> For the complete documentation index, see [llms.txt](https://developer.paddle.com/llms.txt).

# Paddle for AI companies

Usage-based billing, flexible pricing, and global merchant of record for AI businesses — plus MCP servers that let AI agents work directly with Paddle.

---

AI businesses sell software with non-traditional billing patterns — token usage, hybrid pricing (subscription plus overage), pay-as-you-go credits, custom enterprise deals. Paddle's pricing engine, merchant-of-record coverage, and MCP servers make these models work without you rebuilding a billing system.

## What Paddle gives AI companies

{% card-group cols=3 %}
{% card title="Flexible pricing engine" icon="carbon:pricing-consumption" %}
Subscriptions, one-time charges, custom items, and metered usage all live on the same transaction model — combine them however your product needs.
{% /card %}
{% card title="Global merchant of record" icon="carbon:globe" %}
Paddle is the seller of record on every sale. Tax, fraud, and 200+ payment methods covered for you — useful when you're shipping fast and selling globally on day one.
{% /card %}
{% card title="MCP-native integration" icon="mcp" %}
The Paddle MCP server lets AI agents read and act on your account. The docs MCP server gives them current Paddle knowledge. Both ship today.
{% /card %}
{% /card-group %}

## Usage-based billing

The most common AI pricing model is "subscription + usage" — a base plan, plus charges for tokens, API calls, generations, GPU minutes, or whatever unit you meter on.

In Paddle, usage is modeled as **custom items on a transaction**. Each billing period, you call the API to add line items for the customer's usage; Paddle bills the next invoice with the base plan and the metered total. This keeps the contract simple — you handle the metering, Paddle handles the billing.

The high-level shape:

1. **Set up the base plan as a recurring price** with whatever cycle you bill on (monthly, annual). See [create products and prices](https://developer.paddle.com/build/products/create-products-prices.md).
2. **Track usage in your own infrastructure** — your application is the source of truth for metering. Most AI products already have this.
3. **At billing time, add custom items to the upcoming transaction** with the metered amount, unit price, and a description. See [Bill custom items, prices, and products](https://developer.paddle.com/build/transactions/bill-create-custom-items-prices-products.md).
4. **Paddle bills the customer** at the next renewal, including both the recurring plan and the metered usage as separate line items on the invoice.

You can also bill **purely usage-based** (no recurring plan) by creating a transaction with custom items each cycle, or **prepaid credits** by selling credit bundles as one-time products and tracking balance in your application.

For complex models — tiered usage rates, free quotas, overage on top of included credits — combine the building blocks. The pricing engine doesn't enforce a model; you compose what you need.

## Flexible pricing models

The same primitives support every common AI pricing pattern:

- **Pay-as-you-go** — sell credit packs as one-time products; meter consumption in your app.
- **Subscription with included quota** — recurring plan with a quota baked into your provisioning logic; bill overage as custom items.
- **Token-based metering** — recurring base + custom items per million tokens, billed monthly.
- **Per-seat AI tools** — recurring per-seat plan, with optional usage charges per seat.
- **Custom enterprise deals** — manual invoicing for negotiated terms; see [Issue invoices](https://developer.paddle.com/build/invoices/create-issue-invoices.md).
- **Hybrid free tier** — no Paddle entity for free users; create a customer in Paddle when they upgrade.

You can ship multiple models in parallel — some customers on subscriptions, some on PAYG, some on enterprise invoicing — without splitting your billing system across vendors.

## Connect AI tools to Paddle

The Paddle MCP servers are the native integration point if you're building with AI tools or shipping AI features yourself.

{% card-group cols=2 %}
{% card title="Paddle MCP server" icon="carbon:api-1" url="/sdks/ai/paddle-mcp" %}
Lets AI agents read data and take actions in your Paddle account. Build pricing, debug billing, generate reports, onboard enterprise customers — all in natural language.
{% /card %}
{% card title="Docs MCP server" icon="carbon:book" url="/sdks/ai/docs-mcp" %}
Gives AI agents up-to-date Paddle knowledge — every guide, the OpenAPI spec, and SDK references. No more outdated training data.
{% /card %}
{% /card-group %}

Use both together. The docs MCP server tells the agent how Paddle works; the Paddle MCP server lets it do things. Both work with Claude Code, Cursor, Claude Desktop, VS Code, Codex, Raycast, and other MCP-compatible clients. See [Agents and AI](https://developer.paddle.com/sdks/ai.md) for the full picture, or jump straight to the [Build with Cursor](https://developer.paddle.com/get-started/build-with-ai/cursor.md) and [Build with Claude Code](https://developer.paddle.com/get-started/build-with-ai/claude-code.md) guides.

## Built for global scale

AI products tend to be global from day one — your users sign up from anywhere, pay in any currency, and expect tax-compliant receipts. Paddle's merchant-of-record coverage handles all of that:

- **Tax compliance in 200+ countries** without you registering for VAT, GST, or sales tax in each jurisdiction.
- **30+ currencies** with automatic FX handling.
- **Local payment methods** — Apple Pay, Google Pay, PayPal, iDEAL, UPI, Pix, and more — selected automatically based on the customer's region.
- **Card processing** routed to the best acquirer for each transaction to maximize approval rates.

If you're growing fast across regions, this is meaningful: every market you'd otherwise have to register and build for is already covered.

## Onboard enterprise customers

AI companies tend to land enterprise deals early. Paddle handles the manual side:

- **Custom pricing** per customer with negotiated rates.
- **Manual invoicing** with NET 30 / NET 60 terms — see [Create and issue invoices](https://developer.paddle.com/build/invoices/create-issue-invoices.md).
- **Wire transfer payment** alongside the standard card and digital wallets.
- **Tax-exempt billing** for customers with valid VAT/GST registration numbers.
- **Sales-assisted checkout** — generate a checkout link and email it instead of a self-serve flow.

You can use the Paddle MCP server to drive enterprise onboarding from natural language: paste a quote, ask the agent to set up the customer, and review before the agent issues the first invoice.

## Where to start

{% card-group cols=3 %}
{% card title="Python quickstart" icon="carbon:logo-python" url="/get-started/quickstart/python" %}
Recommended for AI teams — install the Python SDK, make your first request, and start scripting.
{% /card %}
{% card title="Set up the MCP server" icon="carbon:api-1" url="/sdks/ai/paddle-mcp" %}
Connect Claude Code, Cursor, or any MCP-compatible client to your Paddle account in under a minute.
{% /card %}
{% card title="Build with Claude Code" icon="carbon:terminal" url="/get-started/build-with-ai/claude-code" %}
Wire up both MCP servers and ship your Paddle integration with AI assistance from your terminal.
{% /card %}
{% /card-group %}