change for brands?
A Brand Becomes Distances and Patterns of Meaning
To AI, a brand is not a creative expression, a story, or a feeling. To AI, a brand is a mathematical representation of meaning. Text and images turn into tokens, each token becomes a vector: a list of hundreds to thousands of numbers encoding its meaning. Your brand dissolves into them — a brand is not stored anywhere as a record or an entry. It dissolves into those numbers, and lives inside AI as relationships, distances, and patterns of meaning, interpreted through everything the model already knows.
AI Doesn't Retrieve Brands. AI Generates Brands.
Which leads to the fact that AI does not retrieve brands. It generates a fresh version of your brand every single time someone asks. I am oversimplifying, but directionally, each time a consumer asks a question that touches your category, roughly the same four things happen.
- The model interprets the question, resolving real intent in meaning, not keywords.
- It selects evidence by meaning — readable content, yours and thousands of other sources, matched by vector distance and ranked.
- It assembles a working context for this one answer.
- It generates the reply token by token, each one chosen by probability and appended, over and over, before the reply is displayed to the user.
Variable by Design
But that's not all. What informs that response is multi-dimensional: the context, the user's intent, whether the evidence comes from training memory or real-time retrieval or both, and hundreds of other signals. The result is variable by design. Same question, different answer, every customer. You can't switch that off; it is how every one of these systems works. That's why measuring how many times your brand is mentioned in/by AI systems is an approximation based on synthetic prompts — be careful with using that data to inform your brand growth decisions.
Until now, every consumer used to get the same version of the brand message, fully controlled by the brand. Now, in AI-mediated journeys, every consumer gets a differently composed version, and it comes with or without the brand's input. Your message is one input among many.
Today, Most Brands Appear Blank to AI
Brands have a lot of work ahead of them. Our minds default to what's known and what seems like an obvious solution — create a ton of content for AI to consume. This is not exactly the path though. Most, if not all, sources from where AI systems get their data were built for human consumption. Most of them are hardly readable to AI, or provide little meaningful value (because marketing content that inspires humans doesn't really add that mathematical representation to AI bots). On top of that, most brand websites are not even accessible to AI crawlers. And no, AI does not see ad campaigns. Today, most brands appear blank to AI.
In the next article, I'll talk about some immediate steps brands need to take.
If you remember one thing today: to AI, a brand is a mathematical representation of meaning. Everything else follows from that.