AI endpoints are structured JSON files served at predictable paths on your website. They are designed to be consumed directly by AI systems, crawlers, and agents ... not by human visitors.

Think of them as an API layer for your website that does not require a database or backend: just static JSON files that AI systems can request at any time.

The Core AI Endpoints

llm.json

The primary AI identity endpoint. Served at /ai/llm.json or /llm.json, it contains:

  • Site name, URL, and canonical domain
  • Site description and primary topics
  • Primary services or content types
  • Constellation/network membership

ai-sitemap.json

The content index endpoint. Served at /ai/sitemap.json, it maps all content by type and priority. See the AI Sitemap Guide for format details.

catalog.json

A content catalog endpoint that exposes the full set of articles, pages, and other content with metadata. Useful for AI systems building a comprehensive understanding of site coverage.

health.json

A machine-readable health status endpoint. Served at /ai/health.json, it confirms the site is operational and provides basic status information for AI agents that need to verify endpoint availability.

Why These Files Matter

AI systems and agents increasingly make direct requests to these endpoints when building their models of what your website is and covers. Websites that serve clean, accurate AI endpoints are easier to understand, more likely to be cited accurately, and better positioned for AI agent interactions.

Implementing AI Endpoints

AI endpoints can be static JSON files you create and maintain manually, or they can be generated dynamically from your content management system. Either approach works ... the important thing is that they exist, are publicly accessible, and accurately represent your site.