In just a few months, generative artificial intelligence has become a staple of e-commerce discussions and is already spreading to consumer practices. According to a study published this week, nearly a third of French people (31%) use ChatGPT or other AI assistants to make their purchases. This share has jumped 46% compared to 2024. And 42% even say they are ready to adopt this technology for their future purchases.
Customers no longer browse a website: they now ask questions to an AI. Major publishers – OpenAI, Google, Microsoft, Salesforce – therefore want to transform their conversational agents into a new gateway to purchasing. The challenge is no longer to integrate AI into its systems, but to be integrated into these new third-party ecosystems: ChatGPT, Perplexity, Gemini, etc.
But these promises are often based on spectacular and futuristic demonstrations, but they are still not very operational. Transactional AI, which would allow purchasing directly from an AI agent, intrigues as much as it troubles industry players. Because if brands rush headlong into this new trend, they risk chasing an innovation without mastering its foundations, thus losing control of their customer experience.
From advice to ordering: AI transforms intention into action
AI no longer just interacts with the consumer. In addition to recommending, it can now act and purchase. It reserves a ticket, fills a basket, validates a payment. This connection between AI and the data it needs was initiated at the end of 2024 with new protocols proposed by Anthropocene (MCP). OpenAI also recently demonstrated this by launching a purchasing feature within ChatGPT, accessible to all its users. No need to go through a search engine: in a few exchanges with the AI, a product is suggested, a link is displayed, and a click is all it takes. Even Google is following suit with its “AI Overviews.” And players like Cdiscount and Leroy Merlin are already starting to adapt to this new playing field.
More generally, small retailers are also taking advantage of it: 84% already use AI to automate their marketing tasks, better target their customers, or manage their inventory (OnePoll). As for visual creation, it is being disrupted by generative AI, which makes it possible to produce seasonal, stylized, and personalized visuals in just a few seconds.
What brands can lose
In this model, the conventional e-commerce site disappears. The end customer no longer visits it: they speak to an AI agent. The interface, the content, the navigation logic... everything passes through an algorithmic filter. The selection criteria become invisible: product availability, data quality, brand reputation, logistical robustness. Traditional SEO gives way to proprietary indexing logic. Anything that is not understood or accessible will not be recommended.
The purchasing journey is also changing. No more lists of results: you ask a question, you get an answer. Simpler, faster, but also more vertical. The risk? A standardization of the offer, a reduction in diversity, and an increased dependence on third-party platforms.
Fnac Darty, for example, has launched a “data factory” to structure its data and remain visible in these new interfaces. As for Club Med, it has deployed its own assistant on WhatsApp, capable of generating personalized quotes in real time. These cases illustrate a turning point: the challenge is no longer technological, but informational. And only rigorous data governance will allow brands to exist tomorrow in the fluid environment of transactional AI.
Tackling the foundations to remain visible
In this context, brands must solidify their fundamentals before giving in to the fad. First, structure and manage the governance of their product data: clear descriptions, standardized attributes, up-to-date content. AI chooses what it understands. Then, make inventory and logistics more reliable: a recommended product must be deliverable quickly. This is a determining criterion for AI models.
E-commerce players today must be able to manage their digital sales from a central console. This ability to manage multiple touchpoints via APIs is essential, as is producing quality, well-structured content to appear in assistant responses. This is the new “SEO for AI.”
With AI, the risk of a false start is real. The commerce of tomorrow will not be limited to conversational agents. It will first and foremost rely on controlled data, an augmented human strategy, and a redesigned customer experience. It's not about slowing down, but about choosing the right pace to engage your employees, your customers, and your brand in a sustainable transformation to maintain control.
0 Comments