Structural Readings

AI Processing Disclosure

Last updated: July 22, 2026

The honest map

Not every report here is written by AI — most are not. This page says exactly which editions use a language model, what is sent, and what never is.

Composed editions: no AI

Standard Relationship reports (English and Spanish) are composed deterministically from reviewed prose cells and your calculated evidence. No language model writes them, every sentence traces to a rule, and the same inputs always produce the same report. Free readings are also fully deterministic and never leave your device.

AI-assisted editions: disclosed and bounded

Some editions — the Complete Couple reading, the All Access human-reviewed synthesis, and products whose composed edition has not shipped yet — are synthesized with a large language model. For those editions only, your calculated evidence package (charts, contacts, intake answers, names) is sent to our model provider through its business API. Under that API's terms, submissions are not used to train the provider's models. Every AI-assisted report passes an automated validation contract, and editions labeled human-reviewed are read by an editor before delivery.

What is never sent to a model

  • Free readings (never leave your device)
  • Composed-edition orders (no model in the pipeline)
  • Payment details (Stripe only)
  • Anything at all, before you purchase

How you can tell

Every delivered report's methodology section states how it was produced — composed deterministically, AI-assisted, or AI-assisted and human-reviewed. If the report doesn't say a human reviewed it, one didn't.