Prompts & Behavior Controls
Prompts and behavior controls let you fine-tune how your agent responds, handles specific situations, and maintains conversation flow.System Prompts
The system prompt is the foundation instruction that shapes all agent behavior.Basic Structure
Effective Prompts
| Good | Bad |
|---|---|
| ”You are a medical receptionist who helps with scheduling." | "Be a helpful AI." |
| "If asked about pricing, refer to our fee schedule." | "Answer all questions." |
| "Confirm the caller’s date of birth for verification." | "Verify identity.” |
Prompt Components
- Identity — Who is the agent?
- Context — What organization, what setting?
- Capabilities — What can the agent do?
- Constraints — What should the agent avoid?
- Style — How should responses sound?
Behavior Modes
Conversation Mode
Control overall conversation style:Task Mode
For transaction-focused interactions:Response Templates
Greeting Templates
Situation Templates
Dynamic Templates
Use variables in templates:| Variable | Description |
|---|---|
{{agent_name}} | Agent’s name |
{{caller_name}} | Caller’s name (if known) |
{{business_name}} | Business name |
{{date}} | Relevant date |
{{time}} | Relevant time |
{{custom.field}} | Custom context data |
Intent Handling
Configure responses for specific intents:Fallback Responses
Handle unexpected inputs gracefully:Turn Management
Control conversation flow:Interruption Handling
| Mode | Behavior |
|---|---|
pause_and_listen | Stop speaking, listen |
complete_thought | Finish current sentence |
acknowledge_and_continue | ”I’ll finish this thought” |
Confirmation Patterns
Ensure accuracy for critical information:Error Recovery
Handle mistakes gracefully:Topic Steering
Keep conversations on track:Context Awareness
Adapt responses based on context:Testing Behaviors
Scenario Testing
Test specific scenarios in the simulator:- Schedule appointment (happy path)
- Cancel appointment (verification needed)
- Unknown question (fallback)
- Frustrated caller (empathy)
- Off-topic question (steering)
Behavioral Metrics
| Metric | Target |
|---|---|
| First-response relevance | > 90% |
| Fallback rate | < 10% |
| Confirmation accuracy | > 95% |
| Transfer rate | < 20% |
Best Practices
Be specific in prompts
Be specific in prompts
Vague prompts lead to inconsistent behavior. Be explicit about expectations.
Test edge cases
Test edge cases
Don’t just test the happy path. Try unexpected inputs.
Use natural language
Use natural language
Templates should sound human, not robotic.
Iterate based on logs
Iterate based on logs
Review real conversations and refine behaviors.
Keep prompts focused
Keep prompts focused
Don’t overload with too many rules. Prioritize what matters.
Next Steps
- Guardrails — Set hard boundaries
- Agent Personas — Define personality
- Pathways — Design conversation flows