Visa has launched a new platform designed to let artificial intelligence agents purchase products on behalf of users, effectively giving AI access to people’s credit cards — with strict guardrails. The system, called Visa Intelligent Commerce, was unveiled last Wednesday at the company’s Global Product Drop event in San Francisco and enables AI assistants to not only recommend products but complete transactions.

“Soon people will have AI agents browse, select, purchase, and manage on their behalf,” said Jack Forestell, Visa’s Chief Product and Strategy Officer, during the announcement. “These agents will need to be trusted with payments, not only by users, but by banks and sellers as well.”

The initiative is built on a network of partnerships with leading AI companies including Anthropic, IBM, Microsoft, Mistral AI, OpenAI, Perplexity, Samsung, and Stripe, among others. This collaboration aims to embed payment capabilities directly into AI systems that are already transforming how consumers discover products and services.

Visa’s new platform addresses a critical gap in the current AI commerce landscape. While AI systems have become increasingly sophisticated at helping users find products, they typically hit a wall when it comes to completing transactions.

“AI commerce is a new commerce experience where AI agents play an active role in helping users shop online,” explained Rubail Birwadker, SVP and Head of Growth, Products & Partnerships at Visa, in an interview with VentureBeat. “Today, agents help largely with product discovery, but with Visa Intelligent Commerce they will start to transact on behalf of users.”

The system works by replacing traditional card details with tokenized digital credentials that can be securely accessed by authorized AI agents. Users maintain control by setting specific parameters, such as spending limits and merchant categories, while the AI handles the transaction details.

For example, a consumer could instruct their AI assistant to book a flight to Cancún under $500, order weekly groceries, or find the perfect gift for a family member. The AI would search across multiple sites, compare options, and complete the purchase — all without requiring the consumer to manually enter payment information at each step.

“There is tremendous potential for the role AI agents will play across a wide variety of commerce use cases, from everyday tasks such as ordering groceries, to more sophisticated search and decision-making like booking vacations,” Birwadker noted.

How Visa plans to make AI transactions secure in an era of digital fraud

The announcement comes at a time when concerns about AI security and data privacy remain high among consumers. Visa appears to have anticipated these concerns by making security a central feature of the platform.

“Visa takes an intelligence-driven approach to understanding new and novel fraud and cybercrime threats against emerging technology,” Birwadker said. “Just like we identified, researched and built controls and best practices for using generative AI, Visa is also committed to identifying, researching and mitigating threat actor activity targeting agentic commerce.”

The company is leveraging its decades of experience in fraud detection and prevention, with executives noting that Visa’s AI and machine learning systems blocked approximately $40 billion in fraud last year alone.

The system includes several key security components. The AI-Ready Cards replace traditional card details with tokenized credentials, enhancing security while simplifying the payment process. The platform also implements identity verification, confirming that a consumer’s chosen AI agent is authorized to act on their behalf. “Only the consumer can instruct the agent on what to do and when to activate a payment credential,” the company emphasized. Additionally, transactions generate signals that are shared with Visa in real-time, enabling the company to enforce transaction controls and assist with dispute management.

“Transactions made by an AI agent will be tokenized, meaning the card details are replaced,” Birwadker explained. “For personalization, Visa uses a data privacy-preserving framework. Data requests are managed through data tokens, which allows for consent management and control by the consumer, payment credential tokenization for the purpose of data sharing, and secure and encrypted transmission of data.”

A distinguishing feature of Visa’s approach is the emphasis on user control. Consumers can set spending limits, specify merchant categories, and even require real-time approval for certain transactions.

Mark Nelsen, Visa’s global head of consumer products, told PYMNTS that the system allows consumers to set parameters such as a “$500 ceiling for a hotel or an airline ticket.” The AI agent then works within these constraints, finding options that meet the consumer’s criteria without exceeding preset limits.

“For personalization, Visa uses a data privacy-preserving framework,” Birwadker told VentureBeat. “Data requests are managed through data tokens, which allows for consent management and control by the consumer, payment credential tokenization for the purpose of data sharing, and secure and encrypted transmission of data.”

This approach reflects Visa’s understanding that consumer adoption hinges on maintaining a sense of agency while delegating shopping tasks to AI. The company has carefully designed a system where convenience doesn’t come at the expense of control.

Visa positions itself at the center of AI commerce revolution with 200-country network

The announcement represents Visa’s effort to position itself at the center of what could become the next major shift in how consumers shop online — potentially as significant as the transitions from physical to digital shopping and from desktop to mobile commerce.

“Just like the shift from physical shopping to online, and from online to mobile, Visa is setting a new standard for a new era of commerce,” Forestell said. “Now, with Visa Intelligent Commerce, AI agents can find, shop and buy for consumers based on their pre-selected preferences.”

Industry analysts note that Visa’s global footprint — spanning more than 200 countries and territories — gives it a significant advantage in scaling this technology. The company’s existing tokenization framework and merchant relationships provide the infrastructure needed to make AI commerce viable on a global scale.

When asked about adoption timelines, Birwadker expressed confidence in the technology’s future: “AI adoption is real and the numbers prove it. When you see metrics like OpenAI topping 400 million users and other GenAI platforms building their audiences, it’s hard to imagine AI commerce not catching up. Visa Intelligent Commerce is a critical step and building block for the industry to start form and help create widespread adoption.”

The company indicated that developers can begin implementing the APIs immediately, with pilot programs expected to launch in the coming months.

Trust remains a key challenge for consumers giving AI access to financial data

Despite the enthusiasm from Visa and its partners, some consumers and privacy advocates remain skeptical about giving AI agents access to payment information. The company appears to be taking a measured approach to address these concerns.

Visa’s system allows consumers to start with small, low-risk transactions before granting their AI assistants greater purchasing authority. The company emphasizes that consumers retain the same protections and fraud safeguards they currently have with their Visa cards, even when transactions are initiated by AI agents.

The extensive partnerships Visa has developed for this initiative demonstrate its commitment to building a comprehensive ecosystem rather than an isolated product. “We’re working with each of the partners you mentioned to bring better commerce experiences to consumers,” Birwadker told us. “For example, with OpenAI we’re enabling agentic commerce securely and safely at scale. And those aren’t the only ones we’re working with – we’re currently with more than 20 payment service providers and AI infrastructure enablers on this program.”

Visa’s announcement comes amid growing competition in the AI commerce space. Several technology companies and financial institutions are exploring ways to integrate AI into the shopping experience, though few have unveiled comprehensive solutions that bridge product discovery and payment.

What distinguishes Visa’s approach is the focus on embedding payment capabilities directly into the AI agents that consumers are already using for other tasks. Rather than creating a separate shopping assistant, Visa is opening its payment rails to existing AI platforms, potentially accelerating adoption.

This strategy also addresses the current fragmentation in online shopping, where consumers often need to create accounts and enter payment information across multiple sites and apps. By allowing AI agents to handle these transactions seamlessly, Visa aims to reduce cart abandonment and simplify the commerce experience.

“Unlike existing e-commerce, which places the task on consumers to search and transact, Visa Intelligent Commerce will enable seamless and scalable transactions through autonomous agents,” Birwadker said.

As Visa prepares to roll out its Intelligent Commerce platform, the boundaries between human and artificial decision-making continue to blur in the commerce landscape. For decades, we’ve entrusted humans—from retail clerks to travel agents—with our shopping needs. Now we stand at the threshold of trusting algorithms with not just our product searches, but our wallets as well.

The ultimate success of Visa’s initiative will depend not on the technology itself, which is increasingly capable, but on the psychological shift it requires from consumers. People who grew up memorizing credit card numbers and guarding them closely must now decide whether the convenience of AI-powered shopping outweighs ingrained caution about financial data.

Throughout history, shoppers have always adapted to new ways to buy — from marketplace haggling to department stores, from mail-order catalogs to e-commerce — each transition has required consumers to place their trust in new systems.

Visa’s AI commerce initiative asks for perhaps the most profound trust leap yet: allowing an artificial agent to not only know what we want but to spend our money acquiring it. As shoppers face this choice, they’re not just evaluating a new technology. They’re deciding whether to outsource the final human element in commerce — the decision to buy.