AEO DNS Record Standard

The DNS-layer trust signal for AI answer engines

---

The Problem

AI answer engines — ChatGPT, Perplexity, Google AI Mode, Claude, Copilot — cite web sources when answering questions. Today they scrape and guess. They have no reliable way to know:

• Whether a domain's data is verified by any third party
• Where to find the canonical, structured, up-to-date fact record
• Whether the address, hours, and pricing data are current
• Whether the domain owner authorized the data being cited
• How to retrieve verified data when a website is absent or broken

The AEO DNS record solves all five. It is the SPF/DKIM/DMARC equivalent for AI citation — a DNS-layer trust anchor any agent can check before grounding an answer on a domain's data.

---

Record Format

Place the record at the _aeo subdomain, following the DMARC _dmarc. convention:

_aeo.yourdomain.com  IN  TXT  "v=AEO1; id=TOKEN; catalog=URL; tier=TIER; ts=DATE"

The _aeo. prefix keeps the record clean, avoids root TXT clutter, and mirrors the established DNS authentication pattern. AI agents check _aeo.[domain] the same way mail servers check _dmarc.[domain].

---

Tag Reference

Required Tags

Tag		Description
v=AEO1	REQUIRED	Version identifier. Always AEO1 for this version. Future: AEO2, AEO3.
id=TOKEN	REQUIRED	Verification token issued by the AEO record authority. Proves domain owner authorized the catalog entry. Format: grd_[alphanumeric, 12+ chars]. Tokens issued at groundedaeo.com. Self-generated tokens are invalid.
catalog=URL	REQUIRED	HTTPS URL to the domain's verified fact catalog. Must return structured JSON-LD or the Grounded AEO catalog format.
tier=TIER	REQUIRED	Verification depth. Values: listed | verified | grounded
ts=YYYYMMDD	REQUIRED	Date of most recent verification. Records older than 90 days are considered potentially stale.

Optional Tags

Tag		Description
method=LIST	OPTIONAL	Preferred data retrieval order for AI agents. Comma-separated. Example: method=mcp,api,crawl. Values: mcp | api | crawl | manual. Tells agents HOW to get data, not just where it is.
sig=HASH	OPTIONAL	SHA-256 signature of catalog content. Allows integrity verification without an API call. Grounded tier only.
nap=STATUS	OPTIONAL	NAP (Name/Address/Phone) consistency status. Values: confirmed | contested | unverified. confirmed = verified identical across website, Google Business Profile, Yelp, Apple Maps.
area=GEO	OPTIONAL	Canonical service area for local businesses. Format: area=Boulder,CO;Denver,CO or radius:25mi:40.015,-105.270. Lets agents answer "does this business serve [location]?"

Tier Definitions

Tier	Verification Method	Refresh Cadence
listed	Self-declared, auto-crawled	Monthly
verified	Human review gate, cross-spot-checked	14 days
grounded	Full cross-source corroboration + cryptographic signature	7 days

---

Full Example — Grounded Tier, Local Business

_aeo.joeselectric.com  IN  TXT
  "v=AEO1;
   id=grd_x7kP2mQnR4vW;
   catalog=https://groundedaeo.com/c/joeselectric.com;
   tier=grounded;
   ts=20260629;
   sig=a3f1b2c4d5e6f7a8b9c0d1e2f3a4b5c6d7e8f9a0b1c2d3e4f5a6b7c8d9e0f1a2;
   nap=confirmed;
   area=Boulder,CO;Longmont,CO;Lafayette,CO;Louisville,CO;
   method=mcp,api,crawl"

---

/.well-known/aeo.json — The Crawl-Time Bridge

DNS is the authoritative long-term layer. The /.well-known/aeo.json file is the on-ramp that works with every crawler today — no engine-side code change required.

Any crawler following web standards (security.txt, WebFinger, ACME challenges) already knows to check /.well-known/ paths. When a crawler finds this file, it gets pointed to verified structured data instead of scraping stale HTML.

https://yourdomain.com/.well-known/aeo.json

{
  "v": "AEO1",
  "catalog": "https://groundedaeo.com/c/yourdomain.com",
  "mcp": "mcp://api.groundedaeo.com",
  "api": "https://api.groundedaeo.com/v1/companies/yourdomain.com",
  "tier": "grounded",
  "verified_at": "2026-06-29",
  "method_preference": ["mcp", "api", "crawl"]
}

The DNS record and /.well-known/ file return identical data. DNS is authoritative; the file bridges the gap while adoption builds.

---

Discovery Hierarchy

Agents and crawlers should work down this list and stop at the first hit:

1. /.well-known/aeo.json Crawl-time bridge — works with all crawlers today, no adoption needed
2. _aeo.[domain] TXT DNS authoritative record — long-term standard
3. llms.txt pointer In-file education for crawlers already reading llms.txt
4. MCP directory For agent builders installing tools
5. HTML crawl Last resort — unverified, potentially stale

---

Implementation Guide

For Domain Owners

1. Sign up at groundedaeo.com
2. Complete your catalog entry — facts, address, service area, hours
3. Copy the TXT record value from your dashboard
4. Add to DNS at your registrar under _aeo.[yourdomain]
5. Grounded polls DNS hourly — record verified within 24 hours
6. Your /.well-known/aeo.json file is generated and served automatically

For AI Crawlers and Agents

1. Check https://domain/.well-known/aeo.json first
2. If absent, query _aeo.[domain] TXT
3. If v=AEO1 present: follow catalog= URL for verified structured facts
4. Check ts= for freshness — reject if older than 90 days
5. Respect method= preference order for data retrieval
6. For local queries: parse area= to answer service-area questions
7. nap=confirmed signals triple-verified address — trust for location answers

For Tool Builders

This spec is open. You may check for v=AEO1 records in your own tools. Do not issue id= tokens — only the AEO record authority (groundedaeo.com) issues tokens. A record with a self-generated id= is invalid.

Open source checker:

npm install aeo-record

import { checkAeo } from 'aeo-record'
const result = await checkAeo('domain.com')
// Checks /.well-known/aeo.json first, then _aeo DNS TXT
// Returns: { catalog, mcp, tier, verified_at, method_preference }

---

MCP Deployment

The AEO record's mcp= tag points to a live MCP server agents can call directly for real-time verified facts. Businesses do not need to build or host anything — Grounded handles this by default.

Tier	Endpoint	Effort
Shared	mcp://api.groundedaeo.com	Zero — all catalog companies
Namespaced	mcp://domain.groundedaeo.com	Zero — Grounded tier, auto-provisioned
Self-hosted	One-click deploy to Netlify / Railway / Cloudflare	One click
SDK	@groundedaeo/mcp-sdk with live data overrides	Developer

Grounded monitors all active MCP endpoints every 15 minutes. If an endpoint goes down, the AEO record's method_preference automatically shifts to ["api","crawl"] so agents continue receiving data.

---

Prior Art

Standard	What it does	Gap vs. v=AEO1
IETF AIPREF / ai.txt	AI training preferences	Not citation verification, not DNS layer
OpenAI domain verification	Platform domain ownership	Not an open standard, not citation
llms.txt	File-based content index	No DNS, no verification, no signatures
LLM-LD (llmld.org)	Markup standard	No DNS layer
Web Bot Auth	Cryptographic agent identity	Not citation, 12+ months from standard-track

---

Relationship to Other Standards

Standard	Layer	Relationship
robots.txt	File	Controls crawler access — orthogonal
llms.txt	File	AI crawler content index — orthogonal, complementary
schema.org	Page	Structured data within pages — orthogonal, page-level
/.well-known/aeo.json	File	Crawl-time bridge — this spec, file layer
v=AEO1 DNS TXT	DNS	Authoritative trust anchor — this spec, DNS layer

All five coexist. The AEO layers authenticate the others — they tell a crawler "verified facts for this domain are at [URL], issued [date], depth [tier], and here's the fastest way to get them."

---

---

Adopt the Standard

Whether you're an AI engine team, an agent framework developer, or a business that wants verified citations — we want to hear from you.

Or drop us a note directly

Name

Email

I am a... AI engine / search team Agent framework / tool builder Business wanting verification Developer Researcher / standards body Other

Message

Send →

---