Protocol architecture and data pipeline
The wallet is not a separate store. It is the interface that brings each layer's permissions and assets together from the user's side.
DATA LIFECYCLE
From collection to revocation
- 01
Collect
The user provides data, or behavioral data is generated in the app.
- 02
Classify
The Data Intelligence Engine classifies and normalizes data by type, source and risk.
- 03
Refine
Data is cleaned, quality-scored and separated by purpose.
- 04
Consent
The user approves a specific purpose, scope and term.
- 05
Credentialize
The Credential Issuer mints a non-transferable ERC-1155 credential.
- 06
Use & Verify
Conditions are proven with credentials, without exposing raw data.
- 07
Update · Revoke
When data changes or consent is withdrawn, the credential is updated, deactivated or burned.
DATA TYPES
Four kinds of data, four sets of rules
Zero-party data
Information the user intentionally and proactively provides. Example: birth date and time, goals, interests. Principle: collect only what the stated purpose needs. Secondary use needs separate consent.
First-party data
Data SATI observes directly in its own services. Example: activity, meditation use, content usage patterns. Principle: clear disclosure, with separate consent for advertising use.
Inferred data
Probabilistic judgments AI derives from declared or behavioral data. Example: likely interests, mood trends. Principle: treated as probability, not fact, with a confidence level and validity period. Users can challenge it.
Sensitive data
Information with elevated health, counseling or privacy risk. Example: depressive symptoms, counseling transcripts. Principle: isolated storage, explicit consent only, and never used for advertising.
Five purpose-specific vaults
| Vault | Purpose | Principle |
|---|---|---|
| Core Service Vault | Data needed to deliver core services such as fortune and AI companion | Retained within the scope of the core service |
| Personalization Vault | Preferences and memory that support personalization | Used only with the user's permission |
| Wellness Vault | Meditation, mental-wellness and emotional-support data | Isolated storage and explicit consent |
| Advertising Credential Store | Low-risk credentials the user approved for advertising | Only approved credentials are used |
| Compliance Vault | KYC, AML and abuse-prevention records | Retained as the law permits |
DATA CREDENTIAL FIELDS
What a Data Credential contains
| Field | Meaning |
|---|---|
| Credential Type | Non-semantic identifier for the credential type |
| Subject Commitment | A commitment that identifies the holder without exposing the user wallet directly |
| Issuer | The entity that issued the credential |
| Source Class | Declared, behavioral, inferred or partner-originated |
| Assurance Level | The verification level |
| issuedAt / expiresAt | Issuance and expiry timestamps |
| Consent PolicyId | Link to the consent policy the credential is bound to |
| status | Active, Paused, Revoked or Expired |
| Proof Commitment | A proof commitment that does not disclose the underlying source data |
On-chain and off-chain boundary
| Where | What is kept |
|---|---|
| SATI internal network | Credential status, consent records, access rights, proof values and audit records |
| Off-chain, encrypted | Raw data such as birth details, conversation content and behavioral logs |