Company profile · 3 min read
Agentcard gives AI agents user-controlled cards for online purchases
Agentcard offers a vault for existing cards and issuing for new one-time or multi-use cards, with spending controls and purchase workflows.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
Agentcard gives agents a way to pay online without handing a raw card number to the agent or merchant automation.
What it does
The company offers two products. Vault stores a user’s existing cards and lets an agent use them through a permissioned flow. Issuing creates new one-time or multi-use cards with spending limits. The current site says card details stay in a secure vault, while the docs describe passkey approval, encrypted cards and user or company funding. Agentcard homepage Getting started docs
The product is aimed at teams building agent browsers, assistants or shopping apps. The purchase path can be a browser automation or Agentcard’s Purchase API, with early integrations such as DoorDash, Good Eggs and Locale shown publicly.
Why I’d look closer
Payments are often the final step where an agent product stops being a demo. Agentcard’s focus is narrow and practical: make a user-approved payment instrument available to the agent while keeping card details away from the application.
The founders’ backgrounds fit that payments problem. The YC profile describes Karen Serfaty’s prior expense-card company and Remote.com cards work, with Felipe Abello’s payments and startup background.
The business model is also explicit: the homepage says issued-card purchases can generate 1% interchange plus dashboard markup. That is a company-published monetization claim.
What could make it the wrong choice
Payments introduce trust, fraud, KYC, chargebacks, merchant login and authorization questions. A developer should understand whether Vault or Issuing fits the product, who holds funds, which countries are restricted, what purchase limits exist and how a user revokes access.
The public success-rate language in the launch material is company-reported. I did not connect an account or test a purchase.
My editorial take
I would shortlist Agentcard for an agent product that already has a real purchase workflow and needs a user-controlled payment boundary. I would not add cards before the product can explain approvals, limits, refunds and receipts. The hard part is not making an agent spend. It is making the spend legible and reversible enough that users will allow it.
Cohort context
Agentcard is listed in Summer 2026. In our 2026-09-18 directory snapshot, 16 of 232 listed companies in that cohort have YC’s primary industry label Fintech (6.9%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:18:31.927Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Not observed in this response |
| H1 or H2 heading | Observed |
| Typed structured data | Not observed in this response |
| Docs/developer link | Observed |
| Pricing link | Observed |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗