That same week, two seemingly unrelated publications told the same story. In France, Fevad released the 2026 edition of its “Key E-commerce Figures”: 94% of e-retailers now use generative AI (primarily for marketing and customer relations), and nearly 7 out of 10 rank agent-based commerce among the most promising innovations for the coming years. Internationally,the PYMNTS/Visa Index ranks AI shopping assistants as retailers’ top investment priority for the next three years. Taken together, these two insights point to the same conclusion: agent-based commerce is no longer just a conference topic—it’s an infrastructure currently being rolled out.
Two studies, one switch
The most telling figure isn’t the adoption rate, but how budgets are shifting. According to the PYMNTS study commissioned by Visa Acceptance Solutions—which surveyed 1,185 merchants in the United States, Brazil, and the United Arab Emirates—37% of retailers now cite AI shopping assistants as their top investment priority. Most importantly, this rise in prominence is accompanied by a relative shift away from technologies previously considered standard: investment in cross-channel capabilities has fallen by nine points, and support for registered payment methods and mobile apps is declining. In other words, retailers aren’t simply adding AI to their existing tech stack—they’re actively reallocating their budgets toward agent-based solutions rather than endlessly piling on new features.
On the demand side, the two studies agree. Fevad notes that nearly one in three online shoppers in France already uses AI during their shopping journey, especially in the early stages, to search, compare, and make decisions. The Visa Index, for its part, finds that 47% of online shoppers used AI during their last purchase, and that 64% expect to use an AI shopping assistant within the next two years. When supply and demand are shifting at the same pace, this is no longer a weak signal.
When Infrastructure Becomes the Standard
There remains a technical issue: until now, opening a catalog to AI agents required specific development for each conversational interface. That barrier is now being broken down. In late June, Microsoft released a public preview of the Dynamics 365 Commerce MCP server, which exposes the essential building blocks of commerce—product search, inventory availability, pricing, shopping cart, payment, and order tracking—through a single access point, so that any compatible agent can call them directly. The Model Context Protocol is thus establishing itself as the open standard that eliminates the need to rebuild a company’s technical stack with every new interface.
That is exactly what elevates a technology from the experimental stage to that of infrastructure: the moment when connecting an agent to a catalog ceases to be a custom project and becomes a convenience. The barrier is no longer technical. It becomes strategic.
The real question is no longer “if,” but “which” agent
If the act of purchasing groceries shifts toward AI agents—and both studies agree on this—then part of the customer journey will take place outside the retailer’s website and app. The relevant question for an e-commerce director is therefore no longer whether an AI agent will one day place an order on behalf of a customer, but which agent will fill that role: a third-party general-purpose assistant that acts as an intermediary between the retailer and its customer, or the retailer’s own agent—integrated into its customer experience, speaking with the retailer’s voice, and keeping the data in-house.
This is our core belief. Faced with the risk of disintermediation, private label is not a defensive option—it’s the way to maintain control over the customer relationship at the very moment it is being reshaped. An agent who starts with a customer’s desire or a recipe, puts together the shopping cart, and suggests the right products at the right time creates more value—for both the customer and the brands—than a generic agent relying on an anonymous catalog. Of course, it must be the retailer’s own catalog.
Conclusion
Two studies, one trend: adoption is here, budgets are following suit, and infrastructure is becoming standardized. Retailers who invest now in a specialized white-label provider—rather than waiting to see which third-party player will step in to take over the relationship—aren’t just following a trend: they’re positioning themselves for the food retail distribution channel of the future. This is precisely where Mealz its retail partners.
