Data Types and Purpose-Based Isolation

SATI divides data into four types and applies a different confidence level and handling principle to each.

Type	Definition	Examples	Default Principle
Zero-party	Information the user voluntarily provides with knowledge of the specific service purpose	Date and time of birth, birthplace, gender, interests, preferences	Explicit purpose-specific consent; editable by the user
First-party	Information observed through direct interaction with SATI services	Feature usage, content choices, reward history, ad responses	Minimal collection, retention limits, access control
Inferred	Attributes computed by the system by combining declared and behavioral data	Interest categories, preference shifts, recommendation fit	Labeled as inferred, with confidence level and right to dispute
Sensitive	Information requiring a high level of protection, such as health, mental care and crisis signals	Counseling conversations, emotional state, risk signals	Isolated from advertising; separate consent; enhanced access control

Inferred data is never treated as equivalent to fact. Credentials include the sourceClass, assuranceLevel, issuedAt and expiresAt fields to express the difference between declared, observed and inferred data and the passage of time. Low-confidence inferences are never used as the sole basis for ad eligibility or high-risk AI actions.

Data is isolated by purpose across five stores:

Vault	Key Data	Advertising Use
Core Service Vault	Date and time of birth, birthplace, gender, interests, preferences	Prohibited without separate consent
Personalization Vault	Preferences, feedback and long-term personalization context	Only general-interest credentials, on a limited basis
Wellness Vault	Meditation, wellness and mental care data	Prohibited in principle
Advertising Credential Store	Non-identifying credentials proving ad eligibility	Permitted
Compliance Vault	KYC/KYB, sanctions and fraud screening results	Prohibited