Introducing AIOX Capsules: The AI-Native Content Format

HTML tells a browser how to render a page. Schema.org tells a search engine what type of thing this is. Robots.txt tells a crawler what it can fetch. But none of them tell an AI assistant everything it needs to use the content intelligently. That gap is what the AIOX Capsule fills.

What a Capsule is

An AIOX Capsule is a self-contained JSON document that packages:

• The content itself, in canonical text form (stripped of layout cruft).
• The schema — Schema.org types plus AIOX-namespaced extensions.
• The intent — informational, commercial, navigational, transactional.
• The audience — declared target reader (developers, shoppers, executives, general).
• The license — per-bot, per-use rights declarations.
• The authorship — author identity, publisher identity, dates.
• A cryptographic signature proving the publisher attests to all of the above.

Every Capsule lives at a stable URL (typically /aiox-capsule/{slug}/) and is referenced from a sitewide manifest at /aiox.json. AI crawlers can fetch the manifest to discover every Capsule on the site, then pull each one in a single HTTP request — much faster than scraping HTML.

Why a new format?

The existing formats were designed for different jobs at different times:

• HTML: layout for browsers (1990s).
• RSS / Atom: syndication for readers (2000s).
• Schema.org: classification for search engines (2010s).
• OpenGraph: previews for social platforms (2010s).
• OpenAPI: specifications for APIs (2010s).

None of them natively express the “intent + audience + license + authenticity” combination that LLMs need. Bolting AIOX-style extensions onto Schema.org would work mechanically but it would muddy the spec — schema.org was deliberately designed for SERP classification, not for AI-content interchange. Building a clean, parallel format gives us room to evolve.

It’s an open spec

The AIOX Capsule format is documented publicly. Any tool — open source or commercial — can read and write Capsules. There’s no licensing fee for the spec. The reference implementation is the AIOX Suite plugin, but adapters for Drupal, Ghost, headless CMSes, and static-site generators are on the roadmap (and the community is welcome to contribute).

This is intentional. A standard that only one vendor supports is just a feature, not a standard. We want AIOX Capsules to become as broadly adopted as Schema.org — used by AI labs, content platforms, and integration tools, with us as one implementation among several.

What it looks like

Here’s a stripped-down Capsule for a hypothetical blog post:

{
  "@context": ["https://schema.org", "https://aiox.org/v1"],
  "@type": "Article",
  "aiox:capsuleId": "cap_abc123",
  "aiox:version": "1.0",
  "headline": "How to Train Your AI to Be a Better Marketer",
  "datePublished": "2026-03-15",
  "author": { "@type": "Person", "name": "Jane Smith" },
  "publisher": { "@type": "Organization", "name": "Acme Inc" },
  "aiox:intent": { "primary": "informational", "secondary": "commercial" },
  "aiox:audience": ["marketing-leaders", "ai-practitioners"],
  "aiox:license": {
    "default": "indexing-and-citation",
    "perBot": { "GPTBot": "indexing-only", "PerplexityBot": "indexing-and-citation" }
  },
  "aiox:signature": "ed25519:eyJ...."
}

That’s enough for an AI system to decide what to do with the content, how to attribute it, and whether it’s authentic.