**Agentive AI** is reshaping the way businesses interact with customers, and it's already on the desks of major companies like COMEX, a French trading group. Retailers are being warned that they need to control access to AI agents on their platforms before larger platforms impose their own rules. Amazon recently blocked the Comet agent from Perplexity, a company that provides AI tools, highlighting that in an AI-driven marketplace, whoever controls access to customers controls the business. Olivier Vigneaux, director of Europe for Havas CX, says European brands must treat agentive AI not as a technical project, but as a strategic business decision, while they still have the option to decide which AI agents they allow. In the autumn of 2025, Amazon took legal action against Perplexity, claiming that Comet, one of Perplexity's AI agents, accessed customer accounts to make purchases directly, bypassing Amazon's usual revenue streams like ads and product recommendations. In March 2026, a court ruled in favor of Amazon, ordering Comet to stop operating on its platform. Perplexity has since appealed the decision, and the case remains unresolved. While this legal battle might seem like a dispute between two American companies, it highlights the growing tension that French retailers will soon face as well. For years, retailers had full control over the customer journey, from the initial search to the final purchase and after-sales service. The arrival of agentive AI has changed this dynamic. Customers now use AI assistants to compare products, filter options, and even make purchases on their behalf. Amazon recognized this shift early and acted accordingly. This transformation has two major implications for brands. When a transaction happens, the brand typically gets standard data like the product, the order, and the payment method. But what is lost is the customer's intention—the way they compared the brand to others before making a decision. This data is controlled by large AI companies like Google, OpenAI, or Anthropic. What brands can manage is which AI agents are allowed to access their platform. Without a clear framework, brands lose track of what each assistant brings and where to invest their resources. Amazon was able to identify that Comet was impersonating customers by hiding its identity, which led to the legal action. However, many AI agents are not even given access to begin with. They remain invisible to the systems that make purchasing decisions for customers, excluded even before the issue of access is addressed. Visibility in AI systems is now critical for commercial survival. According to Adobe, customers who come to a brand through conversational assistants convert 42% better than those from traditional channels, and this conversion rate has improved compared to the previous year. These customers have already made their decisions—they just need to validate the purchase. Brands that are not present in this channel risk cutting themselves off from a segment of customers who are the most active and efficient buyers. Inside companies, the common reaction is to treat agentive AI as a technical project, assigning it to IT teams to connect AI tools and wait for results. However, this approach underestimates the importance of the decision. It needs to be made at the highest level of company management. Amazon took a firm stance by going to court rather than letting customer relationships slip away. European retailers still have time to make this choice, but each month of delay reduces their options. The longer they wait, the more they will have to accept the formats and conditions set by larger platforms, and the less control they will have over their own future.