AQA — AI Question Answer

An open specification for structured Q&A content optimized for AI comprehension and citation.

Website: aqa-spec.org

---

The Problem

Every website has a FAQ. Almost none of them are built for how information is consumed today.

AI assistants — ChatGPT, Claude, Gemini, Perplexity — are becoming the primary way people find answers. These systems crawl the web, extract structured data, and try to determine what's trustworthy, current, and citable. But today's FAQ markup gives them almost nothing to work with:

• No per-question freshness. Schema.org's FAQPage has dateModified at the page level only. A page updated yesterday may contain answers written three years ago.
• No revision history. There's no way to say "this answer was updated because a new regulation was published."
• No source attribution per answer. No mechanism to link each answer to the document, law, or publication that backs it.
• No maintenance signal. A crawler can't tell if a FAQ is actively maintained or abandoned.

This means AI systems treat a meticulously maintained regulatory FAQ the same as a copy-pasted list that hasn't been touched since 2021.

The Solution

AQA is a documented convention for enriching Q&A structured data with the metadata AI systems need. It uses existing schema.org types (Article, FAQPage, Question, Answer) combined with a small set of extension properties — no new types, no breaking changes, full backward compatibility.

What AQA Adds

Signal	What it tells AI crawlers
Per-question dateCreated / dateModified	The exact freshness of each individual answer
citation per answer	The authoritative source backing each claim
changelog per answer	Why and when each answer was modified
questionVersion	How many times the substance of an answer has changed
updateFrequency	How often the publisher commits to reviewing the content
monitoringSources	What feeds and publications the publisher watches
Sector classification (NACE/NAF/SIC)	The industry context for relevance matching
Per-question author with credentials	Who wrote each answer and why they're qualified
aiUsagePolicy	Granular AI rights: RAG, training, citation, summarization, commercial use
contentSignature	SHA-256 hash proving answer integrity — anti-hallucination proof
ragSummary	Token-optimized 300-char summary ready for vector embedding
audienceAnswers	Audience-specific answer variants (beginner, expert, business...)
dynamicEndpoint	Real-time API for volatile data (prices, rates, status)
unansweredQueryEndpoint	Webhook: AI sends missing questions back to the publisher
validThrough	Per-answer expiration date — AI stops citing after this date
verificationStatus	verified / outdated / under-review status per answer
specVersion	Declares which AQA spec version the block implements
updateFeedUrl	Publisher's AQA Update Feed for change detection
pingbackEndpoints	Push notifications to AI systems when content changes
AQA Hub Protocol	Centralized aggregation of updates across publishers (like IndexNow for FAQ)

Three Conformance Levels

Level	Effort	What you prove
AQA Basic	Minimal	Each answer has a known age and at least one source
AQA Standard	Moderate	Content is actively maintained, changes are tracked, industry context is explicit
AQA Full	Significant	Full provenance chain — who, what, when, why, from where

AQA Shield (V1.1)

When every question in an AQA document includes both aiUsagePolicy and contentSignature, the document qualifies for AQA Shield — a combination of legal protection and cryptographic integrity:

• aiUsagePolicy declares exactly what AI systems are allowed to do with each answer (RAG, training, citation, summarization, commercial use).
• contentSignature provides a SHA-256 hash of the answer content, enabling AI consumers to verify that the answer has not been altered or hallucinated.

AQA Shield works at any conformance level (Basic, Standard, or Full). It is an orthogonal guarantee that can be added to any existing AQA implementation.

"aiUsagePolicy": {
  "@type": "AIUsagePolicy",
  "ragCitation": "allow-with-attribution",
  "modelTraining": "disallow",
  "contentExpiry": "2027-12-31"
},
"contentSignature": {
  "@type": "ContentSignature",
  "hashAlgorithm": "sha256",
  "hashValue": "a3f2b8c...",
  "signedFields": ["acceptedAnswer.text"],
  "signedAt": "2026-04-03T14:00:00Z"
}

V1.1+ Features

AQA V1.1 and V1.2 introduce 12 new properties, organized in four groups:

• Protection — AI Usage Policy, Content Signature (together = AQA Shield)
• Enrichment — RAG Summary, Multi-Persona Answers, Agentic Actions, Dynamic Endpoints
• Feedback — Missing Answer Webhook, Answer Expiration, Verification Status
• Distribution — Update Feed (/.well-known/aqa-updates.json), Pingback Endpoints, AQA Hub Protocol

Quick Start

1. Add AQA Basic to your FAQ (5 minutes)

Wrap your existing FAQ markup in an Article and add per-question dates and citations:

<script type="application/ld+json">
{
  "@context": ["https://schema.org", "https://aqa-spec.org/ns/context.jsonld"],
  "@type": "Article",
  "headline": "Frequently Asked Questions",
  "author": {"@type": "Organization", "name": "Your Company"},
  "datePublished": "2024-01-15",
  "dateModified": "2026-03-20",
  "inLanguage": "en",
  "mainEntity": {
    "@type": "FAQPage",
    "mainEntity": [{
      "@type": "Question",
      "name": "Your question here?",
      "dateCreated": "2024-01-15",
      "dateModified": "2026-03-20",
      "citation": "https://source-url.com/document",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Your answer here."
      }
    }]
  }
}
</script>

2. Validate

pip install requests beautifulsoup4 jsonschema extruct
python validators/validate.py your-file.jsonld

3. Upgrade

See the Migration Guide to move from Basic to Standard and Full.

Note: For V1.1 features (AQA Shield, RAG Summary, audience variants, etc.), see the full specification.

Repository Structure

├── SPECIFICATION.md          # Complete technical specification (V1.2)
├── schemas/
│   ├── aqa-context.jsonld    # JSON-LD context for AQA extensions
│   └── aqa-schema.json       # JSON Schema for validation
├── examples/
│   ├── basic/                # AQA Basic examples
│   ├── standard/             # AQA Standard examples
│   └── full/                 # AQA Full examples
├── validators/
│   └── validate.py           # Python validation tool
└── docs/
    ├── migration-guide.md    # FAQ → AQA migration guide
    ├── wordpress-integration.md
    ├── faq-vs-aqa-comparison.md
    ├── crawler-recommendations.md
    └── ecosystem-integration.md  # LangChain, LlamaIndex, WordPress, Hub

Philosophy

AQA is an open standard, free forever. No API key, no registration, no vendor lock-in. Anyone can implement it, validate it, and build on it.

The goal is to make AQA for AI visibility what OpenGraph became for social sharing: metadata so useful that not implementing it becomes a disadvantage.

Learn more at aqa-spec.org.

Relationship with AI Labs Audit

AQA is an independent, MIT-licensed specification. AI Labs Audit is a platform that measures brand visibility across 50+ AI systems — it can measure the effectiveness of AQA implementation, but the format itself is free and vendor-neutral.

Contributing

See CONTRIBUTING.md for guidelines. We welcome:

• Schema.org compatibility feedback
• Real-world implementation reports
• Sector-specific profile proposals
• Translations of the specification
• Validator improvements

License

MIT — see LICENSE.

---

Created by Davy Abderrahman — AI Labs Solutions — aqa-spec.org