Let AI agents
understand your site.
AI-Me is an open standard for machine-readable site identity. One JSON file at a well-known endpoint gives AI agents a structured, explicit picture of who you are and what you offer — without crawling, scraping, or guessing.
AI agents are flying blind.
AI assistants, agent frameworks, and LLM-powered tools increasingly try to understand what websites offer — to answer user questions, route requests, and take actions on behalf of users.
Right now, they have three bad options:
Slow, fragile, and hits bot protections. No reliable structure.
Expensive, slow to aggregate, and still incomplete for agents.
Hallucinated facts, wrong prices, outdated CTAs. The worst outcome.
Schema.org helps search engines but is scattered across pages. llms.txt provides human-readable guidance, not machine-readable structure. There is no standard, compact, authoritative document an AI agent can fetch to understand a site in one request.
That's the gap AI-Me fills.
Agent tooling is forming default discovery behaviors now. Sites that publish structured manifests early will be indexed first.
How AI-Me works
A single well-known endpoint. One HTTP request. Full picture.
Publish
Create /.well-known/ai-me.json — a JSON document conforming to the AI-Me schema.
Agents fetch
Agents built to support AI-Me can fetch that URL in one request — no crawling required.
They understand
Name, summary, canonical pages, offerings, support contacts, and links to MCP/A2A/llms.txt — everything needed to engage correctly.
Start in five minutes
Publish your AI discovery manifest today.
{
"schema_version": "0.1",
"name": "Your Company",
"website": "https://example.com",
"summary": "What you do, in one sentence.",
"canonical_pages": {
"home": "https://example.com/",
"contact": "https://example.com/contact"
},
"last_updated": "2026-03-05"
}Built for the whole web
Any site can publish a discovery manifest. Here's who benefits most.
Web developers and site owners
Your site is increasingly accessed by AI agents, not just browsers. Give them a reliable, authoritative source of truth about your site — and control what they know.
SaaS and API teams
AI assistants are becoming purchase research tools. Make your product, pricing, and API discoverable with structured data that agents can act on — not just read.
AI and agent builders
Building with LLMs? Point your agent at /.well-known/ai-me.json for reliable, structured data about any site that has published a manifest — no scraping required.
No major AI agent reads ai-me.json by default — yet. Publishing now is a forward investment: your manifest doubles as structured documentation, prepares you for agent support as it arrives, and contributes to the critical mass that makes the standard worth implementing.
Works alongside existing standards
Complementary to schema.org, llms.txt, MCP, and A2A — not competing.
llms.txt
llms.txt is human-readable guidance for LLMs. AI-Me is machine-readable structured data. Use both — link to your ai-me.json from your llms.txt.
Schema.org
Schema.org is page-level vocabulary for search engines. AI-Me is site-level discovery for agents. Reference your schema.org data via the related field.
MCP
MCP enables tool and data integration. AI-Me handles discovery — point agents to your MCP server via related.mcp.
A2A
A2A enables agent-to-agent communication. AI-Me acts as the discovery layer — link to your A2A agent card via related.a2a.
An open, evolving standard
AI-Me is a community-driven open specification in active draft. The core data model is stable — we're collecting real-world feedback before v1.0.
All changes proposed and reviewed publicly on GitHub.
Spec: CC-BY 4.0. Code: Apache-2.0. No vendor lock-in.
JSON Schema validation. Machine-verifiable conformance.
IANA registration, verification, conformance test suite.
v0.1 relies on HTTPS domain ownership. DNS-based verification and cryptographic signatures are planned for v1.0.
Make your site AI-ready.
Join the sites publishing a discovery manifest. It takes five minutes and costs nothing.