Architecture Overview
VeilCore OS is built as a modular stack designed to handle the high-frequency demands of AI agents. The system architecture ensures security, privacy, and speed flow seamlessly from the core kernel to the application layer.
High-Level System Design
The architecture consists of four primary layers:
The Settlement Layer (Solana) This is the foundation. We utilize Solana for final settlement and data availability. By using State Compression, we can store merkle roots of agent states cheaply and efficiently on-chain.
The Consensus Layer (Veil Nodes) A decentralized network of nodes that verify the ZK proofs generated by agents. These nodes ensure that the network remains trustless. They check the math, not the data.
The Execution Layer (ZK-VM) This is where the work happens. Agents execute their code (smart contracts or off-chain logic) in this secure environment. This layer handles the heavy lifting of generating proofs for the "Veil Identity Layer" and "ZK Payment Engine."
The Application Layer (Agent Interface) This is where developers interact with the OS. It includes the
agent_terminal.shinterface, SDKs, and the API for deploying and managing agents.
Data Flow Summary
Input: An agent receives encrypted data.
Process: The agent processes the data in the ZK-VM.
Proof: The VM generates a ZK-Proof confirming the work was done correctly.
Verification: The proof is sent to the Veil Nodes for verification.
Settlement: The result is settled on Solana, and payment is released via the ZK Payment Engine.
Last updated