Payment Intelligence Series  |  IntelliPay

Agentic Commerce and AI Payments: What Every Merchant Needs to Know in 2026

AI agents are already completing real purchases on live payment infrastructure. Here is what that means for your business and what you can do to get ready.

What Is Agentic Commerce and Why Does It Matter Right Now?

There is a quiet revolution happening in payments that is easy to miss because it mostly takes place behind the scenes. While consumers are still tapping cards and typing in credit card numbers, a completely different kind of payment infrastructure is being built underneath the surface. One designed for a world where the shopper is a bot.

Agentic commerce is the term the industry uses for transactions initiated and completed by artificial intelligence agents acting on behalf of human users. Instead of a person navigating a website, adding items to a cart, and entering payment details at checkout, an AI assistant does all of that on its own, under rules and spending limits the consumer sets in advance.^[1]

Think about what that actually means for a moment. A consumer tells their AI assistant: "Order my usual coffee subscription when I run low and never spend more than $60." The agent monitors inventory, finds the right product, selects a payment method from the consumer's saved options, and completes the purchase. No browser. No checkout page. No card entry. The merchant gets a sale. The consumer gets their coffee. The whole thing happens without a single human click.

This is not science fiction. In March 2026, Banco Santander and Mastercard announced the completion of what they described as Europe's first live end-to-end payment executed by an AI agent on real production payment infrastructure.^[2] Visa reported that hundreds of agent-initiated transactions had already been completed in collaboration with its partners by late 2025.^[3]

For merchants, the question is no longer whether agentic payments are coming. It is whether you will be positioned to receive them when they do.

The Scale of the Opportunity

Numbers like this deserve a moment of context. McKinsey projects that agentic commerce could drive between $3 trillion and $5 trillion in global consumer transactions by 2030.^[4] The consulting firm Bain projects $7 trillion in embedded payments volume for 2026 across all payment channels, with agentic flows representing the fastest-growing segment within that total.^[5]

Visa's own research found that 47 percent of U.S. shoppers already use AI tools for at least one shopping task, whether that is price comparisons, product recommendations, or deal hunting.^[3] And AI-driven shopping traffic to U.S. retail sites surged by 4,700 percent over the past year, leaving merchants scrambling to figure out which AI agents are legitimate customers and which are bots running card testing or fraud schemes.

PaymentsJournal's analysis from Javelin Strategy and Research identifies embedded payments as one of the three defining merchant payment trends for 2026, noting that technology providers are increasingly bundling financial services directly into their software platforms and that small and medium-sized businesses spend an average of 18 hours per week on banking and financial matters, much of it reconciling data from multiple systems.^[6] Agentic commerce is what happens when that reconciliation starts happening automatically.

For businesses that serve consumers online, whether that is healthcare, insurance, property management, utilities, or retail, this is not a distant technology story. It is a 2026 infrastructure decision.

The Protocols: Visa, Mastercard, and Google Are Racing to Set the Standard

One of the most interesting dynamics in agentic payments right now is that the industry's biggest players are simultaneously competing to set the standards and cooperating to make sure those standards work together. The parallel to the early days of e-commerce, when merchants had to decide which security and checkout standards to support, is very intentional.

Visa Intelligent Commerce and the Trusted Agent Protocol

Visa launched its Trusted Agent Protocol in October 2025 as the security backbone for agent-initiated transactions. The way it works is straightforward. Each AI agent gets a cryptographic identity. When the agent initiates a transaction, it presents that credential alongside the payment. The merchant uses Visa's verification endpoints to confirm that the agent is who it claims to be and that the consumer actually authorized the purchase within the stated limits.^[7]

Visa has signed up AI platform partners including Anthropic, Microsoft, Mistral, OpenAI, and Perplexity. On the commerce side, Nuvei, Adyen, and others have committed to supporting it. Visa has also aligned its protocol with OpenAI's Agentic Commerce Protocol to ensure interoperability across platforms.

Visa's chief digital officer has been public about the ambition here. The goal is to make agentic commerce an extension of the same trust consumers already have in digital payments today, right down to the token-based credentials that power digital wallets. In their framing, an agentic transaction should feel as secure and familiar as Apple Pay, just without a human initiating it.

Mastercard Agent Pay

Mastercard's approach to agentic commerce centers on two technical tools. The first is what they call Agentic Tokens: 16-digit tokens tied to a real consumer card but scoped specifically to agent-initiated flows. The second is Verifiable Intent, a cryptographic artifact that travels with each transaction to prove that the AI agent acted within the limits the consumer actually authorized.^[2]

Mastercard CEO Michael Miebach told analysts in October 2025 that Mastercard's first agentic transaction had taken place on the network that quarter and that the company intended to be "at the center" of agentic commerce going forward. In March 2026, Mastercard and Google co-developed an open standard specifically for making user authorization "explicit, provable, and privacy-preserving in agent-led commerce," with buy-in from Worldpay, Fiserv, Checkout.com, and Adyen.^[8]

Google's Universal Commerce Protocol

At the National Retail Federation conference in January 2026, Google announced its Universal Commerce Protocol in partnership with major retailers including Etsy, Shopify, Target, Wayfair, and Walmart. The protocol allows consumers to check out from participating merchants directly within Google's AI Mode or the Gemini app, with endorsements from Adyen, American Express, Mastercard, and Visa.^[9]

What makes Google's approach particularly significant for merchants is the distribution. Google AI Mode and Gemini together represent one of the largest consumer-facing AI surfaces in the world. A merchant that is not discoverable and transactable within that environment is effectively invisible to a fast-growing segment of AI-assisted shoppers.

The Security Challenge You Cannot Afford to Ignore

With agentic commerce comes a fundamental security problem that simply did not exist before. Merchants can no longer assume the entity completing a checkout is a human. The 4,700 percent surge in AI-driven shopping traffic has already illustrated this challenge. Businesses are struggling to tell legitimate AI shopping agents apart from malicious bots running card testing attacks, account takeover schemes, or synthetic identity fraud.

The industry's answer to this is what it calls "Know Your Agent" capabilities, the AI equivalent of Know Your Customer. Both Visa and Mastercard are building these features into their respective protocols. Visa's Trusted Agent Protocol adds a cryptographic proof of identity to every agent-initiated transaction. Mastercard's Verifiable Intent artifact serves the same function. Together, these are supposed to answer the question that every merchant's fraud system needs to be able to ask: is this agent who it says it is, and did the consumer actually authorize this purchase?

The Federal Reserve flagged in its 2025 payments symposium that rising payment fraud, including AI-enabled synthetic identity schemes and imposter scams, would persist as a serious concern through 2026 and beyond. About three-quarters of respondents to an Association for Financial Professionals survey said their companies were targets of actual or attempted fraud in the previous year.^[10]

The practical implication for merchants is this. Your existing fraud detection tools were built to identify human behavioral anomalies. Unusual mouse movement patterns. Typing speed that does not match normal users. Session durations that look suspicious. AI agents do not exhibit any of those signals. They interact with your site or your payment system programmatically, and they do it quickly and consistently. Security tools calibrated for human behavior will miss attack vectors that come from automated agents entirely.

For a deeper look at how programmatic fraud already affects merchants today, see our article on card testing fraud and how small purchases can signal large losses. The dynamics of bot-driven payment attacks are not new. What agentic commerce does is significantly expand the attack surface.

What Merchants Should Do Now

The infrastructure is being built in real time, and the standards are still converging. But waiting for full standardization before preparing is a mistake. The merchants that integrate with these protocols during the build phase, rather than scrambling to catch up after consumer adoption, will have a meaningful head start on a market McKinsey estimates at $3 to $5 trillion by 2030.

• Audit your AI discoverability. If AI agents cannot find or accurately describe your products, they cannot purchase them. Make sure your product catalog is accessible to AI crawlers, that your structured data is current, and that you are not blocking AI crawlers in your robots.txt file without a good reason. Merchants who block AI crawlers are blocking the revenue that comes through agentic channels.
• Ask your payment processor which agentic protocols they support. Ask specifically about Visa's Trusted Agent Protocol, Mastercard Agent Pay, and Google's Universal Commerce Protocol. Understanding where your processor sits in the agentic ecosystem tells you what you can support today and what will require changes later.
• Review your fraud tools for programmatic transaction patterns. Set velocity limits and behavioral baselines specifically for agent-initiated transactions. Work with your processor to make sure your fraud scoring system can handle the difference between a human shopper and an AI agent making purchases under a defined spending policy.
• Update your payment authorization language. Agent-initiated payments require explicit consumer authorization. Your checkout flow and terms of service should clearly accommodate delegated purchasing authority, both for legal clarity and for compliance with emerging card network rules around consumer consent in agentic transactions.
• Diversify your payment acceptance. Payments Dive research notes that payment companies will need to "jockey for position in payment suggestions offered by agentic shopping tools," similar to how merchants compete for search ranking. Merchants that offer more payment options, including ACH, digital wallets, and BNPL, are more likely to appear favorably in AI agent payment recommendations.^[11]
• Check your API documentation. AI agents interact with merchant systems through APIs and structured data, not through visual interfaces. If your API documentation and product data are incomplete or inconsistent, you are invisible to agents that rely on machine-readable information to make purchasing decisions.

An Honest Timeline: What Is Real and What Is Still Hype

It is worth being clear-eyed here. Agentic commerce has been talked about with a level of enthusiasm that sometimes outpaces the reality. OpenAI launched a product called ChatGPT Instant Checkout in 2025 and quietly shut it down in March 2026 after it underperformed. OpenAI's commerce strategy has since shifted toward ChatGPT Apps, where merchants redirect customers to their own checkout pages rather than completing transactions inside the chat interface.

PaymentsJournal's analysis from Javelin Strategy and Research put it plainly in late 2025: agentic commerce "has not had much impact yet" except in back-office AI applications, where rules-based automation is advancing fastest. The same report identified mainstream agentic consumer purchasing at scale as more likely a 2027 or 2028 story than a 2026 reality.^[6]

Payments Dive reported on the broad gap that still exists between the vision of autonomous bot shopping and the practical reality of what AI agents can actually complete today.^[11] The protocols are real. The partnerships are real. The transactions being completed in controlled environments are real. But the consumer behavior shift that would make agentic commerce a primary sales channel for most merchants is still a few years away from mainstream.

That said, "not yet mainstream" is not the same as "not worth preparing for." Every major payment network and every major card brand is building this infrastructure now. The merchants that make the right infrastructure decisions during the build phase, rather than scrambling to catch up after adoption, will have a structural advantage when it does scale.

What This Means for Your Payment Infrastructure Today

None of the preparation steps above require a ground-up overhaul of your payment stack. What they do require is a conversation with your payment processor about where things are headed and a clear-eyed assessment of where your current setup has gaps.

The merchants who are best positioned for agentic commerce are not necessarily the ones with the most sophisticated technology. They are the ones whose payment infrastructure is flexible, whose payment methods are diverse, and whose fraud tools are calibrated for the reality that not every purchase will be initiated by a human. That is a description of good payment hygiene in 2026 regardless of agentic commerce, which is exactly the point.

A few specific things to look at with your processor:

Settlement speed. Agentic transactions may come in at higher velocity and at unusual times compared to human shopping patterns. Make sure your settlement and reconciliation processes can handle that without creating accounting headaches.

Payment method breadth. AI agents in agentic shopping systems are going to recommend payment methods based on the consumer's history and preferences. Merchants who accept a narrow range of payment types are going to miss transactions that go to competitors who accept the consumer's preferred method. Understanding your payment model options is the starting point for that conversation.

API and integration quality. The merchants who will participate most fully in agentic commerce are those whose systems can be accessed programmatically. If your payment system lives behind a clunky interface with no API access, that is a gap worth addressing now. Developer-friendly payment infrastructure is not a nice-to-have in an agentic commerce world. It is a prerequisite.

Compliance posture. Agentic transactions will come with new card network rules around consumer consent and agent authentication. Your PCI compliance posture and your authorization policies need to be reviewed in light of what those rules will require. Work with your processor to understand what is coming before it becomes a compliance issue.

Frequently Asked Questions

Sources and Citations