With the rise of large language models — ChatGPT, Claude, Gemini, Perplexity, Grok, Deepseek, Meta Llama — consumer behaviour is undergoing a profound transformation. The online user no longer clicks; they enquire. They no longer browse; they listen. In the course of a single conversational exchange, they receive a synthesis, a recommendation. And within that synthesis, your brand is either present — or it is absent. There is no middle ground.
Ninety per cent of all queries directed at AI tools flow through this handful of platforms. They are not merely search engines of a new generation. They are reputation integrators: they aggregate, rank, and render a vision of the world assembled from thousands of distinct sources. And it is that vision — not your own, at least not yet — that will shape the convictions of tomorrow’s consumer.
Yesterday, through SEO (Search Engine Optimization), the imperative was to exist within a list of links. Tomorrow, through GEO (Generative Engine Optimization), the imperative is to exist within a narrative. Generative AI does not return ten results: it tells a story. It synthesises, it adjudicates, it recommends. And the brands that fail to actively nourish this narrative vanish from the account — not because their search optimisation was lacking, but because they never learned to speak the language of today’s algorithms.
This shift is all the more vertiginous when one considers that visitors arriving via a large language model are, by every emerging measure, considerably more disposed to purchase than those arriving through a conventional search engine. AI pre-qualifies, pre-selects, pre-persuades. To be recommended by it is to have already won a decisive engagement.
To secure a place within the responses of conversational AI agents, three imperatives present themselves as absolute priorities:
1. Become a thematic authority. Build a discourse that is compelling, coherent, and distinctive. Large language models reward depth and consistency — not abundance.
2. Align your content with the questions your consumers are genuinely asking. It is they who formulate the queries; it is AI that answers them. To anticipate those questions is to position yourself within the response before the question has even been posed.
3. Earn validation from third parties. Large language models place no trust in brands that speak solely of themselves. They trust independent sources: influential journalists, renowned sommeliers, authoritative guides, online communities, and specialist platforms. To be cited by these voices is to feed directly the AI that will, in turn, feed tomorrow’s customers.
Generative AI does not judge a bottle by its colour, its bouquet, or its finish. It judges a brand by the clarity of its positioning, by the coherence of its messaging across the full breadth of sources that inform it, and by the frequency with which credible, independent voices choose to cite and commend it.
The brands that invest today in their digital ecosystem — that structure their discourse with rigour, multiply their presence across the sources referenced by large language models, and forge concepts durably associated with their name and their terroir — will be the brands that AI recommends tomorrow. This does not mean that the soul of wine disappears. It means, quite simply, that this soul must henceforth be rendered in a language that machines can comprehend, so that it may reach the human beings who listen to them.
Contact Guillaume Jourdan via LinkedIn
