Innovation & Tech

Mastercard's 'Agent Pay' Opens the Digital Mall to AI Shopping Agents

Mastercard's 'Agent Pay' lets AI agents pay retailers and even parking tickets, betting on x402 and stablecoin micropayments to reshape how consumers shop.

By Olivia Hart

3 min read

AI agents are shopping like never before—and tools like Mastercard’s ‘Pay for Machines’ could turbocharge the trend
AI agents are shopping like never before—and tools like Mastercard’s ‘Pay for Machines’ could turbocharge the trendVagaro / unsplash

What's News

  • Mastercard launched 'Agent Pay', enabling AI agents to transact with retailers and government services including parking tickets.
  • A demo showed a granola-bar maker delegating a $400 social media ad campaign to an agent, covering creative assets and platform fees.
  • The tool runs on x402, an open-source payments protocol revived by Coinbase that supports cards and stablecoins.
  • Mastercard's Jorn Lambert called agent shopping 'a bit of a turning point' that is 'accelerating'.
  • Mass-market agents like Meta's Muse and OpenAI's Dots flattened the adoption curve faster than MCP/ACP protocols would have.

Mastercard has rolled out a tool called "Agent Pay" that lets AI agents buy goods and services on behalf of consumers—from retailers all the way to government services, including parking tickets. The move signals that so-called agentic commerce, long predicted but repeatedly deferred, has reached a critical mass of users far sooner than the payments industry expected.

"What we're witnessing is a bit of a turning point. We think it's accelerating…We're rapidly getting to something that will see real adoption," Jorn Lambert, Mastercard's Head of Product, said last week.

Why is agent shopping suddenly accelerating?

The lift-off stems from the unexpected popularity of mass-market personal agents, according to Lambert. A year ago, he and other industry leaders assumed that MCP and ACP—agent-to-agent communication protocols that demand technical skill—would form the pillars of agentic commerce.

That assumption is already outdated. Lambert says the adoption curve has flattened thanks to consumer-facing products such as Muse from Meta and OpenAI's Dots. The result, in his framing, is a critical mass of consumers now willing to send their agents to the digital mall.

Takeaway: consumer products, not technical protocols, made agent shopping mainstream.

What else did Mastercard launch?

Last week the card network also revealed a tool that lets small-business users spin up advertising campaigns. In a demonstration video, the maker of a granola bar gives her agent a $400 budget to build and execute a social media campaign from scratch.

That budget covers more than paying TikTok or Instagram. It also covers acquiring images, music, a script and other creative elements—a process that once would have consumed hours of the business owner's time but can now be dispatched in minutes. Visa and American Express are presumably building similar products, though neither has announced one.

What is x402, and why does it matter?

A significant feature of Mastercard's tool is that it is built on x402, a payments protocol that was baked into the original specifications for the internet but never got built until Coinbase resurrected it a few years ago.

As Lambert explained, x402 is essentially a messaging language that can speak credit card, debit card or stablecoin—or whatever form of payment both parties to a transaction agree upon. Crucially, x402 is open source, meaning anyone can use it or build upon it. That openness lowers the barrier for competitors and startups alike.

Takeaway: an open-source payment rail underpins the entire push, not a proprietary card-network standard.

Could this finally make micropayments work?

Agent Pay opens the door to micropayments—trifling sums such as 27 cents or $1.87 transacted online. Until now, digital merchants and financial networks have shunned such payments as impractical, partly because of processing costs.

Stablecoins change the economics, and agents bring a new behavior: they do not mind haggling over whether a piece of data should cost 50 cents or 75 cents. The likely consequence, the argument goes, is more à la carte micro-offerings in place of every webpage demanding a monthly subscription to access content.

The caveat: micropayment predictions have failed before, and no adoption figures for such offerings exist yet.

What comes next?

Predicting how new technology will be adopted is always perilous. The instinct is to use the new tool—in this case AI agents—to do what we have always done before. A classic example: newspaper publishers initially treated the internet as a place to post PDF copies of their papers.

Lambert predicts that tools like Agent Pay will spur a wave of new businesses founded by college kids tinkering in their dorm rooms. What he does not predict, and what nobody can, is the specific use cases that will emerge and change how consumers shop and how businesses are designed in the first place.

The shopping spree has also had a turbulent backdrop in recent weeks: AI agents have been rummaging around bank accounts, hacking crypto wallets and, according to MIT researchers, using their spare time to play. Whether payments infrastructure can scale faster than agent misbehavior remains the open question for the year ahead.

Original: finance.yahoo.com

Share this article:

More from Olivia Hart

Olivia Hart

Show full bio

Staff writer covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

1,185 articles

Related articles

« Previous articleNext article »