# How Does it Work?

<figure><img src="https://lh7-rt.googleusercontent.com/docsz/AD_4nXeXt1C3nJbeUI9emCTWyz9MpkWL8SNCu5kXlwWKUBKg58hF3tLaZb_AkjXQ5ddnmRTX-v3hHyeRqU20Uile1hdgXv6QbUbC_o3GgxLB_mx8OHLDMh00arCGZi0sLJLs8tan5UfpZg?key=S_AVxe4Pa4t7GMdXHRa1L20C" alt=""><figcaption></figcaption></figure>

### Scenario Overview

Alice is a virtual character creator. She wants to turn her AI-powered persona, "Cyber Diva" into a monetizable experience. This character has a unique voice and dialogue style. By using **LazAI**, she can offer interactive sessions with "Cyber Diva" through a decentralized system, with payment in **DAT tokens** and complete transparency over how the model was trained and who contributed.

#### Key Participants

* **Creator (Alice):** Prepares and submits datasets to model the character’s voice, tone, and behavior.
* **Validator Nodes (iDAO):** Validate data using ZK proofs to ensure integrity.
* **Inference Providers:** Run GPU-enabled nodes that handle AI requests.
* **Consumers (Users):** Interact with the character by paying in AI gas tokens.
* **Governance DAO:** Adjusts parameters like reward distribution dynamically.

#### Workflow Breakdown

1. **Character Modeling Phase**
   1. Alice preprocesses text/audio of the character.
   2. She uses the **Alith Agent Framework** to submit her dataset in **POV** format.
   3. Submission includes:
      1. Text & vocal tensors
      2. Public dialogue prompts (stored on-chain)
      3. ZK proofs for data integrity
2. **Model Minting Phase**
   1. LazAI’s AI Execution Layer (ExEx) detects the submission.
   2. Using GPU co-processors, it fine-tunes the base model with low-rank adaptation (LoRA).
3. **DAT Anchoring Phase**
   1. A **Semi-Fungible Token (SFT)** is minted, containing:

      1. Metadata of the base model
      2. Access controls
      3. Inference endpoints

      This token acts as a verified, tradable proof of data ownership and model provenance.
4. **Interactive Inference Phase**
   1. A user sends a request to talk with "Cyber Diva".
   2. The model runs on a GPU-powered node and returns the interaction result.
5. **Settlement Phase**
   1. Rewards are auto-distributed as follows:
      1. 70% to Alice
      2. 20% to inference node operators
      3. 10% to the governance pool

#### Why This Matters

This use case shows how LazAI supports real-world applications that rely on:

* Trusted data (via DAT)
* Modular model deployment (via Alith & ExEx)
* Automated, decentralized revenue sharing
* Privacy-preserving inference & transparent governance


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.lazai.network/lazainetwork/user-docs/how-does-it-work.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.