Skip to main content

System Architecture

DPO2U is structured into 5 main layers:

1. Application Layer

The entry interface for users and agents:

  • LGPD Kit Generator: Generates compliant privacy policies and documents.
  • MCP Server DPO2U: Exposes tools (like check_compliance_status) for AI agents to natively consume the protocol.

2. Agent Layer

Autonomous agents that operate the system:

  • Expert Agent: Document generation.
  • Auditor Agent: Compliance analysis and Attestation generation. It has its own DID (did:dpo2u:agent:auditor) and linked wallet.
  • Monitor Agent: Continuous monitoring.

3. Storage Layer

Decentralized and public storage for hashes/proofs:

  • Lighthouse (IPFS): Responsible for document storage, returning immutable CIDs.

4. the host chain Layer

The consensus, privacy, and economic viability layer:

  • the native token: Token for funding operations and paying fees.
  • execution fees: Non-transferable token used to cover operational costs (gas/contract execution).
  • Compact Smart Contracts: On-chain business logic (e.g., ComplianceRegistry.compact).
  • zk-SNARKs: Zero-knowledge mathematical proofs that guarantee validation authenticity without exposing data.

5. Analytical Layer

Computation on encrypted data for advanced analytics:

  • OpenFHE (Fully Homomorphic Encryption) — enables mathematical operations on ciphertext. ML models can train on PII-sensitive datasets (e.g., patient health records linked to a partner institution) without ever decrypting the data
  • Risk dashboards — corporate risk dashboards and multi-institutional data sharing immune to data breaches. The calculation occurs on encrypted data without revealing the original plaintext
Future roadmap

The Analytical Layer is on the roadmap and not yet deployed. It will use OpenFHE and the LEANN vector database (C++ backend) to provide encrypted computations for enterprise risk analytics. See the Whitepaper for the full 5-layer architecture description.

Layer status

Layers 1–4 (Application, Agent, Storage, the host chain) are deployed on the host chain testnet. The Analytical Layer (Layer 5 — OpenFHE) is on the roadmap and not yet deployed.

Smart Contracts

For a detailed breakdown of each smart contracts and the zero-knowledge principle, see Smart Contracts.

Data flow (MVP)

  1. Generation — LGPD Kit produces policy.json (schema dpo2u/lgpd/v1).
  2. Storage — Upload to Lighthouse, which returns a CID.
  3. Analysis — DPO2U Auditor Agent evaluates the document and generates the Attestation.
  4. Registration — Attestation is submitted to the the host chain via a smart contracts.
  5. Consumption — The MCP Server reads the chain and exposes check_compliance_status to other agents.

What's next

  • Smart Contracts — smart contracts and the zero-knowledge principle
  • Agents — the autonomous agents that operate each layer
  • Tokenomics — the the native token/execution fees dual-token economy powering the protocol