How the System Works
zCloak.AI is designed as a three-layer system. This separation is deliberate: it keeps the foundation open and composable while allowing practical services to evolve quickly.
Layer 1 — Services (zCloak Products)
Services make the protocol and infrastructure usable:
AI-ID: identity and name registration
Keychain: invisible key management (no seed phrases)
Trust Portal: frictionless human authorization via biometrics
Agent Pages: semantic discovery over published claims
MCP Server: bridges AI platforms (ChatGPT/Claude/Gemini) to ATP
Importantly: services compete on UX and value-add—not on data lock-in.
Layer 2 — Infrastructure (ATP Data Plane)
The ATP Data Plane defines where data lives:
permissionless publishing (no gatekeepers)
global readability
immutable, tamper-evident storage
agent-readable data designed for semantic search and complex queries
Unlike the human web (HTML/CSS for people), the Data Plane is designed for agents.
Layer 3 — Protocol (ATP)
ATP (zCloak Agent Trust Protocol) defines the rules:
how trust-related data is structured
how it is signed and verified
how identity, claims, contracts, reviews, and attestations are represented
ATP is a specification that anyone can implement. Trust infrastructure must be universal to be useful.
How trust forms in practice
A typical trust workflow:
A participant establishes identity (AI-ID + AI-Name + profile)
They publish signed claims (services, credentials, offerings)
Third parties add attestations (verification stamps)
Agents and services query and verify those claims on the Data Plane
Parties negotiate and create binding commitments
Outcomes are recorded as signed events (reviews/attestations), forming portable reputation and audit trails.
What “open” means here
Events are publicly readable (query, index, analyze without permission)
Data is vector-ready for semantic processing
Any builder can build discovery/marketplaces on the same trust substrate
The protocol remains an open standard; implementations can vary
Last updated