Visa Lets AI Agent Payments Start Clicking Buy for You

On Thursday, June 18, Alchemy moved AgentCard onto Visa Intelligent Commerce, a timing signal that matters because AI agents are being pushed from recommendation engines into actual payment actors.

The integration gives Alchemy’s artificial intelligence-powered agent card access to Visa’s agentic commerce infrastructure, with the goal of giving AI agents a “complete identity and payment stack” for online purchases made on a consumer’s behalf, according to PYMNTS.

That’s the real story behind Alchemy AgentCard Visa integration. The hard part was never getting an AI model to suggest groceries, flights, or subscriptions. The hard part is making banks, merchants, card networks, developers, and consumers trust the agent when it actually clicks buy.

June 18: Alchemy AgentCard Visa link turns AI agents into payment actors

Alchemy says AgentCard can work with AI agents built on models from any provider, including OpenAI and Anthropic. The release framed the use cases plainly: book a vacation, order groceries, or renew a subscription “without the consumer ever touching a checkout screen.”

That language matters. It shifts the agent from assistant to delegate.

“Every major computing shift has produced a new kind of economic actor,” Alchemy Co-Founder and CEO Nikil Viswanathan said in the release. “The internet created online businesses. Mobile created the app economy. AI agents are next, and they need to be able to access the global economy, and AgentCard is how that starts.”

Alchemy brings blockchain infrastructure experience around wallets, authentication, and programmable systems. Visa brings the mainstream payment path. Together, they’re trying to make AI agent payments look less like a bot scraping a checkout page and more like an authorized transaction with identity, credentials, and controls attached.

For related context on the same product lane, see Alchemy AgentCard Cracks Visa Payments for AI Shoppers.

Two months after AgentPay, the stack moves from interoperability to execution

The June 18 integration came a little more than two months after Alchemy introduced AgentPay, which PYMNTS described as a tool designed to help AI payment systems work together.

AgentCard appears to move one layer closer to consumer-facing execution. Setup takes developers less than a minute, according to the release, and gives an agent several components through a single API:

• Visa payment token: The credential used to transact through Visa-issued tokens.
• Dedicated email address: A way for the agent to receive confirmations or sign up for services.
• Phone number: A credential for verification flows.
• Cryptocurrency wallet: A programmable wallet layer tied to Alchemy’s infrastructure roots.

This is not the same as a chatbot storing a card number. The pitch is controlled authorization, identity verification, transaction routing, and payment execution.

CoinDesk reported that AgentCard’s routing layer selects the best available payment mechanism for each transaction and falls back to single-use tokens where agent-native protocols are not supported, according to CoinDesk. That detail is important because it shows how Alchemy is trying to bridge a messy transition period, where some merchants may support agent-native flows and others may still need conventional tokenized card payments.

The measurable pieces: under-one-minute setup, Visa tokens, email, phone, wallet

The announcement does not give transaction volume, user numbers, or revenue projections. That limits any hard market read.

But it does give enough operational detail to see the intended architecture. The Alchemy AgentCard Visa stack gives an AI agent identity tools and payment tools in one package. The agent can transact using Visa-issued tokens while “preserving rewards, credit lines and card benefits,” according to the release, without requiring new accounts or credentials.

That is a big design choice. Alchemy and Visa are not asking consumers to abandon existing card relationships. They’re trying to make the agent sit on top of them.

Visa’s Tanner Riche, vice president for growth products and partnerships, framed the need around trust and transaction readiness:

“As AI agents take on a more active role in commerce, they need to demonstrate trusted identity and seamless ways to transact. Visa is working with partners across the ecosystem so developers can bring these experiences to market on our Visa Intelligent Commerce infrastructure designed to support agentic commerce in a manner that is secure, reliable and ready to scale.”

XOOMAR analysis: That wording points to Visa’s commercial incentive. If AI agents become shopping intermediaries, payment networks will want their rails embedded before tech platforms or closed marketplaces dominate the purchase moment.

The same pressure exists across checkout economics. For a separate angle on how payment rails compete at checkout, see Open Banking Payment Providers Slash Checkout Card Fees.

Different players will judge AgentCard by different failure modes

Consumers will judge AI agent payments by control. Can they set spending caps? Can they limit categories? Can they see what the agent bought, when, and why? The source material does not say how AgentCard captures consent or handles cancellation and refund workflows.

Banks and issuers will care about fraud patterns. PYMNTS reported June 10 that AI agents face a security problem because the internet cannot determine if an agent is authorized, who gave the authorization, and what limits it has.

“A person logs in once and leaves a behavioral trail fraud detection can read,” the PYMNTS report said. “Agents act continuously, spawn other agents and carry permissions that spread in ways no one mapped when access was first granted.”

Merchants will judge the stack by conversion and support burden. If an agent can complete routine purchases cleanly, checkout friction falls. If consumers later say the agent misunderstood instructions, merchants may face messier service interactions. The announcement does not say whether merchants will be able to identify an agent-led transaction at checkout.

Developers will care about integration complexity. Alchemy’s less-than-a-minute setup claim is aimed directly at them. A single API that provisions credentials, payment access, and a wallet reduces build time, but reliability and policy controls will decide whether developers trust it in production.

From saved cards to delegated agents, this is a harder handoff

Online commerce has been moving toward delegated payment for years: saved cards, mobile wallets, subscription billing, virtual cards, and tokenized credentials all trained users to let software handle parts of checkout.

Agentic payments change the center of gravity. A subscription follows preset rules. An AI agent may interpret intent, compare options, pick a vendor, and execute the transaction in real time.

That makes identity more important. The agent needs to prove that it is authorized, not merely automated. It also needs payment credentials that don’t expose the consumer’s underlying card details or require new merchant accounts each time.

Alchemy’s blockchain background shows up in the wallet component. Visa’s role keeps the product closer to mainstream payments than pure crypto commerce. That combination is the point: programmable identity and payment behavior, routed through a card network consumers and merchants already know.

The next decision point: narrow permissions before broad autonomous checkout

The first durable use cases for Alchemy AgentCard Visa are likely to be bounded tasks: grocery reorders, subscription renewals, routine travel bookings, bill payments, or low-risk digital purchases. That is XOOMAR analysis, not a company forecast, and it follows from the trust problem PYMNTS highlighted.

The evidence to watch is practical. Do developers adopt the API beyond demos? Do merchants recognize and accept agent-led transactions cleanly? Do consumers get clear permission tools? Do issuers see tokenized agent payments as manageable risk?

The agent that can pay safely will be more commercially valuable than the agent that only answers questions. But the pace will be set by trust, liability, and user control, not by model capability alone.

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

• The partnership moves AI agents closer to making real purchases, not just recommendations.
• Visa’s involvement could help make agent-led transactions more trusted by banks, merchants, and consumers.
• Alchemy’s AgentCard aims to give AI agents the identity, credentials, and controls needed for online commerce.