Artificial intelligence “agents” are envisioned to be more than simple chatbots, with the tech industry promoting AI personal assistants that understand users’ needs and perform real tasks. However, these AI agents have yet to make significant progress. Visa aims to transform this by integrating its credit card system with AI agents. By setting a budget and preferences, these AI-powered assistants—advancements beyond ChatGPT and similar chatbots—could autonomously purchase items like sweaters, groceries, or airline tickets. Jack Forestell, Visa’s chief product and strategy officer, described this development as potentially as transformative as the rise of e-commerce itself. On Wednesday, Visa announced partnerships with leading AI chatbot developers such as Anthropic, Microsoft, OpenAI, Perplexity, and France’s Mistral to link their AI systems with Visa’s payment network. Collaborations with IBM, Stripe, and Samsung are also underway. Pilot projects start immediately, with broader implementation expected next year. As a San Francisco–based payment processor, Visa believes the futuristic concept of AI agents handling routine shopping could soon become a practical convenience. The company has spent six months collaborating with AI developers to surmount technical hurdles necessary for mainstream consumer adoption. For emerging AI firms, Visa’s support could help them challenge major digital commerce giants like Amazon and Google, who are also developing their own AI agents. Currently, many demonstrations of "agentic AI" exist, though few are widely used. Most are variations of large language models—AI that can generate emails, summarize texts, or assist with coding and can search online to provide purchase recommendations. Yet, these models struggle with completing transactions themselves. Forestell noted that while early AI agents perform well in shopping and product discovery, they face significant challenges with payments.

Often, they simply defer the final purchase back to the user. Visa sees itself as a critical enabler, providing trusted access to funds so AI agents can autonomously make purchases. Since AI platforms cannot solve payment issues alone, Visa has initiated collaborations to integrate payment solutions securely. This initiative follows Visa’s recent efforts to modernize U. S. credit and debit card operations, reducing reliance on physical cards and 16-digit numbers. Many consumers are becoming accustomed to digital payments like Apple Pay, where phones replace credit cards. Similarly, AI agents would be authorized through verified digital credentials, assuring buyers, banks, and merchants of transaction legitimacy and dispute resolution by Visa. Forestell emphasized that AI agents might not replace the entire shopping experience but could efficiently handle mundane or complex errands like grocery shopping, home improvement purchases, or travel bookings. In these scenarios, users may prefer agents to handle tasks automatically. Conversely, for luxury shopping—often an enjoyable and immersive experience—many customers would prefer AI agents to assist without overshadowing their involvement. Regarding concerns about credit card debt—U. S. consumer credit card balances reached $1. 21 trillion at the end of last year—Forestell explained consumers would set clear spending limits for their AI agents, maintaining human control. Initially, agents will confirm purchases with users but may eventually gain authority to spend within set budgets, such as up to $1, 500 on airfare. AI developers are drawn to Visa's partnership partly because, with user consent, agents can access extensive data on past credit card transactions. Dmitry Shevelenko, Perplexity’s chief business officer, remarked that this data enables agents to generate personalized recommendations, informed by a user’s transaction history and preferences—for example, advising on the best laptops based on prior purchases.