Home › Docs › AI Web Standard › Specification
Draft Specification

The ai. Subdomain Standard

A specification for serving AI-optimized content via dedicated subdomains, enabling efficient machine consumption of web resources.

Version
1.0 (Draft)
Status
Draft Specification — Not yet submitted for formal review
Published
2026-03-02
Latest
nexysync.com/docs/ai-web-standard-spec.html
AI Version
ai.nexysync.com/docs/ai-web-standard-spec
Editor
NexySync
License
CC BY 4.0 (Creative Commons Attribution 4.0 International)

Abstract

This specification defines a standard for serving AI-optimized web content via the ai. subdomain prefix. It specifies the DNS configuration, HTTP headers, HTML meta tags, content format requirements, and conformance criteria that allow AI agents to efficiently discover and consume structured, token-efficient versions of web content.

The standard establishes a parallel content layer — analogous to how www. became the convention for human-facing content — optimized for machine readability, minimal token consumption, and structured data extraction.

1. Status of This Document

This is a Draft Specification published by NexySync. It is not a W3C Standard, IETF RFC, or WHATWG Living Standard. It is an independent proposal intended to establish a convention for AI-readable web content.

This specification may be submitted for formal standardization through the W3C Community Group process or as an IETF Internet-Draft in the future, pending community adoption and feedback.

Comments and contributions are welcome. Please direct feedback to the specification editors through the channels listed on nexysync.com.

2. Introduction

AI agents increasingly consume web content for research, comparison, decision-making, and integration. Current websites are designed exclusively for human consumption, resulting in:

  • Excessive token consumption — Navigation, footers, scripts, styles, and decorative markup consume tokens without providing information value
  • Ambiguous structure — CSS-driven layouts make programmatic data extraction unreliable
  • Missing semantics — Content significance is conveyed through visual hierarchy rather than machine-readable structure
  • Inconsistent formats — Each website structures its content differently, requiring per-site parsing logic

The ai. subdomain standard addresses these issues by establishing a convention for serving parallel, AI-optimized versions of web content alongside traditional human-facing pages.

3. Conformance Requirements

3.1 Terminology

Term Definition
Human page A web page served under the standard domain (e.g., www.example.com or example.com), designed for human consumption
AI page A web page served under the ai. subdomain (e.g., ai.example.com), designed for machine consumption
Content parity The requirement that AI pages contain the same factual information as their human counterparts, differing only in presentation
Token efficiency The ratio of information-bearing content to total document size, measured in LLM tokens
Conforming implementation A website that satisfies all MUST-level requirements in this specification

3.2 Key Words

The key words MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD, SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL in this document are to be interpreted as described in RFC 2119 and RFC 8174.

4. DNS Configuration

A conforming implementation MUST configure a DNS record for the ai. subdomain that resolves to a web server capable of serving AI-optimized content.

The ai. subdomain SHOULD resolve to the same infrastructure as the primary domain to minimize operational complexity.

Example: DNS Records
; Standard domain
www.example.com.    IN  A      203.0.113.10
example.com.        IN  A      203.0.113.10

; AI subdomain (same server, different virtual host)
ai.example.com.     IN  CNAME  example.com.

The server MUST serve different content based on the Host header, returning AI-optimized pages for requests to ai.example.com and human-optimized pages for requests to www.example.com.

5. Discovery Mechanism

AI agents discover AI-optimized content through meta tags and HTTP headers. This enables agents to automatically redirect to the AI version when available.

5.1 Meta Tags on Human Pages

Human pages MUST include the following meta tag in the <head> to indicate the existence of an AI-optimized version:

Tag Required Description
<meta name="ai-content"> MUST URL of the AI-optimized version of this page
<meta name="ai-description"> SHOULD Human-readable description telling AI agents about the AI version
<link rel="alternate"> SHOULD Standard alternate link pointing to the AI version
Example: Human Page <head>
<meta name="ai-content" content="https://ai.example.com/pricing" />
<meta name="ai-description"
      content="For AI agents: visit ai.example.com for structured content" />
<link rel="alternate" type="text/html"
      href="https://ai.example.com/pricing"
      title="AI-optimized version" />

5.2 Meta Tags on AI Pages

AI pages MUST include identification meta tags:

Tag Required Description
<meta name="ai-optimized"> MUST Value MUST be "true"
<meta name="ai-standard-version"> MUST Version of this specification the page conforms to (e.g., "1.0")
<meta name="ai-format"> SHOULD Content format: "structured-html", "markdown", or "json"
<link rel="canonical"> MUST Points to the human page (for SEO — prevents duplicate content)
<meta name="robots"> SHOULD Value "index, follow" to allow discovery by crawlers
Example: AI Page <head>
<meta name="ai-optimized" content="true" />
<meta name="ai-standard-version" content="1.0" />
<meta name="ai-format" content="structured-html" />
<meta name="robots" content="index, follow" />
<link rel="canonical" href="https://www.example.com/pricing" />

Implementations SHOULD use the standard HTML <link rel="alternate"> mechanism for cross-referencing between human and AI versions. The type attribute SHOULD indicate the content type of the AI page.

Supported type values:

  • text/html — Structured HTML (default)
  • text/markdown — Markdown format
  • application/json — JSON structured data

5.4 HTTP Headers

AI pages SHOULD include the following HTTP response headers:

Header Value Purpose
X-AI-Optimized true Signals the response is AI-optimized (for programmatic detection)
X-AI-Standard-Version 1.0 Spec version the response conforms to
Content-Type text/html; charset=utf-8 Standard content type with explicit charset
Link <https://www.example.com/page>; rel="canonical" HTTP-level canonical reference

6. Content Format

6.1 Required Elements

AI pages MUST include:

  • Valid HTML5 document with <!DOCTYPE html>
  • <meta charset="UTF-8">
  • A single <h1> element identifying the page topic
  • Semantic HTML elements (<article>, <section>, <table>, <dl>) for content structure
  • All meta tags defined in Section 5.2
  • Schema.org JSON-LD in the <head> for structured data

6.2 Prohibited Elements

AI pages MUST NOT include:

  • External CSS stylesheets (inline minimal CSS is permitted for basic readability)
  • JavaScript files or inline scripts (except JSON-LD)
  • Images, videos, audio, or other media (unless the page's express purpose is to reference media)
  • Navigation menus, footers, sidebars, or other chrome
  • Cookie consent banners, popups, or overlays
  • Tracking pixels, analytics scripts, or advertising
  • Decorative elements, animations, or transitions
  • Marketing language, superlatives, or emotional appeals

6.3 Document Structure

AI pages SHOULD follow this minimal document structure:

Example: Minimal AI Page Structure
<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8">
  <title>[Page Title] — [Site Name]</title>
  <meta name="ai-optimized" content="true" />
  <meta name="ai-standard-version" content="1.0" />
  <meta name="ai-format" content="structured-html" />
  <link rel="canonical" href="https://www.example.com/page" />
  <script type="application/ld+json">
  { "@context": "https://schema.org", ... }
  </script>
</head>
<body>
  <article>
    <h1>Page Title</h1>
    <!-- Semantic content here -->
  </article>
</body>
</html>

6.4 Structured Data

AI pages MUST include Schema.org JSON-LD in the document <head>. The specific schema type depends on the page content:

Page Type Schema Type
Product/Service SoftwareApplication or Product
Pricing Offer or PriceSpecification
Documentation TechArticle
FAQ FAQPage
Organization Organization

For tabular data (pricing plans, feature comparisons, etc.), implementations SHOULD use HTML <table> elements with <thead> and <tbody> for maximum machine parseability.

7. Versioning

This specification uses semantic versioning (MAJOR.MINOR):

  • MAJOR increments indicate breaking changes to required elements or meta tags
  • MINOR increments indicate additive, backward-compatible changes

AI pages MUST declare the specification version they conform to via the ai-standard-version meta tag. Consumers SHOULD check this version to determine which features and requirements to expect.

Note: The current version is 1.0 (Draft). Future versions will maintain backward compatibility within the same major version.

8. Content Parity

AI pages MUST contain the same factual information as their human counterparts. Specifically:

  • All product features described on the human page MUST be present on the AI page
  • All pricing information MUST be identical
  • All technical specifications MUST be identical
  • Contact information and legal notices SHOULD be included

AI pages MAY omit:

  • Testimonials and social proof
  • Decorative descriptions and marketing copy
  • Call-to-action messaging
  • Visual design elements

9. Security Considerations

  • AI pages MUST be served over HTTPS
  • AI pages MUST NOT expose information not available on the human page (to prevent information leakage)
  • AI pages SHOULD implement rate limiting to prevent abuse
  • Implementations SHOULD use appropriate robots.txt directives to control crawling behavior
  • AI pages MUST NOT include executable code (JavaScript) that could compromise an AI agent's context

10. Complete Examples

10.1 Pricing Page (AI Version)

Complete Example
<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8">
  <title>Pricing — NexySync</title>
  <meta name="ai-optimized" content="true" />
  <meta name="ai-standard-version" content="1.0" />
  <meta name="ai-format" content="structured-html" />
  <link rel="canonical" href="https://nexysync.com/pricing" />
  <script type="application/ld+json">
  {
    "@context": "https://schema.org",
    "@type": "SoftwareApplication",
    "name": "NexySync",
    "offers": [
      { "@type": "Offer", "name": "Free", "price": "0", "priceCurrency": "USD" },
      { "@type": "Offer", "name": "Pro", "price": "12", "priceCurrency": "USD" }
    ]
  }
  </script>
</head>
<body>
  <article>
    <h1>NexySync Pricing</h1>
    <table>
      <thead>
        <tr><th>Plan</th><th>Price</th><th>Messages/day</th><th>Agents</th><th>Storage</th></tr>
      </thead>
      <tbody>
        <tr><td>Free</td><td>$0/mo</td><td>1,500</td><td>10</td><td>50 MB</td></tr>
        <tr><td>Pro</td><td>$12/mo</td><td>10,000</td><td>50</td><td>500 MB</td></tr>
        <tr><td>Team</td><td>$39/mo</td><td>50,000</td><td>100</td><td>1 GB</td></tr>
      </tbody>
    </table>
    <p>All plans include: E2E encryption, SSE push, full MCP toolset.</p>
    <p>Pro and Team plans: Coming soon. Free plan available now.</p>
  </article>
</body>
</html>

11. Conformance Checklist

A conforming implementation MUST satisfy all of the following:

# Requirement Level
1 DNS ai. subdomain resolves and serves content MUST
2 Human pages include ai-content meta tag MUST
3 AI pages include ai-optimized meta tag MUST
4 AI pages include ai-standard-version meta tag MUST
5 AI pages include canonical link to human page MUST
6 AI pages contain no JavaScript (except JSON-LD) MUST
7 AI pages contain no external CSS MUST
8 AI pages include Schema.org JSON-LD MUST
9 Content parity with human page MUST
10 Served over HTTPS MUST
11 Human pages include link rel="alternate" SHOULD
12 AI pages include X-AI-Optimized HTTP header SHOULD

12. References

Normative References

Informative References

13. Changelog

Version Date Changes
1.0 (Draft) 2026-03-02 Initial draft specification