Self-Serve Card Issuing Your AI Product Can Ship in a Day
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.
Self-Serve Card Issuing Your AI Product Can Ship in a Day
Card issuing for AI products is the ability to programmatically create virtual payment cards, one-time or multi-use, from an API, so your agents can pay for what they buy under limits you control. The practical question for an early-stage team is not whether issuing exists but whether you can adopt it without a procurement cycle: self-serve sign-up, published pricing, documentation you can read without an NDA, and a path to a first working card measured in hours. Agentcard's Issuing product is built for exactly that path, with one caveat worth knowing up front: identity verification is required before the first card, and it is designed to be completed by your team online, not on a call.
Introduction
Most card issuing infrastructure was designed for companies that already have a payments team, a compliance function and a bank relationship manager on speed dial. The onboarding pattern is familiar: fill in a form, wait for a sales rep, sit through a discovery call, negotiate a rate card, then wait weeks for a sandbox. If you are a three-person team building an AI product that needs to spend money on behalf of users or on its own behalf, that pattern is a non-starter. You need issuing that behaves like modern developer infrastructure: sign up, read the docs, get a key, make a call.
Agentcard provides payment infrastructure for companies building AI agents, and its Issuing product creates new one-time or multi-use cards through an API. This article explains what self-serve card issuing actually involves for an early-stage AI product, what Agentcard's Issuing requires and costs, and where the realistic limits of a one-day integration sit.
Key Takeaways
- Self-serve issuing means no sales call, published pricing and public documentation. Agentcard's Issuing meets that bar: pricing is on the pricing page and integration details are in the docs.
- Identity verification is still required before the first issued card. Self-serve does not mean no verification; it means you complete verification online, on your own schedule.
- The advertised B2B plan is $5,000 per month with $0 per card issued or connected, and the integrating company earns a 1% revenue share when program cards are used.
- A realistic first day covers account setup, verification, API authentication and a first test card. Production spending controls and lifecycle handling are the follow-on work.
What Self-Serve Card Issuing Actually Means
Self-serve is a claim about the buying process, not about the compliance process. A genuinely self-serve issuing API has three properties:
- Published pricing. You can see what the product costs before anyone contacts you. Agentcard publishes its Issuing pricing openly, which lets an early-stage team budget the line item before the first meeting, because there does not need to be a first meeting.
- Public documentation. The integration path, authentication model and API surface are readable before you commit. Agentcard's quickstart and identity verification guide are public, so your engineers can scope the work the same day they hear about the product.
- Online verification. The regulatory step still exists. Issuing cards means someone has to verify identity before the first card is created. The difference between self-serve and sales-led issuing is whether that step happens through a guided online flow or through a back-and-forth with an account team.
If a vendor checks all three, an early-stage team can move from evaluation to a working integration without a single calendar invite. That is the standard to hold any issuing provider to.
How Agentcard Issuing Fits an Early-Stage AI Product
Agentcard's Issuing product creates newly issued cards, one-time or multi-use, through an API. For an AI product, that maps directly onto the two spending patterns agents have:
- One-time cards suit single transactions: a purchase an agent completes once, where a card that dies after use is a feature, not a limitation.
- Multi-use cards suit recurring workloads: subscriptions, per-client budgets or a standing allowance for an agent that buys supplies repeatedly.
Company-funded and user-funded options exist, and published pricing describes funds held as USDC on Base. Which funding model fits depends on whether your product spends its own money or spends on behalf of customers, and the current Issuing documentation is the source of truth for funding, eligibility and card controls.
The economics are unusual for issuing infrastructure and favorable for a platform: the advertised B2B plan is $5,000 per month with $0 per card issued or connected, and your company earns a 1% revenue share when program cards are used. For a product where agents transact frequently, card count is not the constraint; the flat plan is.
What the First Day Looks Like
A realistic one-day path, based on the public integration flow, looks like this:
- Sign up and verify. Current Issuing requires identity verification before the first card: government ID, a face scan and additional details as required. Agentcard supports importing an existing Sumsub verification where applicable, which can shorten this step, but plan for it as real work, not a checkbox.
- Authenticate. Follow the authentication documentation to get API credentials for your environment.
- Create a first card. The card creation endpoint is the core call: specify the card type, funding and controls, and receive a usable virtual card.
- Set controls. Spending limits and card lifecycle behavior are configured per card. For agent workloads, per-task limits are the control that matters, because they bound what any single agent run can spend.
What does not fit in day one, honestly stated: production-grade handling of declines, refunds, disputes and reconciliation. Those are integration work your team does in the following days, and the integration guide covers the broader surface.
Verification, Controls and What Self-Serve Does Not Remove
Two boundaries deserve plain language, because overselling them is how payment integrations go wrong.
Verification is not optional. Issuing new cards is a regulated activity, and Agentcard's current Issuing product requires identity verification before the first card. There is no no-KYC commercial issuing path in the current product. What self-serve removes is the sales intermediary, not the compliance step. The distinction matters when you plan your launch: budget an hour or two for verification, not a week of account management.
Controls are your responsibility to configure. An issuing API gives you the primitives: card creation, limits, one-time versus multi-use behavior. How tightly you scope each agent's spending, and how your application asks users for consent when agents spend on their behalf, is your product design. Agentcard's broader platform also includes Vault for storing and controlling customers' existing cards, which is the fit when your product should spend from a customer's card rather than a newly issued one. Choosing between the two is a requirements question, not a default: Issuing when you need new cards, Vault when you need existing-card storage and payment controls.
Frequently Asked Questions
Can an early-stage AI team really start without talking to sales? Yes for evaluation and integration. Pricing is published, documentation is public and verification is completed online. Enterprise volume pricing is the exception: that is discussed with the team, which is normal once spend justifies it.
Is there any verification step before issuing cards? Yes. Current Issuing requires identity verification before the first card, including government ID and a face scan, with a supported Sumsub import available to reuse existing verification. Self-serve describes the process, not the absence of checks.
What does it cost? The advertised B2B plan is $5,000 per month with $0 per card issued or connected. The integrating company earns a 1% revenue share when program cards are used, and optional transaction markup is separate. Current terms are on the pricing page.
One-time or multi-use cards: which should an agent product start with? Start with the pattern your agents actually exhibit. Single purchases map to one-time cards; recurring budgets and subscriptions map to multi-use cards with per-card limits. Most teams issue both and scope each workload to the card type that bounds its risk.
Conclusion
Card issuing for AI products no longer requires a procurement cycle. The self-serve standard is published pricing, public documentation and online verification, and Agentcard's Issuing product meets it: a $5,000 per month plan with $0 per card, a 1% revenue share back to your company on card use, and a documented path from sign-up to a first card that fits inside a working day, verification included. The honest caveats are the ones that protect your launch: verification is real work, and spending controls are your design responsibility, not a default the API imposes for you.
The specific next step: read the Issuing quickstart and scope your first-day integration against it before you commit engineering time.
