Overview
ZapDev’s conversational interface allows you to iteratively build and improve your application through natural language. Each message you send triggers an AI agent that understands context, generates code, and provides interactive previews.How AI Chat Works
The AI chat operates in a continuous conversation loop:- You send a message describing changes or new features
- AI agent processes your request with full project context
- Code is generated in isolated sandbox environments
- Preview updates in real-time with the new changes
- Iterate based on the results
Sending Messages
Open Your Project
Navigate to any existing project to access the message interface at the bottom of the screen.
Compose Your Message
In the message textarea, describe what you want to add or change:Good examples:
Enhance Your Prompt (Optional)
Click the sparkles ✨ icon to automatically enhance your prompt with AI:Before:After enhancement:
Add Visual Context (Optional)
Attach images to provide visual guidance:
- Click the image icon to upload screenshots or mockups
- Add multiple images per message
- AI analyzes images to understand design requirements
Message Types
ZapDev handles different types of requests:Feature Additions
Add new functionality to your application:Modifications
Update existing features or styling:Bug Fixes
Address issues in the generated code:Refactoring
Improve code quality or architecture:AI Model Selection
Choose different AI models based on your task complexity:- Auto (Recommended)
- Claude Haiku 4.5
- Gemini 3 Pro
- GPT-5.1 Codex
- Z-AI GLM 4.7
- Kimi K2.5
Automatically selects the best model for your specific request.Best for: Most tasks, hands-off approach
Message Workflow
Understand what happens when you send a message:Message Status Indicators
Messages display different states:| Status | Indicator | Meaning |
|---|---|---|
| Pending | ⏳ Spinner | Message queued for processing |
| Streaming | 💬 Animated | AI is actively generating code |
| Complete | ✅ Checkmark | Code generated successfully |
| Error | ❌ Error icon | Generation failed |
Conversation Context
The AI maintains context throughout your conversation:- Previous messages - All prior requests and responses
- Current code - The complete state of your application
- Framework - Your selected framework and conventions
- Attachments - Images and imports from earlier messages
Attachments and Imports
Image Attachments
Attach visual references to guide AI:Upload Images
Choose one or more images from your device:
- Screenshots
- Design mockups
- UI inspiration
- Reference materials
Design Imports
Import from external sources:- Figma - Import designs directly from Figma files
- GitHub - Import existing repositories
Usage and Credits
Each message consumes 1 credit from your daily allowance:The usage indicator at the top of the message form shows:
- Remaining credits
- Time until next credit resets
- Your current plan
Best Practices
Write Clear Requests
Break Down Complex Changes
Instead of: “Build a complete user authentication system with profile management, password reset, email verification, and social login”Try:
- “Create a login form with email and password fields”
- “Add password reset functionality”
- “Implement email verification flow”
- “Add Google OAuth integration”
Provide Specific Feedback
Instead of: “The button doesn’t look right”Try: “The submit button should be larger with rounded corners and use the primary brand color #3B82F6”
Use Examples
Reference specific websites or patterns:“Create a pricing table similar to Stripe’s pricing page with three tiers”
Troubleshooting
Message Won’t Send
- Check credits - Ensure you have available credits
- Verify prompt length - Messages must be 1-10,000 characters
- Wait for uploads - Let image uploads complete
No Code Generated
- Check error messages - Look for specific error details
- Simplify request - Try breaking into smaller steps
- Try different model - Switch to a different AI model
Unexpected Results
- Add more context - Provide additional details or examples
- Use images - Upload visual references
- Iterate - Send follow-up messages to refine