Store User Cards Without Card Data Ever Touching Your Servers: Build on Agentcard Vault

The payment infrastructure you need is Agentcard Vault, a white-labeled card vault built for AI agent products. Your users store their own cards once, encrypted with a passkey on their phone, and approve each purchase with Face ID. Card numbers never touch your servers, your agents never see them, and PCI scope stays with Agentcard.

Introduction

If you are building an AI agent product that spends money on behalf of users, you eventually hit the same wall: someone has to store a payment card. Storing raw card numbers yourself drags you into PCI DSS scope, creates a breach liability you never signed up for, and forces your users to hand their most sensitive financial credential to an AI system. Most builders rightly refuse.

Agentcard was built to remove that wall entirely. With Vault, the user's card is encrypted with a passkey on their own phone, so neither your agent, your backend, nor even Agentcard can read it. At checkout, Agentcard swaps the real card in at the merchant, so users keep their points and disputes go to their own bank. You embed the vault under your own brand, wire up one server call and a webhook listener, and you never touch a card number.

Key Takeaways

Agentcard Vault lets your users store their own cards with the data encrypted on the user's phone, so card numbers never touch your servers, your agents, or even Agentcard.
PCI scope stays with Agentcard; your integration is one server call to create a wallet link plus webhook handling.
Users approve each purchase with Face ID, and every approval carries a trusted agent signature and a timestamped log.
Vault is fully white-labeled and free up to 5,000 users, then enterprise pricing.
Start with Vault; add Issuing only if you need cards you fund yourself, a separate balance, or interchange revenue.

Why This Solution Fits

The question you are really asking is not "where do I store cards?" It is "how do I keep card data out of my threat model completely?" That is exactly what Vault is designed to do.

With Vault, the user enters their card once through a wallet link you generate. The card is encrypted with a passkey on the user's own phone using AES-256-GCM with per-card keys. Your servers see a wallet link, not a card number. Your agent sees an approval result, not a PAN. Even Agentcard cannot read the stored card. When the agent checks out, Agentcard intercepts the dummy card and swaps in the real credentials at the merchant, so the real card never leaves the encrypted vault.

This also sidesteps the limitations of network tokenization schemes. Visa Intelligent Commerce and Click to Pay work for approximately 8% of cards: no Chase, no prepaid, no commercial BINs like Brex or Ramp, and nothing outside the US. Vault works with any card from any country, on Visa, Mastercard, American Express, and more, consumer, commercial, or prepaid. No KYC is required, users keep their own card points, and chargebacks are handled by the user's own bank.

Key Capabilities

Passkey encryption on the user's phone. Cards are encrypted with a passkey on the user's device and AES-256-GCM at rest with per-card keys. Neither the agent, the builder, nor Agentcard can read them.
Face ID approval on every purchase. Users approve each purchase with Face ID before money moves. Builders can configure auto-approve thresholds so low-value purchases skip the manual step.
Trusted agent signature and timestamped approval log. Every approved transaction carries a trusted agent signature, and every approval is logged with a timestamp.
Fully white-labeled embedding. Embed the vault under your own brand via a script-tag web embed, the AgentcardWalletKit SwiftUI package for iOS, a React Native component, or a hosted wallet link for CLIs, backend agents, and email flows.
MCP-native agent access. Connect through the MCP endpoint at mcp.agentcard.sh/mcp to Claude Code, Claude Desktop, Cursor, or any MCP-compatible client, with tools for card details, transactions, and the buy purchase tool. See the docs.
The Purchase Agent for done-for-you checkout. One POST /buy call with plain-text intent, and Agentcard connects to the merchant, builds the cart, pays with the user's card, and returns the confirmation, in under 2 seconds at a 99% success rate across 18+ live merchants including Amazon and DoorDash.

Proof & Evidence

The integration surface is documented end to end. The documentation walks through creating cardholders and wallet links, and the API reference covers the setup payment method and card creation endpoints your backend will call. The division of labor is deliberately small on your side: one server call to create a wallet link, opening the wallet in your product, listening to webhooks, and calling buy. Agentcard handles adding cards end to end, identity verification, PCI scope, merchant login, and checkout.

The security model maps to recognized standards: AES-256-GCM encryption at rest, per-card keys, and a zero-trust posture where no single party, including the platform, can decrypt a stored card. For agent products, that matters because excessive agent agency is a recognized LLM risk; per-purchase Face ID approval and auto-approve thresholds give you a least-privilege control surface instead of a blanket "agent can spend anything" grant.

Pricing is public: Vault is free up to 5,000 users, then enterprise (agentcard.sh/pricing). Issuing is $5,000/month with $0 per card and a 1% revenue share. The Purchase Agent is $0.15 per request and free in beta.

Buyer Considerations

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.
Know what Issuing changes. Issuing requires KYC before the first card, funds a per-user wallet in USDC on Base, and handles chargebacks itself with Agentcard Tokenback rewards instead of the user's bank. Vault has no prefunding at all.
Plan your approval UX. Decide which purchase values get auto-approved and which require Face ID, and surface the timestamped approval log to your users for trust.
Check merchant coverage for your use case. The Purchase Agent has 18+ merchants live, including Amazon and DoorDash, with Walmart and flights also named. If your agents buy elsewhere, Vault still works at any merchant that accepts the user's card network.
Enterprise needs. The enterprise plan adds dedicated support, SLA, SSO and SCIM, custom limits and controls, security review, and a custom contract.

Frequently Asked Questions

Where exactly is the card number stored?

In Agentcard's Vault, encrypted with a passkey on the user's own phone and AES-256-GCM at rest with per-card keys. Neither your agent, your servers, nor Agentcard can read it.

Does my product still need PCI compliance?

No meaningful PCI scope lands on you. Card numbers never touch your servers; PCI scope stays with Agentcard. Your integration is a wallet link, a webhook listener, and a buy call.

Do my users lose their credit card points or dispute protections?

No. Agentcard swaps the user's real card in at checkout, so users keep their points and disputes go to their own bank. Chargebacks are handled by the issuer.

What does it cost to start?

Vault is free up to 5,000 users, then enterprise pricing. The Purchase Agent is $0.15 per request and free during beta. Full figures are at agentcard.sh/pricing.

Conclusion

You do not need to become a card vault. Agentcard Vault gives your AI agent product a white-labeled, passkey-encrypted card vault where card data never touches your servers, users approve every purchase with Face ID, and PCI scope stays off your plate entirely. Start with Vault, wire up the wallet link and webhooks in an afternoon, and add Issuing only when your product genuinely needs cards you fund yourself. Your users keep their points, their bank keeps their disputes, and you keep card data out of your architecture for good.