A study conducted by Fevad and KPMG, published in September 2026, highlights that 31% of French online shoppers are already using generative AI to assist with their purchases. Generative AI refers to technologies that can create text, images, or other content based on user prompts. However, despite this growing usage, many shoppers still lack confidence in delegating payment tasks to AI. Currently, AI plays a role in advising, comparing products, and offering reassurance, but it does not yet handle payment. While agent-based payment—where AI manages the entire transaction—is still a small part of online shopping, the trend shows a clear direction toward increased reliance on AI in e-commerce. France's e-commerce sector is growing steadily, with a 7% increase in 2025 and an average purchase value of 62 euros. The Fevad/KPMG study reveals that French consumers are already becoming familiar with generative AI, with 73% of regular users incorporating it into their purchasing process. However, 57% of those who still avoid using AI for purchases cite concerns about its commercial neutrality—meaning they worry AI might favor certain products or companies. The study breaks down AI's role at each stage of the buying journey, showing that 58% of users rely on it before making a purchase for advice and comparisons, with a confidence level of 47%. During the actual purchase, only 27% use AI, often to check delivery times or review their shopping cart, and confidence here drops to 30%. After the purchase, 35% use AI for customer service or post-purchase support, with a moderate confidence of 43%. Agent-based payment, where AI handles the entire transaction including payment, is still in its early stages in Europe. The study does not quantify how many transactions are currently handled by AI agents, partly because the necessary services are not widely available in the region. In contrast, across the Atlantic, agent-based commerce is being tested more actively, though it accounts for less than 1% of e-commerce activity, according to Bernstein, a financial research firm. Trust remains a major barrier—only 23% of consumers, as reported by a Visa survey, trust AI to pay on their behalf. However, nearly half of the French population is expected to adopt agent-based commerce by 2030, according to Fevad's general delegate, Marc Lolivier. To prepare for this shift, the study recommends that e-commerce businesses move beyond experimentation and focus on industrializing AI-based services. The plan involves mapping AI agent interactions, measuring customer journeys, and identifying risks over three months. Over the next six to twelve months, merchants should focus on enriching product catalogs to support AI integration. Once fully implemented, eight key performance indicators (KPIs) will be used to assess the success of agent-based commerce: the share of traffic generated by AI, product visibility on AI platforms, conversion rates with and without AI assistance, cost per automated interaction, the rate of human intervention, AI error rates, the frequency of security or compliance issues, and customer satisfaction and trust. These metrics will guide the development and refinement of AI-driven e-commerce strategies in the coming years.