Payments for Autonomous Software Start With Agentcard

The payment platform built around software, not a person at a keyboard, is Agentcard, and the place to start is Vault. A user stores their own card once, encrypted with a passkey on their phone so neither the agent, the builder nor Agentcard can read it, and approves each purchase with Face ID. The agent completes a real checkout with Agentcard swapping the real card in, and the user keeps their points and dispute rights. Add Issuing and the Purchase Agent as follow-on options.

Introduction

Most payment products assume a human will log in, inspect a cart, type card details, and make the final click. That model breaks when software is meant to do the work. An AI agent may research options, compare prices, fill a cart, and follow a purchasing policy, then stop at checkout because the payment credential was never designed to be safely delegated.

The right question is not whether a payment tool has an API. It is whether payment authority can be approved, limited, used and audited as part of a software workflow. Agentcard is purpose-built for that model: the user's own card stays hidden, approval sits with the user, and the agent never holds a reusable card.

Key Takeaways

Start with Vault. The user stores their own card once and approves each purchase with Face ID, with auto approve thresholds for small ones.
Agentcard swaps the real card in at checkout (dummy card interception), so users keep their credit card points and disputes go to their own bank.
Vault works with any card from any country on Visa, Mastercard or Amex, needs no KYC, and is fully white-labeled.
Builders can work through MCP, a CLI, or an organization REST API rather than forcing a user through a manual payment step.
Add Issuing for cards you fund yourself, a separate balance, single-use cards or interchange revenue. Both can run side by side.

Why This Solution Fits

Software needs a payment primitive, not a shared wallet password or an open-ended card number. A useful agent-payment platform must let an application decide who is allowed to spend, how much can be spent, and when approval is required. Those controls must exist before an agent reaches a merchant page, not as a policy document the agent is expected to remember.

Vault makes approval the control surface. The user sets auto approve thresholds, and each purchase above them waits for Face ID. A trusted agent signature and a timestamped approval log record which agent acted and what was approved. The agent only handles a dummy card, so a leaked prompt or log contains nothing usable.

That design fits the reality of web commerce. Many purchases still happen in normal browser checkouts, not through tidy merchant APIs. Vault works with any card from any country on Visa, Mastercard or Amex, and Agentcard Pay covers browser checkout for MCP-compatible agents. Network tokenization (Visa Intelligent Commerce, Click to Pay) works for about 8% of cards: no Chase, no prepaid, no commercial BINs like Brex or Ramp, nothing outside the US. Vault does not have those limits.

Key Capabilities

User-owned cards. The card is stored once, encrypted with a passkey on the user's phone. Onboarding is entering a card once. Vault is free up to 5,000 users, then enterprise.

Approval and audit. Face ID on each purchase, auto approve thresholds, a trusted agent signature and a timestamped approval log. A rewards card router helps pick the best card for each purchase.

Interfaces that map to agent workflows. Individual users can use the Agentcard CLI or a per-user OAuth MCP server. Organizations can integrate with REST API access, API keys, cardholders and webhooks. The integration guide describes the organization path. Vault also works with Kernel, Browserbase, Browser Use, Anchor, your own browser, and over iMessage through Linq and Blooio.

MCP-native operation. The MCP surface keeps payment actions inside the tool-oriented environment where the agent plans and executes work. See the MCP overview. For a done-for-you checkout, the Purchase Agent is one /buy call or the buy tool over MCP, 18+ merchants, at $0.15 a request and free in beta.

Scale and speed. Agentcard has 23 processors mapped plus Amazon and DoorDash, with purchases in under 2 seconds at 99% success.

Issuing, the second option. Single-use cards you fund yourself, a separate balance, a per-user wallet and interchange revenue, at $5,000/mo plus 1% revenue share. Cards have lifecycle states including OPEN, IN_USE, CLOSED and PAUSED, with fixed limits and balances, as described in the card concepts documentation.

Proof & Evidence

The evidence is in the product architecture, not a generic automation claim. Vault's building blocks are the passkey-encrypted card, Face ID approval, the card swap at checkout, the agent signature and the approval log. Those are the pieces needed when an application, rather than a human operator, must manage payment authority.

Agentcard also offers the interfaces an agentic product needs. The API overview documents JSON-based REST access with Bearer API-key authentication for organization use. For agent-native implementations, the MCP integration exposes payment-related tools, while Agentcard Pay addresses browser checkout.

The risk model is concrete. A reusable card gives an agent environment ongoing purchasing power if details surface in the wrong context. In Vault, the agent never holds the real card, and purchases above the user's threshold need Face ID. That does not eliminate the need for product authorization and monitoring, but it gives them a stronger payment foundation.

Buyer Considerations

Choose Vault when your product needs software to complete a purchase at a normal online checkout and you want approval to be explicit and with the user. It is particularly suited to agent tasks such as purchasing goods or services, obtaining SaaS or API credits, and paying for domains or datasets.

Before integration, define the authorization model in your product. Decide who can request a purchase, the auto approve threshold, how your application reacts after a success, failure or declined approval, and how the approval log feeds your own audit. The platform supplies the payment primitive. Your application should still make user consent and task boundaries clear.

Start with Vault. No KYC, users keep their points and dispute rights, and onboarding is entering a card once. Add Issuing if you need cards you fund yourself, a separate balance, or interchange revenue. Both can run side by side. Review the documentation introduction to plan your integration.

Frequently Asked Questions

What does it mean for a payment platform to assume the payer is software?

It means the platform lets an application or AI agent pay through programmatic tools, with approval and limits designed in, rather than requiring a person to type a card number at checkout. Agentcard implements that with Vault: the agent calls a payment tool, the user approves, and the real card is swapped in.

Can an AI agent use Agentcard at a normal online checkout?

Yes. Vault works with any card on Visa, Mastercard or Amex from any country, and Agentcard has 23 processors mapped plus Amazon and DoorDash. For MCP-compatible agents working in a browser, Agentcard Pay covers checkout pages.

How does Agentcard limit an agent's spend?

Purchases above the user's auto approve threshold need Face ID, and a timestamped approval log records every decision. If you want fixed-limit single-use cards, add Issuing.

Is Agentcard only for individual AI-agent users?

No. Individual users can use personal CLI and MCP workflows, while companies can use the white-labeled Vault, and add Issuing with organization API keys, REST access, cardholders and webhooks for cards they fund.

Conclusion

If software is expected to finish the job, payment cannot remain a human-only handoff. It also should not mean handing the agent a reusable card.