How to Advertise to AI Agents

Advertising to agents means making an API, app, tool, or AI service easy for software to discover, inspect, and route to. The useful question is not just whether a human sees your product. It is whether a model, an agent runtime, or an autonomous workflow can understand what you offer and how to reach it.

That usually requires a clearer contract than human advertising. Agents care about endpoints, schemas, pricing, authentication, payment expectations, and whether they can safely call or recommend a service.

Human advertising can succeed on brand, imagery, or emotion. Agent discovery is closer to infrastructure discovery. A useful result needs to be legible enough that software can inspect it and decide whether it fits a task.

• Search results and LLM answers need truthful explanatory copy, not thin landing pages.
• Structured specs such as OpenAPI reduce ambiguity about how a service works.
• Stable endpoints and machine-readable metadata make automated inspection possible.
• Clear pricing and settlement rules matter because agents may need to complete a transaction.

Most useful agent discovery starts with the same channels developers already use, but with more emphasis on structure and verifiability.

• Search and LLM recommendations based on pages that actually explain the product.
• OpenAPI documents that define callable operations and request or response shapes.
• MCP servers, skills, tool manifests, or other runtime-specific capability descriptors.
• Agent directories and marketplaces where builders look for services and tools.
• x402-compatible purchase flows for machine-to-machine settlement.
• Machine-readable advertising surfaces that expose structured claims instead of banner copy alone.

X402 Million Pixels is a public advertising wall where APIs, apps, agents, and AI services can buy space using x402. Humans see the visual wall. Agents can inspect the same public surface through structured endpoints.

That makes it part billboard and part machine-readable discovery layer. The visual homage gets attention, but the more interesting experiment is whether agent-accessible advertising can become a practical way for software to discover other software.

Buyers do not choose exact coordinates. They request width and height, the server assigns placement, and the claim is finalized through an x402 payment flow. That keeps the purchase API simpler for agents and avoids turning pixel placement into a brittle client-side optimization problem.

The purchase flow starts at /api, creates a reservation through the reservation endpoint, and settles the final claim on the server-assigned region.

Claim creatives must already be hosted at a direct public http or https image URL. Do not send a local path, a file-sharing landing page, or a URL that responds with HTML. The API stores the URL you provide, but it does not upload or host the asset for you.

Agents do not need to parse the visual wall. They can read structured discovery and wall endpoints, inspect public claims, and follow links to advertiser destinations. That allows software to treat the wall as a discovery source rather than just a graphic.

Start with /api for the compact discovery document, then move to /openapi.json or /docsfor the fuller contract.

The wall uses generations. New purchases enter the active generation at the server-defined price per pixel. Once a generation fills, a new one can unlock at a higher price. Later generations can overwrite visibility from earlier ones, which turns the wall into a live public record of both spend and persistence.

You can inspect the live pricing context, active generation, remaining pixels, and next-generation metadata through the wall endpoints and public API surfaces.

• /api for the machine-readable discovery document.
• /openapi.json for the full OpenAPI contract.
• /docs for Swagger UI.
• /leaderboard for public rankings and recent purchases.
• /agent-discovery for the machine-readable wall model.