AI Behavior Configuration
1. AI Response Style
- Choose between response modes such as:
- Professional (formal, concise)
- Conversational (friendly, informal)
- Minimal (short, data-focused)
- These options affect Aura AI’s tone, sentence length, and intent confidence thresholds.
2. AI Fallback Strategy
- Configure how Aura handles uncertainty or failed attempts:
- Escalate to Human Agent after X failed responses.
- Offer Help Bubbles after fallback events.
- Tag conversations with system labels for future analysis.
3. AI Escalation Thresholds
- Define thresholds to trigger fallback or human escalation:
- Max unanswered questions (e.g., 3 per conversation)
- Flagged intents (e.g., detected frustration or urgent keywords)
Trigger Logic Enhancements
Advanced users can fine-tune their conditions and logic across flows and triggers:
- Use AND/OR Logic to combine conditions like:
- Visitor page URL contains
pricing
AND time spent > 30s
- Chat started OR intent = “book a demo”
- Nested Logic Support for more nuanced targeting:
- Example: If (Location = “France” AND Returning Visitor = true) OR (UTM source = “email-campaign”)
- Custom Visitor Properties can be created and injected via JavaScript to be used in conditions.
Session Configuration & Privacy
- Session Timeout Settings: Control how long visitor sessions are retained before reset.
- Cookie Behavior: Toggle Emplorium’s use of cookies for tracking anonymous sessions.
- Tracking Consent Management: Show consent banners if GDPR or CCPA applies.
AI Memory & Persistence (Coming Soon)
While Emplorium does not yet support persistent memory across sessions, upcoming features will allow:
- AI to remember prior conversations (opt-in).
- Cross-session profile enrichment.
- Memory reset rules per conversation/session.
Custom Variables & Dynamic Data Injection
Administrators can pass custom variables into the chat widget or session context:
- JavaScript snippet to inject:
window.emploriumSettings = {
visitor_name: "John Doe",
user_id: "12345",
subscription_plan: "Premium"
};
- These variables are accessible by flows, triggers, and the AI for personalized experiences.
Environment Modes
- Live Mode: For production/live visitor interactions.
- Test Mode: Used inside the Virtual Visitor tab for simulating visitor conversations and flows without real traffic.
- Debug Mode: Show live logs of trigger matches, intent detection, and fallback reason codes.
Best Practices
- Use Test Mode before publishing any new flows or triggers.
- Regularly review fallback tags to improve AI and rule quality.
- Enable consent banners if operating in regulated regions (e.g., EU).
- Inject only safe and minimal variables to avoid data leakage.
Responses are generated using AI and may contain mistakes.