Skip to content

Using AI Chat

Master the AI chat feature and start earning while you learn, code, and create!

🎯 What is AI Chat?

Kawai's AI Chat gives you access to powerful language models that can help you with:

  • 💻 Code: Write, debug, and explain code
  • ✍️ Writing: Create content, emails, documents
  • 🎓 Learning: Explain concepts, answer questions
  • 🔍 Research: Summarize information, analyze data
  • 🎨 Creative: Brainstorm ideas, write stories

Best part? You earn 5% cashback on every request!

🚀 Getting Started

Opening the Chat

  1. Launch Kawai Desktop App
  2. Main interface = Chat (default view)
  3. Click "New Chat" to start fresh

Your First Message

  1. Type your question or request
  2. Press Enter or click Send
  3. Wait for AI response (usually 2-10 seconds)
  4. Continue the conversation!

Example:

You: What is blockchain?
AI: Blockchain is a distributed ledger technology...

💬 Chat Features

Multi-Turn Conversations

The AI remembers your conversation context.

Example:

You: Write a function to calculate Fibonacci numbers
AI: [provides Python code]

You: Can you make it more efficient?
AI: [provides optimized version with memoization]

You: Now convert it to JavaScript
AI: [provides JS version]

Tip: Each conversation has full context - no need to repeat yourself!

Code Highlighting

Code in responses is automatically highlighted with syntax.

Supported languages:
- Python, JavaScript, TypeScript
- Go, Rust, C++, Java
- HTML, CSS, SQL
- And 50+ more!

Markdown Support

AI responses support full Markdown formatting:

  • Bold and italic text
  • Lists (bullet and numbered)
  • Tables
  • Inline code and code blocks
  • Links and images
  • Headers and sections

Copy & Export

  • Copy code: Click the copy icon on code blocks
  • Copy message: Right-click any message
  • Export chat: Save conversations to file (coming soon)

🎨 Chat Management

Creating New Chats

Click "New Chat" to start a fresh conversation.

When to start new:
- Different topic/project
- Conversation getting too long
- Want fresh context

Chat History

All your chats are saved automatically.

Accessing history:
1. Click the history icon (sidebar)
2. Browse your past conversations
3. Click any chat to continue

Features:

  • ✅ Auto-saved (no manual save needed)
  • ✅ Searchable (find old conversations)
  • ✅ Organized by date
  • ✅ Unlimited storage

Renaming Chats

  1. Click the pencil icon next to chat title
  2. Enter new name
  3. Press Enter

Tip: Use descriptive names like "Marketing Campaign Ideas" or "Python Tutorial".

Deleting Chats

  1. Right-click the chat
  2. Select "Delete"
  3. Confirm

Note: Deleted chats are gone forever. Be sure!

🧠 Model Selection

Available Models

Currently using local LLM models via llama.cpp:

  • Fast Mode: Quick responses, lower quality
  • Balanced: Good speed and quality (default)
  • Quality Mode: Best responses, slower

Switching Models

  1. Click settings icon in chat
  2. Select "Model"
  3. Choose your preferred model

When to use each:
- Fast: Quick questions, simple tasks
- Balanced: Most use cases
- Quality: Complex reasoning, long content

📚 Knowledge Base Integration

What is Knowledge Base?

Upload documents so AI can reference your files.

Use cases:
- Code repositories → AI helps with your codebase
- Documentation → AI answers from your docs
- Research papers → AI summarizes your research
- Business reports → AI analyzes your data

Adding Documents

  1. Click "Knowledge Base" in sidebar
  2. Click "Upload"

  3. Select files:
    - PDF documents
    - Word files (.docx)
    - Markdown (.md)
    - Text files (.txt)
    - Code files

  4. Wait for indexing (automatic)

  5. Ask questions about your docs!

Using Knowledge Base

Once documents are uploaded, just ask questions:

Example:

You: What were the Q3 sales numbers?
AI: According to the Q3_Report.pdf you uploaded,
     sales were $2.5M, up 25% from Q2...

Pro tip: AI automatically knows which documents are relevant!

Managing Documents

  • View all: Click "Knowledge Base" to see all docs
  • Remove: Click X on any document
  • Update: Upload newer version (same filename)

💰 Earning While You Chat

Use-to-Earn (5% Cashback)

Every AI request earns you KAWAI tokens!

Example:

Request cost: 0.05 USDT
Cashback: 0.05 × 5% = 2.5 KAWAI tokens

The more you use, the more you earn!

Tracking Your Earnings

  1. Go to Wallet → Rewards → Use-to-Earn
  2. See your accumulated rewards
  3. Claim weekly after settlement

Cost Breakdown

Average costs:
- Simple question: 0.01-0.02 USDT
- Code generation: 0.03-0.05 USDT
- Long analysis: 0.05-0.10 USDT

With 100 USDT balance:
- ~2,000-10,000 AI requests
- Earn 250-500 KAWAI tokens back
- Net cost: 95 USDT (after cashback)

🎯 Pro Tips

Getting Better Responses

1. Be specific

❌ "Write code"
✅ "Write a Python function that calculates compound interest"

2. Provide context

❌ "Fix this"
✅ "Fix this JavaScript function - it should validate email addresses but returns false for valid emails"

3. Use examples

✅ "Create a function like this: input: [1,2,3], output: 6"

4. Break down complex tasks

✅ "First, explain the algorithm. Then, show the code."

Using System Messages

System messages guide the AI's behavior.

Examples:

System: You are a Python expert. Be concise.
System: Explain to a beginner. Use simple terms.
System: Act as a code reviewer. Be critical.

How to use:
1. Click settings → System Message
2. Enter your instructions
3. Applies to entire chat

Iterating on Responses

Don't hesitate to refine!

Example flow:

You: Write a blog post about AI
AI: [provides 500 words]

You: Make it shorter and more casual
AI: [provides 300 words, casual tone]

You: Add 3 examples
AI: [adds examples]

🔧 Advanced Features

Code Execution (Coming Soon)

Run code directly in the chat:

print("Hello, Kawai!")
# Output: Hello, Kawai!

Image Understanding (Coming Soon)

Upload images and ask questions:

You: [uploads diagram] Explain this architecture
AI: This diagram shows a microservices architecture...

Voice Input (Coming Soon)

Speak your questions instead of typing.

Collaborative Chats (Coming Soon)

Share chats with team members.

📊 Performance & Limits

Speed

  • Local inference: 20-50 tokens/second
  • Average response time: 2-10 seconds
  • Long responses: up to 30 seconds

Factors affecting speed:
- Your hardware (CPU/GPU)
- Model size
- Response length
- System load

Token Limits

  • Input: ~4,000 tokens per message
  • Output: ~2,000 tokens per response
  • Context window: ~8,000 tokens total

What's a token? ~4 characters or 0.75 words.

Rate Limits

  • No hard limits on chat requests
  • Limited only by your USDT balance
  • Fair usage policy applies

🆘 Troubleshooting

AI Not Responding

Solutions:
1. Check internet connection
2. Verify USDT balance > 0
3. Restart the app
4. Check logs: Settings → Advanced → View Logs

Slow Responses

Solutions:
1. Switch to Fast Mode
2. Ask shorter questions
3. Close other apps (free up RAM)
4. Check system resources

Unexpected Responses

Solutions:
1. Rephrase your question
2. Add more context
3. Try a different model
4. Start a new chat (fresh context)

"Insufficient Balance"

Solution:
1. Check your USDT balance
2. Deposit more USDT
3. Or claim free trial

Context Confusion

If AI seems confused:
1. Start a new chat
2. Or explicitly reset: "Forget the above, new topic..."
3. Be more explicit with context

💡 Use Cases & Examples

For Developers

Code review:

You: Review this code: [paste code]
AI: [provides detailed review with suggestions]

Debugging:

You: Why is this throwing a null pointer exception?
AI: [analyzes and explains the issue]

Documentation:

You: Write documentation for this function
AI: [generates comprehensive docs]

For Content Creators

Blog posts:

You: Write a 500-word blog about Web3
AI: [provides article]

Social media:

You: Create 5 tweets about our new product
AI: [provides tweet thread]

Video scripts:

You: Script for a 2-minute YouTube intro
AI: [provides script with timing]

For Students

Explanations:

You: Explain quantum computing like I'm 10
AI: [simple explanation]

Study help:

You: Quiz me on photosynthesis
AI: [asks questions]

Essay help:

You: Outline for essay on climate change
AI: [provides structured outline]

For Business

Emails:

You: Write a professional follow-up email
AI: [drafts email]

Analysis:

You: Analyze this sales data [paste data]
AI: [provides insights]

Proposals:

You: Create project proposal outline
AI: [provides structure]

✅ Checklist

For best results:

  • Have USDT balance ready
  • Understand token costs
  • Know how to use Knowledge Base
  • Practice with clear, specific prompts
  • Track your cashback rewards
  • Save important chats

🚀 Next Steps

Explore more features:

  1. Image Generation - Create AI art
  2. Rewards Dashboard - Track earnings
  3. Referral Program - Earn by sharing
  4. Trading - Trade KAWAI tokens

Need help? Check FAQ or join Discord.