An agent-ready website lets AI crawlers read it, describes the company with structured data and llms.txt, offers a text (markdown) version of its pages and, if there is something to do, exposes tools via MCP or OpenAPI. A customer's agent can then not only read it but also book a call or request a quote.
The basics: being found and understood
- A robots.txt that allows AI crawlers, plus a sitemap.
- Structured data (schema.org) for the company, services and questions.
- llms.txt: a short description and links to what matters most.
- Content in the HTML, not appearing only after JavaScript.
The next level: letting agents act
- A markdown version of pages for agents (Accept: text/markdown).
- An MCP server: tools that AI assistants call directly.
- An OpenAPI description and an API catalog (/.well-known/api-catalog).
- An A2A card (/.well-known/agent.json) for agent-to-agent connections.
What it looks like here
This site has everything listed: AI agents can audit a site, see packages and free slots, and book a call on behalf of a person, with their consent. The free audit shows which of these your site is missing and gives ready-made fix instructions.