Something quietly significant happened in March 2026. An AI agent, acting on behalf of a Banco Santander customer, used a Visa card to buy a book. No human approved the transaction. No one entered a PIN. The AI identified the item, selected the payment method, completed authorization, and settled the purchase, from start to finish, without touching a screen.

Visa calls this “agentic commerce,” and they just built the infrastructure to make it mainstream.

What Visa Launched

On March 17, 2026, Visa announced its Agentic Ready program, a structured framework that allows issuer banks to connect AI agents to Visa’s payment network in a controlled, production-grade environment.

The launch included 21 European issuer banks as founding partners, among them Barclays, HSBC UK, Revolut, Commerzbank, Raiffeisen Bank International, and Banco Santander. The program uses Visa’s existing tokenization infrastructure and biometric authentication, not new payment rails, but a secure bridge that lets AI agents initiate and complete transactions using existing consumer credentials.

The Santander demo validated the model in practice: an AI agent purchased a book through Santander’s platform, completing every step of the payment lifecycle autonomously. The transaction was real. The settlement was real. The human was not involved.

Visa projects that millions of consumers will be using AI agents to make purchases on their behalf by the 2026 holiday season. That is not a decade-out forecast. It is eight months away.

Why This Matters More Than It Looks

This is easy to file under “interesting fintech news” and move on. That would be a mistake.

What Visa has done is solve one of the practical blockers to agentic AI adoption in business: payments. An AI agent that can research, recommend, and draft but not transact is an assistant. An AI agent that can complete a purchase is an operator.

Think about what becomes possible once AI agents can spend money:

• A procurement agent that identifies a supplier, negotiates terms, and issues a purchase order without a human in the loop for routine orders under a certain threshold
• A travel agent that books flights, hotels, and ground transport the moment a meeting is confirmed
• A reorder agent that monitors inventory levels and places supplier orders automatically when stock drops below par
• A media buying agent that allocates ad spend across platforms in real time based on performance data

These are not theoretical. These are the kinds of workflows that Omni Ops clients are already building, and the missing piece has consistently been secure, automated payment execution. Visa just built that piece.

The Security Architecture

The most important thing about how Visa structured this: they did not build new payment infrastructure. They built a governance layer on top of what already exists.

Agentic Ready uses:

• Tokenization: AI agents never handle real card numbers. They work with single-use tokens that expire and are scoped to specific merchants or transaction types.
• Biometric authentication: The consumer authorizes the AI agent to act on their behalf once, with biometric verification. Subsequent transactions by that agent do not require re-authentication, but remain auditable.
• Spending controls: Banks and cardholders can set limits covering merchant categories, geographic restrictions, maximum transaction values, or time-of-day rules.

This matters because the biggest objection to agentic AI in finance has always been accountability. Who is responsible when an AI makes a bad purchase? Visa’s architecture provides a clear answer: the transaction is traceable, reversible through normal dispute channels, and the agent’s authority is explicitly bounded by rules set at enrollment.

The Gartner Signal

The research firm Gartner added context that should make every procurement and operations leader pay attention: they project that 90% of B2B buying will be AI agent-intermediated by 2028.

That is not 90% of B2C impulse purchases. That is 90% of enterprise procurement, supplier negotiations, and business-to-business commercial activity, handled by agents acting on behalf of companies, not individual consumers.

If that projection is anywhere close to accurate, every business that has not thought about how their operations interact with AI agents, both their own and those of their customers and suppliers, is already behind.

What This Means for Business

For business owners and operations leaders, the Visa Agentic Ready announcement is a signal, not just a story.

If you sell to other businesses: Your customers will increasingly send AI agents to research, evaluate, and purchase your products. Those agents need to be able to transact with you programmatically. If your purchase process requires a phone call, a PDF form, or a meeting, you are invisible to an agent-driven buyer.

If you run operations: The friction of routine purchasing, reorders, SaaS renewals, travel bookings, advertising spend, is now solvable with automation that goes all the way through to payment. That friction is currently absorbing human attention that could be spent on higher-value work.

If you are evaluating AI agents for your business: Payment capability is no longer a blocker. The infrastructure is live, major banks are participating, and the governance model is sound. The question is no longer whether AI agents can transact. The question is what transactions you want to automate first.

If you’re building with Claude or Codex right now, grab the free Working With Claude field guide. Thirty-two pages on the full ecosystem, Claude Code in depth, and how to roll agents out properly. Get the free guide.

The book Banco Santander’s AI agent bought in March 2026 is a small purchase. But it marks the beginning of something much larger.