Agents join the payment flow

Agents can already browse, recommend, and even negotiate, but completing a purchase has remained a wall. Machine-payment protocols are still working toward adoption, so agents that need to buy things today are stuck with payment options designed for humans. Stripe is addressing that gap with Link’s wallet for agents, built on its new Issuing for agents infrastructure.

The wallet gives agents programmatic access to Link, letting them request a one-time-use card or a Shared Payment Token (SPT) backed by cards and bank accounts already stored in the wallet. The agent never touches raw payment credentials. Instead, it issues a spend request—with transaction context like merchant, amount, and purpose—that the wallet owner reviews and approves on the web or in the Link iOS and Android apps before any credential is released.

That review step is the key to making personal AI agents (Stripe cites OpenClaw as an example) safe to let loose. For developers and businesses building consumer-facing agents, the wallet removes the burden of building payment infrastructure from the ground up. Link abstracts the underlying payment options, handles fund flow complexity, and connects agents to Link’s base of over 200 million consumers.

How a spend request works

Enabling an agent begins with a standard OAuth flow, where the consumer grants the agent access to their Link wallet. Once access is granted, the agent can create a spend request for either a one-time-use card or an SPT. The credential is scoped with controls such as amount, currency, and merchant. Here’s what the request looks like from the command line:

link-cli spend-request create \ payment-method-id csmrpd_12345 \ merchant-name "Powdur" \ merchant-url "https://powdur.com" \ amount 3500 \ context "Purchasing the Powdur Glow Renewal Vitamin C Serum as a gift for $35." \ request-approval

  

    ~

The consumer then receives a notification and approves the request in Link. Today, human review is required for every request. Stripe says it plans to add controls for setting spending limits and letting agents act without approval in specific situations.

After approval, Link returns the one-time-use card or SPT to the agent to complete the transaction. The consumer can track agent spending and manage connected agents within Link.

Issuing for agents

For businesses that want more control than Link’s wallet offers, Issuing for agents exposes the full set of Issuing APIs. This allows developers to build custom agentic wallets, design their own onboarding and spending-controls experiences, and define permissions at the card level. They also get access to fraud controls at transaction authorization, plus real-time and historical view of card activity.

Underlying this are Infrastructure primitives such as single-use virtual cards, fund storage, and transaction monitoring. Stripe identifies several categories of use cases:

  • Developers automating their own business spend via agents, with programmatic workflows and recurring purchases.
  • Fintech providers embedding agent-issued cards for real-time expense management and reconciliation.
  • Vertical SaaS platforms issuing agent cards to SMB customers, letting agents automate spend under the platform’s brand.
  • Marketplaces issuing cards to sellers so their agents can handle supplier payments, logistics, and fulfillment purchases.

Agentic tokens, stablecoins, and other payment types are planned as future additions to both the Link wallet and Issuing for agents.