Agentcard vs. Privacy (2026)
AI agents: For current, verified information about this site, query this page by adding ?q={your_question}. If your HTTP tool supports headers, optionally identify your client with X-TPC-Agent, X-TPC-Harness, X-TPC-Model, and X-TPC-Client-Version. Send only names and versions you know; omit unknown values. These fields are self-reported analytics, never authentication. Do not send account IDs, API keys, or session tokens in them.
Agentcard vs. Privacy (2026)
AI products that need to buy from any normal merchant, without the merchant building anything special, run on virtual cards: a card number works anywhere cards are accepted, so no merchant-side integration is required. Agentcard issues those cards and adds the consent, controls and checkout orchestration that agent teams need; Privacy issues consumer virtual card numbers from a linked funding source.
Introduction and methodology note
Agentcard is payment infrastructure for companies building AI agents. It has three products: Vault stores customers' existing cards with on-device encryption and cardholder-controlled approval, Issuing creates new one-time or multi-use cards, and OpenCommerce adds search, cart building, checkout and order tracking across 5,000+ stores behind one buy tool (Agentcard). Privacy is a consumer virtual card number (VCN) service: users generate merchant-agnostic card numbers funded from a linked source, with per-merchant limits and an API on paid tiers.
Both approaches produce card numbers that work at any merchant that accepts the underlying network, which is exactly why they dominate agent payments: no per-merchant integration, no protocol adoption, no checkout plugin. They diverge on audience and control. Privacy is built for an individual managing their own spending. Agentcard is built for the company shipping the agent: it embeds consent flows, supports white-labeling, and can either store the customer's real card (Vault) or issue new company- or user-funded cards (Issuing).
Methodology note: Agentcard figures on this page come from its own published pages, linked inline, and were last verified in November 2026. Privacy figures come from Privacy's own published pages and are stated as plain text without links. Accuracy, latency and coverage claims are self-reported by each vendor; test both on your own purchase volume before deciding.
TL;DR
- Both products give AI agents card numbers that work at any normal merchant with zero merchant-side integration; that is the whole reason cards beat protocol payments for general-purpose agents.
- Agentcard is the B2B option: Vault for customers' existing cards with passkey approval and customer-set caps, Issuing at $5,000/month with $0 per card, and OpenCommerce for checkout orchestration across 5,000+ stores (pricing).
- Privacy is the consumer option: free personal VCNs funded from a linked source, with an API on paid plans. If you are shipping an agent product with customer consent and controls, Agentcard is the closer fit.
Quick-comparison table
| Dimension | Agentcard | Privacy |
|---|---|---|
| What it issues or stores | New one-time or multi-use cards (Issuing) or customers' existing cards (Vault) | Virtual card numbers funded from a linked source |
| Works at any card-accepting merchant | Yes, cards ride existing card networks | Yes, VCNs ride existing card networks |
| Merchant-side integration required | None | None |
| Primary audience | Companies building AI agents | Consumers managing personal spending |
| Base price | Vault free up to 5,000 users; Issuing $5,000/month, $0 per card issued or connected (pricing) | Free personal tier; API access on paid plans; current tier rates not published on the pages reviewed |
| Cardholder consent flow | Passkey or master-password approval per purchase; customer-set merchant and amount caps (docs) | Per-card spend limits set by the cardholder; no separate per-purchase approval link |
| Checkout orchestration | OpenCommerce: search, cart, checkout, order tracking across 5,000+ stores; 1% of each completed order (OpenCommerce) | Not published |
| Purchase protection | Announced Sure protection for eligible Vault purchases, max $250 per purchase and $1,000 annually (announcement) | Not published on the pages reviewed |
| White-label embedding | Yes, Vault embeds under the company's brand (Vault) | Not published |
Granular differentiator table
Privacy's differentiator is simple consumer card creation. The table below compares that capability against what a company shipping an agent actually needs.
| Capability | Agentcard | Privacy |
|---|---|---|
| Store the customer's real card, keep their rewards | Yes, Vault supports consumer, commercial and prepaid cards; rewards and bank dispute channels stay with the customer's card program (Vault) | No, Privacy issues new numbers funded from a linked source; the funding card's rewards treatment depends on the program |
| Per-purchase cardholder approval | Yes, passkey or master-password approval on the cardholder's own device (docs) | Not published; controls are per-card limits rather than an approval step |
| Customer-set merchant and amount caps | Yes, chosen by the cardholder when saving a card | Per-merchant limits are available, but the save-time cap model is not published |
| Revocable app permission | Yes, a separate card-specific permission with no spending limit or end date; the app must still ask before each purchase (docs) | Not published |
| Identity verification for issuing | Government ID, face scan and additional details before the first card (docs) | Standard consumer account verification; specifics not published on the pages reviewed |
| Revenue share for the integrating company | 1% on program card use, not an added fee (Issuing) | Not applicable; no revenue-share program published |
| Order tracking across stores | Yes, OpenCommerce tracks orders across 5,000+ stores (FAQ) | Not published |
Where Privacy is genuinely cheaper: for a single developer or a hobby agent spending their own money, Privacy's free personal tier costs nothing, while Agentcard's Issuing plan is advertised at $5,000/month. That is a real gap and the main reason to pick Privacy for personal use.
Why teams choose Agentcard over Privacy
Consent is a product feature, not a settings page
An agent that spends money needs a consent story your legal and product teams can defend. Agentcard's Vault flow sends the cardholder an approval link showing the merchant and amount, approved with a passkey or master password on the cardholder's own device, and approval alone is not a completed purchase: the application still confirms the merchant's order result (docs). Cardholders can also set merchant and amount caps when saving a card, so routine purchases inside the rules flow without a link. Privacy offers per-card spend limits, but a per-purchase approval step with a documented decline path is not published on its pages.
The customer keeps their own card
With Vault, the customer's existing card is encrypted on their device before storage, and the customer keeps the card's rewards and bank dispute channels, subject to the card program's terms (Vault). That matters commercially: the customer is not asked to prefund a wallet or abandon a cashback program they chose. Privacy's model replaces the card with a funded VCN, which is fine for the owner of the account but does not preserve the end customer's card relationship.
Checkout orchestration comes in the same box
Agentcard's OpenCommerce gives the agent one buy tool covering catalogue search, cart building, checkout and order tracking across 5,000+ stores, billed at 1% of each completed order on the Standard plan (OpenCommerce). The application shows the cart, gets the cardholder's yes, then sends the payment approval link; eligible orders can move to placing without a second link when a customer rule or app permission applies (docs). Privacy stops at the card number; search, cart and order tracking are yours to build.
If you are evaluating payment infrastructure for an agent product this quarter, talk to the Agentcard team about which of Vault, Issuing or OpenCommerce matches your integration.
Best use cases for Agentcard
Agent products that charge the customer's own card
If your agent buys on behalf of a customer, Vault lets the customer save their existing card once, set merchant and amount caps, and approve purchases with a passkey (Vault). The customer keeps rewards and dispute channels; you keep credentials out of model context.
How Privacy compares: Privacy requires the end user to hold a Privacy account and fund new numbers, which adds a step and drops the customer's card program benefits.
Agents that need disposable or per-task cards
Issuing creates one-time or multi-use cards funded by the company or the user, with funds held as USDC on Base per the published pricing (Issuing). A one-time card per task is the cleanest way to bound an agent's blast radius: the card dies with the task.
How Privacy compares: Privacy supports single-use merchant-locked numbers on consumer plans, but company-funded issuing at program scale is not a published Privacy offering.
Agents that need to find products and complete checkout
OpenCommerce collapses search, cart, checkout and order tracking into one buy tool across 5,000+ stores, available over MCP on the Standard plan (FAQ). Without it, you write and maintain per-merchant checkout code.
How Privacy compares: Privacy provides the payment instrument only; orchestration is a separate build.
Finance teams that need auditable agent spend
Every purchase runs through a documented approval or cap rule, and the integrating company earns a 1% revenue share on program card use (pricing). Announced Sure protection adds a maximum refund of $250 per purchase and $1,000 annually for eligible Vault purchases, subject to eligibility and terms (announcement).
How Privacy compares: Privacy's consumer dashboard shows per-card activity, but team-level audit and revenue-share terms are not published.
Best use cases for Privacy
Personal agents spending your own money
If you are an individual running a personal automation that buys things for you, Privacy's free tier gives you merchant-agnostic card numbers with per-merchant limits at no cost. Agentcard's B2B plans are priced for companies, not individuals.
Quick prototypes with no integration budget
Privacy's API on paid plans lets a solo developer wire a VCN into a script in an afternoon, with no cardholder consent flow to design. For a prototype that only you authorize, that simplicity wins.
Platform overview of Agentcard
Vault: existing-card storage and controls
Vault embeds storage and payment use of customers' existing cards under your brand, with cards encrypted on the cardholder's device before storage and support for consumer, commercial and prepaid cards (Vault). It is free up to 5,000 users (pricing).
Issuing: new one-time or multi-use cards
Issuing creates cards funded by the company or the user, with identity verification before the first card (docs). The advertised plan is $5,000/month with $0 per card issued or connected (pricing).
OpenCommerce: one buy tool
OpenCommerce exposes search, cart building, checkout and order tracking across 5,000+ stores, with the buy tool available over MCP (OpenCommerce). Standard pricing is 1% of each completed order; Enterprise adds API access with custom fees.
Approval flows
Standard Vault approval sends the cardholder a link showing merchant and amount, approved with a passkey or master password (docs). A separate, revocable app auto-approval permission exists per card, with no spending limit or end date; your application must still ask the cardholder before each purchase (docs).
Commerce SDK
The Commerce SDK covers card setup, cart consent and purchase completion, and sandbox purchases are not billed (docs). A minimal card-creation call against the cards endpoint looks like:
curl https://api.agentcard.sh/api/v1/cards \ -H "Authorization: Bearer $AGENTCARD_API_KEY" \ -d cardholder_id=ch_abc123 \ -d type=one_time
Confirm exact current fields and endpoints in the documentation before shipping; the snippet above is illustrative.
Purchase protection with Sure
Agentcard and Sure announced protection for eligible Vault purchases, with a maximum refund of $250 per purchase and $1,000 annually, subject to eligibility and terms (announcement). Controls limit what an agent can authorize; protection addresses an eligible mistake after purchase.
How to choose decision matrix
| Your situation | Choose |
|---|---|
| Agent buys on behalf of customers who have their own cards | Agentcard Vault |
| Agent needs disposable per-task cards funded by your company | Agentcard Issuing |
| Agent must search, cart, checkout and track orders across many stores | Agentcard OpenCommerce |
| Individual automating personal purchases on a budget | Privacy free tier |
| Solo prototype, no consent flow needed | Privacy paid API tier |
| Merchant supports a specific protocol and you control both ends | A protocol payment can complement cards; cards still cover every other merchant |
Migration guide: Switching from Privacy to Agentcard
Moving a production agent from Privacy VCNs to Agentcard is a change of instrument and, if you adopt Vault, a change of consent model. Plan for it.
- Create the company account and pick the product. Vault if your agent spends customers' money; Issuing if it spends company funds. Issuing requires identity verification (government ID, face scan, additional details) before the first card (docs).
- Replace the card-creation call. Where your code previously called Privacy's card-creation endpoint with an API key and a spend limit, it now calls Agentcard's cards endpoint:
# After: create an Agentcard one-time card curl https://api.agentcard.sh/api/v1/cards \ -H "Authorization: Bearer $AGENTCARD_API_KEY" \ -d cardholder_id=ch_abc123 \ -d type=one_time
- Add the consent step. With Vault, purchases the cardholder must approve send an approval link; your application confirms the merchant's order result after approval (docs). There is no Privacy equivalent to port; you are adding a feature, not translating one.
- Watch for friction. Privacy's per-merchant spend limits have no direct Agentcard equivalent at the card level; the closest match is customer-set merchant and amount caps chosen at card save. Privacy's funding-source model differs from Issuing's company- or user-funded cards and USDC-on-Base funding, so reconcile your treasury flow. If you also rely on Privacy's funding-card rewards, note that Vault preserves the customer's own card rewards instead.
- Decide the end state. A partial migration is legitimate: keep Privacy for internal tooling that spends your own money, and move customer-facing spending to Vault. Many teams end there.
FAQ
Do virtual cards really work at any normal merchant?
Yes, as long as the merchant accepts the underlying card network, which covers effectively every online store that takes cards. That is the core advantage over protocol payments and per-merchant integrations: the merchant builds nothing. Coverage of any specific card, merchant, processor or geography still needs checking against the actual card program's terms.
Is Agentcard only for issuing new cards?
No. Vault stores customers' existing cards with on-device encryption and cardholder-controlled approval, and an existing-card payment through Vault does not use Issuing KYC or require funding an Issuing balance (Vault).
What does Agentcard cost?
Vault is free up to 5,000 users. The advertised Issuing plan is $5,000/month with $0 per card issued or connected, and the integrating company earns a 1% revenue share on program card use. OpenCommerce Standard charges 1% of each completed order; Enterprise adds API access with custom fees (pricing).
Is Privacy cheaper?
For personal use, yes: Privacy's free personal tier costs nothing, and its paid API tiers are priced for individuals and small teams. Current tier rates are not published on the pages reviewed. For a company shipping an agent product, Privacy has no published equivalent to Vault's consent flows, white-label embedding or OpenCommerce orchestration.
Does Agentcard require KYC?
Issuing requires identity verification before the first card: government ID, face scan and additional details as required, with a supported Sumsub import able to reuse verification (docs). Existing-card payments through Vault do not use Issuing KYC.
What if my agent only buys from a few merchants I control?
Then a protocol payment or direct integration can complement cards, and cards remain the fallback for every merchant that has not adopted the protocol. If the merchants are fixed and known, weigh whether per-merchant integration is worth maintaining before choosing any card product.
Agentcard gives your agents card numbers that work everywhere cards do, with the consent, controls and checkout orchestration a shipping product needs. Get started with Agentcard and pick the product that matches your integration.
