On March 2, 2026, Santander and Mastercard completed Europe’s first live end-to-end payment executed by an AI agent. Not a demo. Not a sandbox transaction. A real payment processed through Santander’s live banking infrastructure, inside a regulated framework, initiated and completed by an autonomous AI system. Three days later, Mastercard announced Verifiable Intent, an open-source framework that creates tamper-resistant records of what an AI agent was told to do and what it actually did.
Meanwhile, Visa has enrolled 21 European bank partners in its Agentic Ready program, including Commerzbank, DZ Bank, Raiffeisen Bank International, Barclays, HSBC, Revolut, and Banco Santander. Visa’s bet: get the banks testing agent-initiated transactions in production-grade environments before Mastercard scales its own pilots.
Two payment networks. Two fundamentally different strategies for the same prize: becoming the default rails that AI agents use when they spend your money.
Why Payment Networks Are the Bottleneck
Agentic commerce already has its shopping layer. Google launched the Universal Commerce Protocol. Stripe shipped its Agentic Commerce Suite. Shopify, Amazon, and dozens of startups are building agent-friendly storefronts. An AI agent can browse products, compare prices, and pick the right item. The hard part was never finding the product. The hard part is paying for it.
When a human buys something online, the payment flow depends on identity verification that assumes a human is present: passwords, biometrics, 3D Secure challenges, CAPTCHAs. An AI agent cannot complete any of those. It also cannot prove it has authorization to spend your money, that it is acting within the limits you set, or that the transaction it initiates matches the instruction you gave it three hours ago.
That is the gap Visa and Mastercard are racing to fill. Not better shopping. Better paying.
According to Visa’s own data, 47% of U.S. shoppers already use AI tools for shopping tasks like price comparison and product recommendations. Visa predicts millions of consumers will use AI agents to complete purchases by the 2026 holiday season. But “complete purchases” requires payment infrastructure that trusts a machine to act for a human, and that infrastructure did not exist until this year.
Visa’s Strategy: Trusted Agent Protocol and Network Scale
Visa is approaching agentic commerce as an infrastructure play. Its core strategy rests on three pillars: identity, ecosystem scale, and interoperability.
Trusted Agent Protocol
In October 2025, Visa introduced the Trusted Agent Protocol (TAP) alongside more than 10 launch partners. TAP is an open framework built on existing web infrastructure that solves a specific problem: how does a merchant know whether the entity trying to buy something is a legitimate AI agent acting on behalf of a real customer, or a bot running a credential-stuffing attack?
TAP works by establishing a chain of trust. The agent registers with Visa, gets verified, and receives credentials that merchants can check in real time. Akamai provides the behavioral intelligence layer, distinguishing agent traffic from bot traffic based on patterns, not just credentials. Visa has aligned TAP with OpenAI’s Agentic Commerce Protocol and Coinbase’s x402 standard, signaling it wants TAP to work across agent platforms, not just within Visa’s own ecosystem.
Visa Intelligent Commerce
The broader initiative, Visa Intelligent Commerce (VIC), is Visa’s umbrella program for agentic payments. The numbers tell the story of Visa’s approach: 100+ partners working across the ecosystem, 30+ actively building in the VIC sandbox, and 20+ agent developers integrating directly. Visa has completed hundreds of real-world agent-initiated transactions across multiple geographies.
In the U.S., Visa demonstrated concrete use cases: Skyfire built a Consumer Reports integration where an agent researches products, selects the best option, and pays via Visa. Nekuda enables fashion purchases through AI agents using Gensmo and Rye checkout. Ramp integrated agent-driven B2B purchasing with automatic cashback.
Agentic Ready Program in Europe
For European banks, Visa launched the Agentic Ready program with 21 issuing partners in the first phase. The list reads like a who’s who of European banking: Commerzbank, DZ Bank, Raiffeisen Bank International, Barclays, HSBC UK, Revolut, Banco Santander, Nationwide Building Society, and Nexi Group. Participating banks run agent-initiated transactions in production-grade testing environments alongside selected merchants. The goal: help issuers understand how agentic payments behave in live systems before rolling them out to customers.
Rubail Birwadker, Visa’s SVP and Head of Growth Products, put it plainly: “In 2026, AI agents won’t just assist your shopping. They will complete your purchases, powered by Visa’s global scale, standards leadership, and unparalleled commitment to secure agentic commerce.”
Mastercard’s Counter: Agent Pay and Verifiable Intent
Where Visa leads with ecosystem breadth, Mastercard leads with regulated-market proof points. Its strategy is built around two products: Agent Pay for the transaction mechanics and Verifiable Intent for the trust layer.
How Agent Pay Works
Mastercard Agent Pay, unveiled in April 2025, takes a different approach to agent authentication than Visa’s TAP. Agent Pay requires trusted AI agents to be registered and verified on the Mastercard network. Once verified, agents can initiate secure payments on behalf of users through conversational interfaces, with enhanced tokenization technology handling the payment credentials.
The difference from Visa’s approach is subtle but significant. Visa focuses on letting merchants verify agent identity at the point of sale. Mastercard focuses on pre-registering agents within the banking system itself. Visa’s model is merchant-centric. Mastercard’s is issuer-centric.
Europe’s First Live Agentic Payment
Mastercard’s strongest card right now is Santander. On March 2, 2026, Santander processed Europe’s first live end-to-end AI agent payment using Agent Pay within Santander’s regulated banking framework. This was not a sandbox experiment. The transaction went through Santander’s live payments infrastructure.
Why does this matter? Because European financial regulation is the hardest test case. PSD2 strong customer authentication requirements, DSGVO data protection rules, and national banking regulations all create constraints that a sandbox demo does not face. Completing a real agentic payment within that regulatory framework proves the concept works where it matters most.
Mastercard has also completed Australia’s first authenticated agentic transactions and New Zealand’s first with Westpac. Three continents, three “firsts.” That is Mastercard’s playbook: prove it works in regulated environments faster than Visa can scale its sandbox.
Verifiable Intent: The Cryptographic Audit Trail
On March 5, 2026, Mastercard introduced Verifiable Intent, an open-source framework that may be the most technically ambitious piece of the entire agentic payments stack. It addresses a problem neither company had publicly tackled before: what happens when an AI agent executes an instruction that was issued hours or days earlier?
Consider this scenario: you tell your agent on Monday to book the cheapest flight to Berlin for next Friday. On Wednesday, the agent finds a deal and books it. How does the airline know the agent had current authorization? How does your bank verify the purchase matches what you asked for? What happens if the agent books business class instead of economy?
Verifiable Intent solves this by linking the consumer’s identity, their specific instructions, and the transaction outcome into a single, tamper-resistant cryptographic record. It uses Selective Disclosure, sharing only the minimum information each party needs. The framework is built on standards from FIDO Alliance, EMVCo, IETF, and W3C, and it interoperates with Google’s Agent Payments Protocol (AP2) and Universal Commerce Protocol.
Pablo Fourez, Mastercard’s Chief Digital Officer, framed it precisely: “As autonomy increases, trust cannot be implied. It must be proven.”
Partners backing Verifiable Intent include Google, Fiserv, IBM, Checkout.com, Basis Theory, and Getnet. Google specifically endorsed it as “strong, interoperable trust infrastructure compatible with Agent Payments Protocol.”
The Standards War: Where the Real Fight Is
The technology works. Both companies have proven that. The real battle is over whose standards become the default.
Where They Agree
Both Visa and Mastercard agree on several fundamentals. AI agents need verified identities. Transactions need cryptographic proof of authorization. The frameworks need to be open and interoperable, not proprietary lock-ins. Both are aligning with Google’s protocols (UCP and AP2). Both are working with FIDO Alliance and EMVCo on authentication standards.
Where They Diverge
The divergence is strategic. Visa’s TAP is merchant-facing: it helps merchants distinguish real agents from bots. Mastercard’s Agent Pay is issuer-facing: it registers agents within the banking system. Visa is building the widest ecosystem first (100+ partners, 21 European banks). Mastercard is stacking regulated-market firsts (Europe, Australia, New Zealand).
For businesses designing their B2A (Business-to-Agent) strategy, this means supporting both. A merchant that only integrates with Visa’s TAP misses the Mastercard-enrolled agents, and vice versa. The winning move, at least in 2026, is implementing both protocols and letting the market decide which becomes dominant.
The Interoperability Question
The most important development might be Mastercard’s Verifiable Intent being open-source and standards-based. If Visa adopts or aligns with Verifiable Intent (or builds something compatible), the standards war resolves into a standards convergence. If they build competing intent-verification systems, merchants and banks get stuck supporting two parallel stacks.
PYMNTS Intelligence data suggests the market is ready for either outcome: 43% of CFOs expect high impact from agentic AI on their operations, with another 47% expecting moderate impact. That is 90% of financial leaders who believe this shift is real. The question is not whether agentic payments happen, but whose rails they run on.
What European Banks and Merchants Should Do Now
European businesses face a unique situation. Both Visa and Mastercard are aggressively targeting the European market, and European regulation creates both constraints and advantages.
For banks and issuers: Join both programs. Visa’s Agentic Ready program is enrolling partners now. Mastercard’s Agent Pay is live in regulated European environments. Running parallel pilots with both networks gives you data on which model works better for your customer base. Commerzbank, Raiffeisen, and Santander are all hedging by working with both networks simultaneously.
For merchants: Start with your payment processor. Fiserv has already integrated Mastercard Agent Pay into its merchant platform. Ask your PSP about Visa TAP support. If you are building direct integrations, implement both MCP and A2A protocols as the foundational layer, then add the payment-specific protocols on top.
For compliance teams: Mastercard’s Santander pilot is the template. It proves agentic payments can work within PSD2 and DSGVO constraints. Document it internally and use it to build the case for your own pilots. The regulated-first approach means compliance is a feature, not a blocker.
The 2026 holiday season is the target both networks have set for mainstream adoption. That gives European businesses roughly eight months to go from “watching” to “running production pilots.”
Frequently Asked Questions
What is the difference between Visa’s Trusted Agent Protocol and Mastercard’s Agent Pay?
Visa’s Trusted Agent Protocol (TAP) is merchant-facing: it helps merchants verify whether an AI agent is legitimate before processing a transaction. Mastercard’s Agent Pay is issuer-facing: it requires agents to register and be verified within the banking network before they can initiate payments. Both achieve agent authentication but from opposite ends of the payment chain.
Can AI agents already make real payments in Europe?
Yes. In March 2026, Santander and Mastercard completed Europe’s first live end-to-end payment executed by an AI agent using Mastercard Agent Pay within Santander’s regulated banking framework. Visa has also enrolled 21 European banks in its Agentic Ready program for production-grade testing. Both are targeting mainstream rollout by the 2026 holiday season.
What is Mastercard’s Verifiable Intent framework?
Verifiable Intent is an open-source framework announced by Mastercard on March 5, 2026. It creates tamper-resistant cryptographic records that link a consumer’s identity, their specific instructions to an AI agent, and the transaction outcome. This solves the problem of verifying that an agent had authorization when it executes a purchase hours or days after receiving the original instruction.
Which European banks support agentic commerce payments?
Visa’s Agentic Ready program includes 21 European issuing partners: Commerzbank, DZ Bank, Raiffeisen Bank International, Barclays, HSBC UK, Revolut, Banco Santander, Nationwide Building Society, and Nexi Group among others. Santander is also working with Mastercard on Agent Pay. Several banks are hedging by working with both networks.
Should merchants integrate with Visa or Mastercard for agentic payments?
Both. In 2026, neither standard has won the market. Merchants should implement foundational protocols like MCP and A2A, then add both Visa TAP and Mastercard Agent Pay support. Fiserv has already integrated Mastercard Agent Pay into its merchant platform, and Visa’s TAP is available through its developer program. Supporting both networks ensures you do not miss agent-initiated transactions from either ecosystem.