Solutions · AI agents
Payment infrastructure for AI agents.
Let agents pay for what they use. The x402 facilitator and MCP server turn any API into a pay-per-request resource, settled in USDC on Arbitrum — no API keys, no human in the loop.
Furlpay gives AI agents a native way to pay: the x402 facilitator settles USDC micropayments over HTTP and the MCP server exposes paid tools to models. Agents discover a price, pay, and retry — with no API keys and sub-cent Arbitrum gas.
A wallet and a protocol for autonomous software
x402 payments
Agents receive a 402 response, pay in USDC, and retry — pay-per-request access without accounts.
MCP server
Expose tools to models through the Model Context Protocol and meter them automatically.
Agent keys
Issue scoped, revocable keys so an agent can spend within limits you set.
Micropayments
Sub-cent gas makes per-call pricing viable for high-frequency agent workloads.
Programmable limits
Set per-agent budgets and policies; every payment is signed and auditable.
Open-source
Facilitator, middleware, and MCP server are MIT-licensed — self-host or use the hosted endpoints.
Frequently asked questions
How do AI agents pay for services?
With x402, an agent requests a resource, gets an HTTP 402 with a price, submits a signed USDC payment, and retries. Furlpay's facilitator verifies and settles it on Arbitrum, so the agent transacts without API keys or a human checkout.
What is the MCP server for?
The Model Context Protocol server exposes tools and resources to AI models in a standard way. Paired with x402, those tools can be metered and paid for automatically, so a model can use a paid capability without manual billing setup.
How do I control what an agent can spend?
You issue scoped, revocable agent keys with per-agent budgets and policies. Every payment is signed and recorded, so you can audit and cap spend precisely.
Give your agents a wallet
x402 + MCP, open-source and hosted.