“Consumers Are Turning to AI Agents to Complete Purchases” Says Mastercard CPO

AI tools like ChatGPT and Copilot have become mainstream and more people are turning to conversational agents, not traditional search engines for product discovery. But while AI can now personalize recommendations and streamline decision-making, there’s still a disconnect when it comes time to purchase. Consumers often have to leave these platforms to complete a transaction, breaking the flow and reintroducing friction.

Mastercard sees this gap as both a challenge and an opportunity. With the launch of Agent Pay, it aims to bridge that divide by embedding secure payments directly into AI-powered experiences and enabling a new model of commerce where agents can search, recommend, and that Mastercard will be the trusted engine underneath it all.

At the Fortune Brainstorm Tech conference this week, Mastercard Chief Product Officer Jorn Lambert unveiled the company’s ambitious vision for “agentic commerce” , a future where AI agents not only help users discover products, but also transact on their behalf. 

From Browsers to Bots

For over two decades, online shopping has been shaped by one behavior: the search. Consumers typed “best running shoes” into Google, scrolled through endless listings, and clicked through websites to compare options. But as Lambert points out, that behavior is already changing. Now, people are asking the same questions of AI agents like ChatGPT, expecting not just a list, but a curated, personalized answer.

“Consumers are increasingly turning to AI agents to search, discover, and now even complete purchases,” said Lambert. “It’s a paradigm shift in commerce.”

Mastercard’s Agent Pay is designed to enable that shift. Rather than abandoning a conversational platform to complete a purchase, consumers can now authorize an AI agent to carry out the transaction for them securely and autonomously. The solution integrates Mastercard’s trusted payments infrastructure directly into AI platforms, allowing transactions to unfold seamlessly inside the same experience.

Building Trust in a Machine-Led Marketplace

The cornerstone of Agent Pay is security and transparency. Central to this is Mastercard Agentic Tokens, which build upon the company’s existing tokenization capabilities. These tokens replace sensitive card details with unique, cryptographically generated codes that can only be used by the verified AI agent on the user’s behalf.

“Even if a bad actor gains access to that credential, we can bind it back to its origin,” Lambert explained. “That’s the power of tokenization.”

Every AI agent that participates in this ecosystem must be registered and authenticated. Users can set clear permissions such as spending limits, merchant categories, or frequency—and all purchases must be verified using Mastercard’s biometric authentication tools, such as on-device face or fingerprint recognition.

This proactive design also addresses a growing concern in the digital economy: fraud. In 2024 alone, online shopping fraud reached $750 million, according to the U.S. Federal Trade Commission. While many traditional fraud prevention tools rely on post-transaction checks, AI agents when armed with embedded security protocols can prevent fraud in real-time. Carlo Bruno, VP of Product at Adyen, noted the delicate balance between security and autonomy but acknowledged that AI offers new tools to fight fraud more effectively.

Agentic Commerce in Action

So what does agentic commerce look like in the real world? Mastercard paints several compelling scenarios.

A soon-to-be-30-year-old can chat with an AI agent to curate party outfits from both online retailers and local boutiques, personalized based on her style, budget, and even the venue’s weather forecast. Once she confirms her preferences, the agent completes the purchase, suggesting the best payment method, such as Mastercard One Credential.

A small textile manufacturer might use an AI agent to source fabrics from international suppliers, negotiate terms, and execute payments using a virtual Mastercard corporate card token, handling logistics and delivery in one continuous workflow.

In both cases, Agent Pay ensures that every participant—consumers, banks, merchants, and AI developers can recognize, verify, and track the transaction at every step.

Collaboration for Scale

To realize its agentic commerce vision, Mastercard is aligning with some of the biggest players in enterprise AI and payments. The company is working with Microsoft to integrate payment capabilities into Azure OpenAI Service and Copilot Studio, enabling both consumer-facing and B2B applications. It’s also partnering with IBM’s watsonx Orchestrate to streamline B2B automation and is in talks with major checkout providers like Braintree and Checkout.com to scale tokenization for merchants.

“This isn’t just about launching a product,” said Lambert. “It’s about setting the standards for how agentic payments should work. We’re collaborating with the industry to ensure transparency, trust, and scale.”

Mastercard is also developing new merchant interfaces to distinguish trusted AI agents from malicious actors. By advancing protocols such as the Model Context Protocol to Secure Remote Commerce, the company is laying the groundwork for a standardized agentic economy.

For all the hype surrounding generative AI from silly memes to corporate productivity tools, its impact on commerce may be among the most transformative. If Mastercard’s vision succeeds, the act of shopping could become less about searching and more about simply asking and trusting your agent to handle the rest.

“The world of commerce is being redefined,” Lambert said. “With Agent Pay, we’re creating the rails for the next generation of payments—where intelligence, identity, and intent converge.”