Introducing the AI Discovery Standard: An Open Specification for How the Web Communicates with AI
Today we announce the AI Discovery Standard — an open specification (CC-BY-4.0) that gives every website a structured, machine-readable way to tell AI agents who it is, what it does, and how to verify the answer.
The standard defines a single endpoint at /.well-known/ai that any AI agent can read. It's the missing protocol layer between the web and AI — the equivalent of what robots.txt did for search engines, but for intelligent agents that need to understand, not just crawl.
The Problem
50 billion AI crawler requests hit the web every day. When ChatGPT, Perplexity, or Google's AI Overview answers a question about your company, it's guessing from scraped HTML. The hallucination rate for company and person queries is 33-51%. The cost of AI hallucinations reached $67.4 billion in 2024.
The web has standards for search engines, for security, for identity. It has no standard for AI. 200 million websites are invisible to AI agents — not because they lack content, but because they lack a machine-readable format for their identity.
The Standard
The AI Discovery Standard provides a three-tier architecture:
- Discovery (
/.well-known/ai) — Organizational identity, mission, products, team, contact, and core concepts - Knowledge — Deep organizational encyclopedia for comprehensive AI understanding
- Feed — AI-optimized news and updates, structured for machine consumption
Every manifest can be cryptographically signed with a SHA-256 content hash and blockchain-anchored Digital Name. AI agents verify origin. Users decide trust. We call this principle Origin, Not Trust — the same measurement approach that the Trusted Platform Module brought to computer security 25 years ago, now applied to the web.
Founding Members
The standard launches with [N] founding members:
- Rootz Corp — Standard author and reference implementation
- Made in USA Inc. — Enterprise certification and government deployment through the Veritize™ platform
- [Additional founding members to be announced]
"In 1993, Mosaic gave humans a browser for the web. In 2026,
/.well-known/aigives AI agents one. Every AI agent is already a browser — this standard gives them a URL bar."— Steven Sprague, Chair, AI Discovery Standard
Getting Started
The standard is free to implement, today:
- WordPress sites: Install the Rootz AI Discovery plugin — generates a complete three-tier implementation in 2 minutes
- Any website: Create a 10-line JSON file at
/.well-known/ai— see the Getting Started guide - Enterprise: Contact Made in USA Inc. for managed implementation and certification
Scan any site now at rootz.global/ai-discovery.
Join Us
The founding member enrollment window is open. Organizations that join now shape the specification, receive Steering Committee representation, and are permanently recognized as founders of the standard.