> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agentcard.sh/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> Payments infrastructure for AI agents

Agentcard issues virtual cards that AI agents can use to pay on the web — and a buying agent that completes whole purchases for you.

## What companies can do with Agentcard

There are two ways of using Agentcard:

### 1. Card issuing

Issue a virtual card per purchase, on behalf of any connected user, and pay anywhere online. This is the right path when you're implementing **your own browser automations for payments** — you mint the card through the [MCP server](/companies/mcp/overview), your automation enters it at the merchant's checkout.

### 2. The buying agent

Tell the [buying agent](/companies/mcp/tools/buy) what the user wants and it runs the entire transaction. It integrates directly with vendors like **DoorDash, Uber Eats, Good Eggs, and Rappi**, and automates the buying transaction on behalf of the user — search, cart, confirmation, payment, done.

Its main features:

* **No browser automations** — direct vendor integrations, so nothing brittle to maintain
* **Super fast** — checkout happens server-side in seconds
* **We handle authentication and the complete purchase** on behalf of the user, end to end

This documentation describes how to implement Agentcard so you can enable these features for your users and customers.

## Start here

The fastest path is the [wizard](/companies/getting-started/wizard) — one command that implements the integration in your repo and verifies it in sandbox. Prefer to wire it yourself? Follow [Manual Implementation](/companies/getting-started/manual-implementation).