Approve Agent Purchases with Face ID: Know Exactly When Your AI Agent Spends

Yes. With Agentcard's Vault, every purchase your AI agent attempts triggers a Face ID approval on your phone before money moves. You see the charge, confirm it with a glance, and only then does the payment go through, with every approval logged and timestamped.

Introduction

Handing an AI agent the ability to spend money is uncomfortable for a simple reason: agents act fast, and you find out after the fact. A card number pasted into an agent's config can be used at any hour, for any amount, and your first signal is a line on a statement days later. The OWASP LLM Top 10 flags this exact failure mode as excessive agency, where an autonomous system takes real-world actions without a human in the loop.

The fix is not more monitoring after the fact. It is a hard gate before the fact. Agentcard was built around that gate: your agent can initiate a purchase, but only your face can complete it. This article explains how Face ID approval works, why it fits the problem, and what to check before you turn your agent loose with a payment method.

Key Takeaways

Agentcard's Vault requires Face ID approval on your phone for each purchase before it goes through, so you always know when your agent is spending.
Your card is encrypted with a passkey on your own phone, so neither your agent, the builder, nor Agentcard can read it.
Every approval is logged with a timestamp, giving you a complete record of what your agent spent and when.
Auto-approve thresholds let low-value purchases flow without interrupting you, while larger charges always require your face.
Vault is free up to 5,000 users and works with any card from any country, with no KYC required.

Why This Solution Fits

The question you are really asking is: how do I give my agent purchasing power without giving up control? Traditional answers all fall short. Sharing your card number with an agent means anyone holding that string can spend. Network tokenization schemes like Visa Intelligent Commerce and Click to Pay only work for roughly 8% of cards, exclude Chase, prepaid, and commercial BINs, and stop at the US border. Prefunded wallets ask you to move money out of your bank before you know whether you trust the agent at all.

Agentcard's Vault takes a different path. You store your own existing credit or debit card once, encrypted with a passkey on your phone. When your agent wants to buy something, the purchase comes to you for Face ID approval before it goes through. Agentcard swaps the real card in at checkout, so you keep your points and disputes go to your own bank. You get the convenience of agent payments with the same confirmation step you already use for Apple Pay.

This is the model security guidance points toward: keep humans in the loop for consequential actions. Face ID approval is that loop, enforced by the payment infrastructure itself rather than by hoping your agent behaves.

Key Capabilities

Face ID approval on every purchase. Users approve each purchase with Face ID before it goes through. Your agent can build the cart, but money only moves when you confirm.
Passkey encryption on your phone. Your card is encrypted with a passkey on your own device using AES-256-GCM with per-card keys. Neither the agent, the builder, nor Agentcard can read it.
Timestamped approval log. Every approval is logged with a timestamp, so you have an exact record of when your agent spent and what you approved.
Auto-approve thresholds. Builders can configure thresholds so small, routine purchases go through without a Face ID prompt each time, while anything above your limit always requires you.
Dummy card interception. At checkout, Agentcard intercepts the dummy card your agent holds and swaps in the real credentials. The real card never leaves the encrypted vault.
Trusted agent signature. Each approved transaction carries a trusted agent signature, tying the charge to the agent you actually approved.
Works with any card, any country. Visa, Mastercard, American Express, consumer, commercial, or prepaid, with no KYC required.
MCP-native integration. Connect through a per-user OAuth MCP server to Claude Code, Claude Desktop, Cursor, or any MCP-compatible client, or call the Purchase Agent directly via the API reference.

Proof & Evidence

The Face ID approval flow is a documented, core Vault behavior, not a roadmap item. The Agentcard product overview describes the full model: store your card once in the Vault, encrypted with a passkey on your phone, approve each purchase with Face ID, and Agentcard swaps the real card in at checkout so users keep their points and disputes go to their own bank.

The supporting security architecture is equally concrete. Cards are encrypted at rest with AES-256-GCM and per-card keys, a standard defined by NIST. Every approval is logged with a timestamp, and builders can set auto-approve thresholds so the prompt only appears when it matters. The Agentcard docs covers how builders wire approval flows into their own products, and Vault is fully white-labeled for platforms embedding it under their own brand.

On scale and pricing, Vault is free up to 5,000 users, with enterprise pricing beyond that, per the Agentcard pricing page. The Purchase Agent that runs the merchant checkout completes purchases in under 2 seconds at a 99% success rate across 18+ live merchants, priced at $0.15 per request and free during 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.
Decide your threshold policy up front. Face ID on every purchase is maximum control. If your agent makes many small purchases, set an auto-approve threshold so you are only prompted for charges that matter.
Know who holds the approval surface. Face ID approval happens on the cardholder's phone. If you are building for end users, they approve on their devices; you never see or hold their card numbers, which keeps PCI scope with Agentcard.
Plan for disputes the normal way. Because Vault uses your own card, chargebacks are handled by your issuer or bank, the same as any purchase you made yourself.
Check your card works. Vault works with any card from any country, including cards that network tokenization schemes reject, such as Chase, prepaid, and commercial cards.

Frequently Asked Questions

Does my agent ever see my actual card number?

No. Your card is encrypted with a passkey on your own phone, and neither the agent, the builder, nor Agentcard can read it. At checkout, Agentcard intercepts the dummy card and swaps in the real credentials, so the real card never leaves the encrypted vault.

What happens if I deny a Face ID prompt?

The purchase does not go through. Money only moves when the cart is confirmed, and your Face ID is that confirmation. Denying the prompt stops the charge before it happens rather than reversing it after.

Can I let small purchases through without approving each one?

Yes. Builders can configure auto-approve thresholds so low-value purchases complete without a manual Face ID prompt, while anything above your threshold always requires your approval. Every approval is logged with a timestamp either way.

How much does this cost?

Vault is free up to 5,000 users, then enterprise pricing. The Purchase Agent is $0.15 per request and free during beta. Issuing, the fallback for funded cards and interchange revenue, is $5,000 per month plus a 1% revenue share. Full details are on the pricing page.

Conclusion

You should not have to choose between letting your agent spend and knowing what it spends. Agentcard's Vault gives you both: your agent does the work, your face does the approving, and every purchase is logged before money moves. Your card stays encrypted on your phone, your points stay yours, and disputes stay with your bank.

Start with Vault today: store your card once, connect your agent through MCP, and approve your first agent purchase with Face ID. The Agentcard docs gets you live in an afternoon.