Skip to main content

Contributor Guide

Welcome to the AIA Chain! This guide will help you get started as a contributor - someone who submits AI inference requests to the network.

Prerequisites​

  • Modern web browser (Chrome, Firefox, or Edge)
  • MetaMask wallet installed
  • Access to Polygon Amoy testnet
  • Small amount of MATIC for gas fees (get from Polygon Faucet)

Quick Start​

1. Connect Your Wallet​

  1. Open the AIA Chain application
  2. Click "Connect Wallet" in the top right
  3. Approve the connection in MetaMask
  4. Ensure you're on Polygon Amoy Testnet

2. Generate Your Keys​

On first use, the application will automatically:

  • Generate an RSA-OAEP key pair
  • Store your private key securely in IndexedDB
  • Register your public key on-chain

You'll see: "Registering contributor public key on-chain..."

3. Submit Your First Request​

  1. Navigate to the Chat interface
  2. Type your question or prompt
  3. (Optional) Adjust settings:
    • Number of validators (default: 1)
    • Priority flag (faster processing, higher fee)
  4. Click Send

4. Wait for Response​

The application will:

  1. Find an active validator
  2. Encrypt your message with the validator's public key
  3. Submit to the blockchain
  4. Poll for the encrypted response
  5. Decrypt and display the AI's answer

Typical response time: 10-30 seconds

Understanding Fees​

Fee Structure​

Total Fee = Base Fee + (Validators × Fee per Validator) + Priority Fee

Current Rates (Amoy Testnet):

  • Base Fee: 0.001 MATIC
  • Per Validator: 0.0002 MATIC
  • Priority: +0.005 MATIC

Examples:

  • 1 validator, normal: 0.0012 MATIC
  • 3 validators, normal: 0.0016 MATIC
  • 1 validator, priority: 0.0062 MATIC

The fee display shows:

  • Estimated fee
  • Whether you can afford it
  • Balance check

Privacy & Security​

Your Data is Protected​

  1. End-to-End Encryption: Your messages are encrypted before leaving your browser
  2. On-Chain Privacy: Only encrypted ciphertext is stored on the blockchain
  3. Targeted Decryption: Only your selected validator can decrypt your message
  4. Response Encryption: The validator encrypts the response for you

Key Management​

Your Keys:

  • Stored in browser IndexedDB
  • Private key is non-exportable
  • Public key registered on-chain

Backup Your Keys (Coming Soon): Currently keys are browser-specific. Clearing browser data will delete your keys.

What Validators See​

❌ Validators CANNOT see:

  • Your original message
  • Messages from other contributors
  • Other validators' responses

✅ Validators CAN see:

  • That you submitted a request (wallet address visible)
  • The encrypted ciphertext
  • Request parameters (number of validators, priority)

Advanced Features​

Multiple Validators​

Request validation from multiple validators for:

  • Redundancy: Backup if one fails
  • Consensus: Compare multiple AI responses
  • Quality: Choose the best answer
validatorCount = 3;  // Request 3 validators

Priority Requests​

Mark requests as priority for:

  • Faster processing
  • Higher validator incentive
  • Better quality responses

Cost: +0.005 MATIC

Conversation History​

Your chat history is stored locally in your browser:

  • Saved automatically
  • Persists across sessions
  • Export/import (coming soon)

Troubleshooting​

"No active validators found"​

Cause: No validators are online Solution:

"Insufficient balance"​

Cause: Not enough MATIC for gas + fees Solution:

  1. Get test MATIC from Polygon Faucet
  2. Ensure you have at least 0.01 MATIC

"Contributor public key not registered"​

Cause: Registration transaction pending or failed Solution:

  1. Check MetaMask for pending transactions
  2. If stuck, try speeding up or canceling
  3. Refresh the page to retry registration

"Decryption failed"​

Cause: Response encrypted for wrong key Solution: This shouldn't happen. If it does:

  1. Clear browser data (will delete keys)
  2. Regenerate keys
  3. Report the issue

Privacy Budget​

Each contributor has a lifetime privacy budget of 10.0 epsilon:

Privacy Remaining = 10.0 - Σ(epsilon used)

Current Implementation:

  • Each request uses ε = 0.001
  • You can make ~10,000 requests

Purpose: In future differential privacy implementations, this prevents privacy leakage from too many queries.

Best Practices​

1. Clear Prompts​

Write specific, well-formed questions:

  • ✅ "What is the capital of France?"
  • ❌ "france capital?"

2. Reasonable Expectations​

AI models running in browsers have limitations:

  • Smaller models (1-7B parameters)
  • Longer response times than cloud services
  • Occasional mistakes or hallucinations

3. Network Reliability​

The network is decentralized and permissionless:

  • Validators can go offline
  • Response quality varies
  • Request multiple validators for critical queries

4. Cost Management​

Monitor your spending:

  • Check fee estimates before submitting
  • Use normal priority for non-urgent requests
  • Request fewer validators when testing

Next Steps​

Support​