CODEXA.Technologiez
All insightsAI

Charging AI agents for your content

Bilal Ashraf, Senior React Native Engineer · Oct 25, 2026 · 3 min read

Two business models have carried the web for twenty years. Advertising sells attention. Subscriptions sell access to someone who remembers a password. Both assume there is a person at the other end.

An AI agent reading your documentation does neither. It will not see a banner and it will not create an account. For most sites that leaves two outcomes: the agent takes the content for free, or you block it and lose the citation. Cloudflare has announced something aimed at a third.

What is x402?

An open protocol built on HTTP 402 Payment Required, a status code reserved in the original HTTP specification and left unused for thirty years. The flow has four steps. A client requests a resource. The server answers 402 with a price, an accepted asset, and a payment destination. The client pays and retries with proof of payment. A facilitator verifies that proof, and the server returns the resource. No account, no API key, and no prior relationship between buyer and seller.

What Cloudflare announced

Monetization Gateway, an engine that enforces x402 payments at Cloudflare's edge across more than 330 cities, so the payment handshake happens near the buyer and never reaches your origin. You can price anything sitting behind it: pages, datasets, APIs, and MCP tools. Rules are written as expressions and managed from the dashboard, the API, or Terraform.

  • Per-route or per-verb pricing, so a read costs less than a write.
  • Variable pricing tied to how much compute a request actually consumes.
  • Returning 402 instead of 401 to a caller with no credentials, which turns an unauthenticated request into a sale rather than a rejection.
  • Usage-based pricing on accounts you already have, with agents identified through Web Bot Auth.

Settlement is in stablecoins, USDC and Open USD at launch, which you can hold or redeem for fiat. Cloudflare has not said what it will take as a fee. Sub-second settlement is described as a design goal rather than a guarantee. The product is on a waitlist with no announced release date.

Why would an agent pay instead of just scraping?

Because the alternative is a 402 and no content. The leverage is positional rather than moral: the payment check runs at the edge, before the request reaches anything worth taking. That only works when what you are selling is genuinely hard to get elsewhere. An article that exists in twenty other places will not sell. A proprietary dataset, a calculation, or a support answer that actually resolves a ticket might, because the agent's operator is already paying for tokens and would rather spend a cent than fail the task.

Should you start charging agents now?

Almost certainly not, and not only because the product is unreleased. Most sites earn more from being cited than they would from metering, because a citation inside an AI answer is what ranking used to be. The maths changes only when your content is the product rather than the marketing for it. If you publish to be found, keep publishing to be found, and spend the effort on the structured data worth adding instead.

What would it actually take to use this?

Three commitments, each independent of the others. Your traffic has to run through Cloudflare, which for a self-hosted site means moving DNS and putting a proxy in front of an origin that currently answers directly. You have to be accepted off the waitlist. And you have to be willing to settle in stablecoins, which is a finance and compliance question long before it is an engineering one. None of those are reversible in an afternoon.

The part worth paying attention to

Not the payments. It is that the web is being asked to price itself per request, and the unit of value is a machine completing a task rather than a person giving you their attention. Every assumption behind a content strategy was built for the second thing.

That shift is real whether or not this particular product ships. If you are deciding what to build for agents at all, shipping AI features without the hype applies the same instinct to product work: understand the mechanism first, adopt it when it earns its place.

Bilal Ashraf, Senior React Native Engineer

Senior React Native engineer with more than five years in mobile and twelve apps live on the App Store and Google Play, mostly health-tech and AI, from wearable firmware through to the app. Also teaches production React Native.

Enjoyed this? Let's work together.

Start a project