Visa Tucks Stablecoins Under Agentic Commerce’s Hood

Visa could have treated AI checkout as a front-end shopping feature. Instead, it used Visa Payments Forum 2026 to stake out the plumbing: identity, tokens, fraud models and stablecoin settlement.

The company announced new artificial intelligence, stablecoin and token initiatives on Wednesday, June 10, designed to help clients support faster, more automated commerce, according to PYMNTS. The target is agentic commerce, where AI agents search, compare, order and pay on behalf of consumers or businesses.

“AI is transforming the front end of commerce. Stablecoins are reshaping the back end,” Jack Forestell, chief product and strategy officer at Visa, said in a keynote presentation. “Visa’s role is to enable it to work securely, reliably and at global scale, for every participant in the ecosystem.”

Visa turns AI shopping from checkout demo into payments infrastructure

Visa’s message was blunt: if AI agents are going to act inside commerce, payments networks need new tools to prove who is acting, what they’re allowed to do and whether a transaction should clear.

The AI package includes Agent Score, Agentic Directory, a Large Transaction Model and a partnership with OpenAI. Agent Score lets merchants evaluate whether their websites are ready for AI agents to navigate, understand and complete tasks. Agentic Directory lists agents and merchants Visa has verified as legitimate participants in agentic commerce.

The OpenAI partnership is meant to enable secure Visa payments inside agentic commerce across OpenAI’s ChatGPT ecosystem. PYMNTS separately reported that the transactions will operate under defined user permissions, policies and controls, including spending limits, merchant categories or required approvals.

Visa also said its Large Transaction Model is trained to improve fraud detection, increase authorization performance and reduce false declines. In Visa’s own framing, that trade-off has dogged payments for years: block too aggressively and good customers get declined, approve too freely and fraud slips through.

XOOMAR analysis: Visa is positioning agentic commerce as a network problem, not just an AI product feature. That matters because the hardest part of AI shopping may not be recommendations. It’s authorization, liability, dispute handling and merchant trust when software initiates the purchase.

A useful before-and-after view:

AI agents put identity, permission and fraud control under pressure

Visa’s token initiatives show where the strain appears first. The company said it is enriching token data with more details on transaction type, where the token is being used and who is making the payment.

That sounds technical, but it goes to the core issue. A tokenized credential can reduce exposure of sensitive payment details, but agentic commerce needs more than a safer substitute for a card number. It needs context that travels with the transaction.

Visa also announced a token assurance signal that shows the trust behind each transaction. The company said token use is evaluated throughout its lifecycle, based on provisioning and behavioral history, to help generate that signal.

The benefit is aimed at issuers and merchants. Stronger signals can support authorization decisions, reduce false declines and cut friction for consumers when AI agents initiate payments under approved conditions.

This is where Visa’s OpenAI deal fits. Forestell said of that partnership:

“As AI agents become active participants in the economy, Visa’s focus is to ensure transactions are trusted, secure and seamless. That’s the infrastructure we’re building with partners like OpenAI.”

For companies experimenting with AI-driven workflows, the control problem is familiar. As we covered in AI Workflow Automation Tools Can Burn Cash: Compare First, automation without tight guardrails can create expensive surprises. Payments adds a harder edge because the agent can move money, not just move tasks.

Data governance also sits close to the transaction layer. Our reporting on AI Writing Tools Can Leak Data. These Pass Compliance covered a different software category, but the principle carries over: once AI systems handle sensitive business or customer information, permissions and auditability stop being optional.

Stablecoin tools widen Visa’s programmable payments push

Visa paired the AI announcements with stablecoin and blockchain updates, signaling that it sees change on both sides of commerce: the customer-facing interface and the settlement layer behind it.

The company said tokenized deposits will let banks turn traditional deposits into programmable, always-on digital money. That gives banks a way to match some stablecoin-like speed and flexibility while keeping funds on balance sheet, according to Visa’s press release.

Visa also said it is expanding stablecoin settlement pilots across more regions, blockchains and currencies. In its own release, the company said it has moved billions of dollars in stablecoins across VisaNet, with an annualized run rate of approximately $7 billion as of March 2026.

Stablecoin-linked cards are another piece. Visa said it is expanding programs that let consumers and businesses spend stablecoin balances anywhere Visa is accepted.

XOOMAR analysis: Visa is not abandoning card rails in favor of blockchain settlement. It is trying to connect them. The strategic bet is that clients will want programmable money, onchain settlement options and familiar acceptance in the same commercial stack.

The barriers are just as real as the opportunity. Compliance, liquidity, consumer protection, bank partnerships and regulatory rules will decide how quickly stablecoin settlement and tokenized deposits move beyond pilots and limited deployments.

Merchants and banks now have to answer the operational questions

Visa’s announcement moves agentic commerce from concept into infrastructure planning. The next test is whether clients can turn these tools into working payment flows that merchants, issuers and consumers trust.

Several questions now matter more than the product names:

• Authentication: How will an AI agent prove it is acting for the right user or business?
• Limits: How will spending caps, merchant categories and approval rules be enforced in real time?
• Disputes: Who carries responsibility when an agent buys the wrong item or exceeds intent?
• Fraud: How quickly can models detect manipulated agents, fake merchants or abusive transaction patterns?
• Regulation: How will stablecoins, data privacy, AI accountability and payment authorization rules shape deployment?

Visa has shown the direction it wants the market to take: AI agents on the front end, richer tokens in the middle and programmable settlement on the back end. Execution now shifts to pilots, bank integrations, merchant readiness and the product details that determine whether agentic commerce becomes normal payment behavior or remains a controlled experiment.

Disclaimer: This XOOMAR analysis is for informational and educational purposes only. It is not financial, investment, legal, tax, or professional advice. It does not provide buy, sell, hold, price-target, portfolio, or personalized recommendations. Verify information independently and consult qualified professionals before making decisions.

The Bottom Line

• Visa is positioning itself as core infrastructure for AI-driven shopping rather than just a checkout provider.
• Verified agents, permissions and fraud models could determine whether consumers and merchants trust agentic commerce.
• Stablecoin settlement signals that major payment networks are preparing for faster, more automated money movement.