An AI agent could soon compare deals, book flights and pay the bills

Major payment and tech companies are racing to build the infrastructure for what they see as the next evolution of global commerce: artificial intelligence agents that can perform searches, compare prices, and make purchases on behalf of consumers. The trend is referred to as “agentic commerce,” and reflects consumers’ growing reliance on chatbots for everyday tasks, including searching for products and deals online. However, until recently, those tools have lacked a critical function. Shoppers could search and compare inside a chatbot, but still had to leave the interface to complete a purchase. Payment giants like Visa and Mastercard say that is changing. Over the past year, both companies have been racing to forge the systems and partnerships needed for this next step in commerce, with early pilots of the technology already underway. Payment executives told CNBC that the technology will become a reality in 2026 and could be more transformative than the rise of e-commerce platforms such as Amazon . “A big shift in commerce happened when payments moved from a mostly brick-and-mortar world to an e-commerce world,” said Sandeep Malhotra, Mastercard ‘s EVP for Core Payments in Asia Pacific. “Now, we are seeing the next shift, which is moving from the e-commerce world to an agentic commerce world,” he said. “We have gone from cash to digital, now we’re going from digital to intelligent.” How it will work While many of the finer details of how and where agentic commerce will function are still being worked out, the term generally refers to AI systems that act on behalf of users to discover products, compare deals and complete payments within the chatbot. This could make shopping more seamless by curating options based on specific requests, instead of navigating multiple websites or apps in traditional e-commerce. When price discovery and shopping ubiquity become the norm rather than the exception, it will be fascinating to see how companies adapt. Visa’s APAC Head of Products and Solutions T.R. Ramachandran Payment executives say one of the obvious early use cases could be flight and vacation bookings. For instance, a user might ask an AI commerce agent: “Find me the cheapest red-eye flight from Singapore to Tokyo under $500 with no stops.” The agent will then be able to scan, provide options, book tickets and pay using the user’s stored payment credentials — all within the chat interface. Mastercard’s Malhotra said the technology could also allow shoppers to authorize agents to make purchases even when they are offline, such as automatically buying a product if its price drops below a preset threshold. Early pilots Visa and Mastercard have been rolling out their initial frameworks to secure bot-driven transactions, and have already completed pilot programs with selected users and merchants. T.R. Ramachandran, Visa’s APAC Head of Products and Solutions, told CNBC that the commercial use of personalized, secure agent transactions could come as early as the first quarter of 2026. With over half of Visa’s overall volume already through e-commerce and data showing demand for AI to assist with shopping, the ground is fertile for agentic commerce, Ramachandran said. In December, a Visa survey found that nearly half of U.S. shoppers are now using AI to enhance their shopping experience, whether to find gifts or compare prices. Meanwhile, a study from Adobe found that AI-driven traffic to retail sites in the U.S. increased by 4,700% in July from a year earlier. Agentic commerce transactions are expected to take place through AI platforms commonly used by consumers, such as ChatGPT and Google’s Gemini, as well as through merchant, bank and app-specific agents. Companies like Mastercard and Visa have been working closely with AI giants such as OpenAI to prepare for the shift. OpenAI launched a ” Buy it in ChatGPT ” feature in September that allowed instant checkouts within its platform. Meanwhile, Perplexity partnered with PayPal and rolled out a free agentic shopping product for U.S. users in November. Worried about potential price pressures and losing direct access to customers, large merchants are also testing agentic commerce tools independently. Amazon began testing its ” Buy For Me ” earlier this year, while working to block external AI agents from crawling its website. Potential issues Despite early momentum, agentic commerce still raises many potential structural, security and liability concerns. Among the main focuses of payment companies has been the creation of so-called ‘agentic tokens,’ which use cryptographic authentication to verify authorized AI agents from human users and distinguish them from malicious bots. For example, Visa launched its “Trusted Agent Protocol” in October with Cloudflare ,…