Structured Data for AI Agents: Beyond Schema.org

Schema.org is brilliant for what it was designed for: helping search engine crawlers understand the content of HTML pages. When you add LocalBusiness schema to your website, Google can extract your business name, address, hours, and ratings from the page markup.

But AI agents do not crawl HTML pages. They make API calls. They use MCP tool connections. They query structured data sources. Asking an AI agent to parse Schema.org from an HTML page is like asking a modern application to parse data from a printed PDF — technically possible, but architecturally wrong.

An ACN content object is not an HTML annotation. It is a standalone, typed data structure with approximately 30 fields purpose-built for AI consumption:

• Identity fields — id, slug, domain, name (unique identification)
• Location fields — address (street, city, state, zip, lat, lng) for geographic queries
• Operations fields — hours, services, priceRange, paymentMethods for practical questions
• Semantic fields — facts, qa, description for natural language citation
• Trust fields — provenance chain (hash, signature, verified status, public key)
• Intelligence fields — citation count, completeness score, last cited timestamp
• Attribute flags — 18 boolean fields (walkIns, wheelchair, delivery, etc.) for precise filtering

This is what "structured data for AI agents" actually means — not annotations on a web page, but a typed data object designed from the ground up for machine consumption.