# Architecture Overview

**Orby** AI introduces a modular zkML execution stack on Ethereum — optimized for decentralized inference, verifiable outcomes, and on-chain integration with DAOs, DeFi, and Real World Asset (RWA) flows.

**🔁 End-to-End Flow:**

```
User/Protocol → Validator → Inference Node → zk-Proof + Output → Verifier Contract (Ethereum) → On-Chain Log + Action
```

**🧬 zk-Inference Lifecycle:**

1. **User or DAO defines the task** and selects a registered AI model.
2. **Validator dispatches** the job to a zk-compatible inference node.
3. **Inference node executes** the model off-chain and generates a zk-proof.
4. **Validator verifies proof** and logs the result via Ethereum smart contract.
5. **On-chain events** trigger further actions: payouts, governance, RWA royalty distribution, etc.

***

**⚙️ Compatibility:**

* Supports multiple zk formats: Halo2, Risc0, zk-WASM, zkVM.
* Ethereum-native verifier contracts and ERC interfaces.
* Ready for L2 integrations: Optimism, Base, Arbitrum for low-cost inference pipelines.

***

**🌍 RWA Embedding:**

**Orby** AI’s architecture treats every verified inference as an **on-chain income event**. This makes it ideal for:

* **Tokenizing AI model usage rights** as ERC-20/721/1155 assets.
* **Distributing yield** based on zk-verified activity (inference frequency, model accuracy).
* **Embedding RWA logic** directly into protocol-level contracts — no need for external oracles or trust assumptions.

This makes **Orby** AI more than just a zkML layer — it's a programmable revenue fabric for AI-powered RWAs, where every output is a cryptographically secured microtransaction.