> For the complete documentation index, see [llms.txt](https://docs.viralmind.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.viralmind.ai/training-gym/how-it-works.md).

# How It Works

The **Training Gym** is designed to make AI training **intuitive, scalable, and rewarding**. It allows users to **record demonstrations, process them into structured training data, and contribute to AI development**—all while earning $VIRAL for their contributions.

Here’s a step-by-step breakdown of how it works:

## **1. Record Your Demonstration**

* Connect your wallet.
* Open the Training Gym and **start recording** your task.
* Perform actions as you normally would—**clicking, typing, navigating software, and executing workflows.**
* The AI **observes and records** every step in real time, creating a raw demonstration.

📌 **Example:** A user records themselves executing a DeFi swap, clicking through a wallet interface, confirming transactions, and optimizing slippage settings.

***

## **2. Review & Process Your Recording**

* Once a demonstration is recorded, **head to your Recording History** to review it.
* Click **Details** to replay your recording and analyze your steps.
* Press **Process** to break down the recording into **structured steps** AI can learn from.

📌 **Example:** AI segments the DeFi swap process into steps like "Connect Wallet," "Select Token Pair," "Adjust Slippage," and "Confirm Transaction."

***

## **3. Upload to Contribute & Earn Rewards**

* When ready, press **Upload** to submit your demonstration to the ViralMind dataset.
* Your submission will be **evaluated by the data quality agent**, which scores the clarity, effectiveness, and accuracy of your demonstration.
* If your submission meets quality standards, **you receive immediate $VIRAL rewards**.
* **The better your demonstration, the higher your payout.**

📌 **Example:** A well-structured demonstration of executing a trade earns **85% quality**, resulting in a payout of **$0.17 in VIRAL tokens** (with the remaining $0.03 returned to the training pool).

***

## **4. AI Learns & Adapts Over Time**

* Your demonstration is now **part of the open AI dataset**, allowing AI models to learn from real-world human actions.
* The AI **analyzes patterns, decision-making logic, and workflow optimizations**, improving its ability to complete tasks independently.
* As more demonstrations are uploaded, **AI agents continuously refine their accuracy and efficiency.**

📌 **Example:** After multiple high-quality submissions, an AI model **perfects the DeFi swap process**, executing transactions autonomously based on real human behavior.

## **5. Tracking AI Progress in the Skill Tree**

* Every successful demonstration **contributes to AI mastery** in specific skills.
* The **AI Skill Tree** visualizes progress, showing which skills have been mastered and which need more training.
* Users can **focus on growing specific branches**, ensuring AI models develop expertise in high-demand areas.

📌 **Example:** The **DeFi trading branch** of the Skill Tree now shows **95% mastery in executing swaps** but only **60% mastery in liquidity pool interactions**—indicating that more training is needed for staking and farming strategies.

## **Why This System Works**

🔹 **AI Learns Like a Human** – Instead of training on **static datasets**, AI improves **through real-world human interaction**.

🔹 **Open & Decentralized AI Training** – Anyone can **contribute and earn**, making AI development **community-driven**.

🔹 **Real-Time AI Optimization** – AI models **continuously improve** as more demonstrations are uploaded.

🔹 **Monetized AI Contributions** – Users **earn $VIRAL** for training AI, making AI development an **incentive-driven process**.