Customization Options
Fine-tune your integration’s behavior to match your brand voice and specific use cases through advanced prompt engineering and configuration options.System Prompts
System prompts define how your integration behaves and responds to users.Default System Prompt
Custom Prompt Engineering
1
Define Role & Personality
Establish your AI’s role and personality:
2
Set Response Guidelines
Define how the AI should structure responses:
3
Handle Edge Cases
Define behavior for edge cases:
Response Templates
Create standardized responses for common scenarios:Greeting Templates
Fallback Templates
Behavior Settings
Response Length
Control how verbose your integration’s responses are:- Concise (50-150 words): Quick, direct answers
- Detailed (150-300 words): Comprehensive responses with context
- Comprehensive (300+ words): In-depth explanations with examples
Personality Settings
| Setting | Conservative | Balanced | Friendly |
|---|---|---|---|
| Tone | Professional, formal | Professional, approachable | Casual, warm |
| Emoji Usage | None | Occasional | Frequent |
| Humor | None | Light | Playful |
| Formality | High | Medium | Low |
Context Handling
- Memory Length: Number of previous messages to remember (1-10)
- Context Relevance: How strictly to stay within knowledge base
- Topic Switching: How to handle when users change subjects
Advanced Prompt Examples
E-commerce Support
SaaS Product Support
Testing Your Customizations
A/B Testing Prompts
- Create Variations: Develop 2-3 different prompt versions
- Test Systematically: Use the built-in chat interface to test various scenarios
- Measure Performance: Track user satisfaction and task completion
- Iterate: Refine based on feedback and performance data
Quality Assurance Checklist
- Responses stay on-brand and on-topic
- Information provided is accurate and up-to-date
- Tone matches your organization’s voice
- Edge cases are handled gracefully
- Fallback responses are helpful and actionable
- No sensitive information is accidentally shared
Best Practices
- Start Simple: Begin with basic prompts and gradually add complexity
- Be Specific: Clear, specific instructions work better than vague guidelines
- Include Examples: Show the AI what good responses look like
- Regular Updates: Review and update prompts as your product/service evolves
- User Feedback: Incorporate user feedback to improve responses over time