02. Prompting

Steps for Creating a Receptionist AI Assistant

1. Create a New Assistant

1. Navigate to the top-right corner of the dashboard and select Create New Assistant.
2. Choose from the following options:
   • Import: Bring in an assistant from another account using its ID.
   • Blank Model: Start from scratch.
   • Generate Template: Use a pre-made prompt template (recommended for beginners).
3. Enter a description to guide the assistant's behavior. Example:
   • Name: Emma
   • Role: Emma is a receptionist for Apex Concrete, responsible for answering questions, asking qualifying questions, and booking estimates.

2. Set Up the Assistant’s Identity and Style

• Identity: Describe the assistant’s role, demeanor, and expertise. Example:
  arduino

  Copy code

  You are Emma, a receptionist for Apex Concrete. Your role includes answering customer inquiries, qualifying leads, and scheduling appointments. You maintain a professional, approachable, and efficient tone.
• Response Guidelines:
  • Use concise and clear responses.
  • Maintain a conversational, proactive, and helpful tone.
  • Lead the conversation back to the topic if the user veers off-course.
  • Ensure human-like phrasing with diverse language.

3. Build the Task-Based Framework

Structure the conversation in sequential steps:

1. Greet the user warmly:
   • Example: "Hi, I’m Emma, the receptionist for Apex Concrete. How can I assist you today?"
2. Handle user queries:
   • Provide specific information based on the query.
   • Use conditionals:
     • If the user asks about services, explain them.
     • If the user inquires about estimates, ask qualifying questions.
3. Ask qualifying questions:
   • Example questions:
     • What type of project is this? (e.g., foundation, driveway)
     • What is the project size or budget?
     • When do you need the project completed?
4. Schedule appointments:
   • Confirm the date and time with the user.
   • Request contact details (name, email, phone).
5. End the conversation:
   • Thank the user for their time and confirm the next steps.

4. Connect Knowledge Base

To enhance responses:

1. Go to the right-hand tab labeled Knowledge Base.
2. Link the appropriate knowledge base (e.g., "Concrete Example").
3. Ensure the knowledge base contains FAQs, services, and other relevant details.

5. Test the Assistant

1. Use the Chat Lab or Voice Lab to test the assistant.
2. Example interactions:
   • User: “Hi, can you tell me about your services?”
     • Assistant: “We offer concrete repair, resurfacing, and stamping. Are you looking for a driveway, patio, or another project?”
   • User: “I need a foundation repair.”
     • Assistant: “Sure, can you tell me the size of the project and your timeline?”
   • User: “I’d like to schedule an estimate.”
     • Assistant: “Great! Could I get your name, email, and phone number to schedule the appointment?”

6. Refine the Prompt

If the assistant fails to meet expectations:

• Edit the identity or task framework.
• Add conditionals to improve response specificity.
• Test again until satisfied.

Best Practices

1. Use Clear Vocabulary:
   • Refer to users as “User” and the assistant as “Agent” in prompts.
2. Incorporate Conditionals:
   • Define steps for specific scenarios (e.g., booking estimates, asking for contact details).
3. Keep Iterating:
   • Refine prompts based on test results.
4. Link to Real Calendars:
   • In advanced setups, connect calendars to enable real-time appointment scheduling.

Troubleshooting Tips

• Incomplete or confusing responses: Check if the knowledge base is linked and properly populated.
• Generic answers: Add more specific examples or conditionals in the prompt.
• Appointment issues: Ensure the calendar integration is active (covered in future tutorials).

Conclusion

With this structure, your AI assistant will be capable of efficiently handling customer inquiries, gathering relevant details, and scheduling appointments. For further customizations, explore adding calendar integrations and advanced conditionals to fine-tune its behavior.