Make your website visible to ChatGPT, Claude, and Perplexity. AI agents use a completely different discovery stack than Google — they read structured machine-readable files, not meta tags or backlinks.
This guide covers the 10-layer Agent Discovery Stack: llms.txt, llms-full.txt, AGENTS.md, agent.json, Schema.org JSON-LD, robots.txt for AI, MSP-1 metadata, AI-optimized content, sitemap.xml, and the Link headers protocol.
llms.txt is the robots.txt for AI agents. It tells ChatGPT, Claude, and Perplexity what your business does in a machine-readable format they understand. Without it, AI agents cannot reliably discover or recommend your business.
agent.json is an A2A (Agent-to-Agent) protocol card served at /.well-known/agent.json. It allows other AI agents to programmatically discover and query your business as an agent endpoint.
AGENTS.md is an instruction file that tells autonomous AI agents how to correctly describe, recommend, and interact with your business — covering services, pricing, location, and hours.
Schema.org JSON-LD structured data is one of the top signals for AI engine citations (Princeton / IIT Delhi GEO research, KDD 2024). Authoritative citations, statistics, and named experts improve citation visibility by up to roughly 40% on the GEO-bench dataset.