AI changes the equation for documents.

The world runs on documents — tax filings, invoices, statements, disclosures, agreements, applications, and every other structured record that governs how business and government operate.

Today, the infrastructure behind these documents is scattered. Every system reimplements its own templates, validation, and rendering. Schemas live in one place, field logic in another, output formats in a third.

AI agents are now expected to fill, validate, and generate these same documents — but they have no structured context to work with. They infer fields from pixels, guess at validation rules, and reconstruct logic that already exists somewhere else.

What's needed is a single, explicit definition of each document — its fields, rules, parties, and output formats — that both software and agents can share.