AI Agents: Comprehensive User Guide
Our AI agents are advanced, speech-driven assistants that can engage in natural conversations, handle calls, and perform designated tasks—all without relying on separate speech-to-text (STT) or text-to-speech (TTS) systems. This guide explains what agents are, how to create, configure, and enhance them, and offers best practices for maintaining and improving their performance.
Table of Contents
- Overview
- What Is an Agent?
- Getting Started: Creating Your First Agent
- Configuring Your Agent
- Enhancing Your Agent
- Best Practices for Agent Management
- Understanding Our Native Speech-to-Speech Model
- Common Use Cases
- Troubleshooting and Adjustments
- Getting Support
Overview
This guide provides a complete walkthrough for setting up and using our AI agents. These agents leverage a unique speech-to-speech model, which means they natively process spoken language—delivering a seamless, human-like conversation experience without intermediate text conversion.
What Is an Agent?
An agent is an AI-powered assistant that can:
- Engage in Conversations: Handle spoken dialogue naturally.
- Perform Tasks: Execute actions such as checking order status or scheduling appointments.
- Represent Your Brand: Customize responses, tone, and personality to match your business identity.
Our agents use a native speech-to-speech model, which allows them to directly understand and generate speech. This creates fluid and authentic interactions.
Getting Started: Creating Your First Agent
Basic Setup
- Name: Choose a descriptive name that reflects the agent’s role (e.g., "Customer Support Sarah").
- Description: Provide a concise summary of the agent's purpose (e.g., "Assists with account inquiries and basic troubleshooting").
Choosing Language and Voice
- Language Selection: Decide the language in which your agent will communicate.
- Voice Selection: Choose a voice that aligns with your brand identity. The voice affects how the agent sounds during live interactions.
Configuring Your Agent
Greeting and Introduction
Craft a welcoming message that introduces your agent to the user. For example:
"Hello, I'm Sarah from Customer Support. How may I assist you today?"
This greeting sets the tone and helps users understand who they are interacting with.
Defining the Base Prompt
The base prompt shapes your agent's personality and core knowledge. Think of it as its training manual. For example:
"You are a friendly customer service representative specializing in account inquiries and technical support."
This prompt guides the agent's behavior, ensuring consistency in responses.
Providing Context and Custom Instructions
Provide additional background and guidelines:
-
Context Example:
"You work for a software company that provides project management tools. Your main responsibilities include handling subscription issues and troubleshooting basic problems."
-
Custom Instructions:
- Always ask for an order number when handling delivery inquiries.
- Escalate complex technical issues to a human supervisor.
Designing Follow-Up Questions
Prepare follow-up questions to gather detailed information. Examples include:
- "What specific features are you interested in?"
- "Can you describe the issue in more detail?"
These questions help refine the conversation and ensure the agent collects all necessary details to assist effectively.
Enhancing Your Agent
Integrating Knowledge Bases
Enhance your agent's responses by linking:
- Documents and FAQs: Provide direct access to product manuals, support articles, and FAQs.
- Websites: Integrate relevant web pages to keep responses current and accurate.
Assigning Tools
Tools extend your agent's capabilities beyond conversation, enabling real-world actions:
User Tools
Custom tools you create to integrate with your business systems:
- API Integrations: Connect to your CRM, order systems, or databases
- External Services: Integrate with payment processors, calendars, or notification systems
- Business Logic: Implement custom workflows specific to your operations
How to Assign:
- Navigate to Tools → User Tools
- Find your tool and click "Assign"
- Select which agents should have access
- Tools become immediately available during calls
System Tools
Built-in platform tools for common call management:
- Hang Up: End calls when conversation is complete
- Play DTMF Sounds: Navigate phone trees or enter codes
How to Assign:
- Navigate to Tools → System Tools
- Click "Assign" on the desired tool
- Select agents that need this functionality
- System tools are available during live calls
Tool Usage During Calls
When an agent handles a call, it automatically:
- Recognizes when a tool might be useful based on conversation context
- Extracts required information from the customer's speech
- Executes the appropriate tool (API call or system action)
- Responds with results in natural language
For detailed tool creation and management, see our Tools Guide.
Defining Specific Actions
Program your agent to perform targeted tasks, such as:
- Order Tracking: Automatically retrieve order status.
- Appointment Scheduling: Book and manage appointments.
- Customer Updates: Update customer information directly in your CRM.
Best Practices for Agent Management
- Start Simple: Begin with core configurations and basic functionalities. Gradually add features as needed.
- Test Extensively: Simulate various scenarios to ensure the agent handles different cases correctly.
- Iterate Regularly: Update your knowledge base and prompts based on user feedback and common interaction patterns.
- Maintain Consistency: Keep a uniform tone and style across all communication to build trust and clarity.
Understanding Our Native Speech-to-Speech Model
Our platform’s unique feature is the native speech-to-speech model:
- Direct Interaction: The agent understands spoken language directly, eliminating the need for intermediate STT or TTS processing.
- Natural Communication: This results in faster, smoother conversations that feel more natural and human-like.
- Simplified Workflow: With native processing, integration becomes simpler and response times improve, enhancing the overall user experience.
Common Use Cases
- Customer Service: Answer inquiries, provide product details, and manage support tickets.
- Appointment Scheduling: Efficiently book, reschedule, and cancel appointments.
- Sales Support: Qualify leads, provide pricing details, and answer product-related questions.
- Technical Troubleshooting: Guide users through common issues and escalate when needed.
Troubleshooting and Adjustments
Common Issues and Fixes
- Too Formal Responses: Adjust the base prompt to adopt a more conversational tone.
- Vague or Incomplete Answers: Increase the specificity of the context and follow-up questions.
- Missing Information: Enhance your knowledge base and ensure all relevant data sources are integrated.
Tips for Continuous Improvement
- Monitor Interactions: Use analytics to track common queries and adjust prompts accordingly.
- Iterate Based on Feedback: Regularly update your agent’s training materials based on real-world usage.
- Keep Knowledge Current: Ensure any linked documents or FAQs are frequently reviewed and updated.
Getting Support
If you need further assistance:
- Review the Documentation: Revisit this guide for detailed configuration and troubleshooting tips.
- Test in a Controlled Environment: Experiment in a sandbox environment before deploying changes live.
- Contact Support: Reach out to our support team for personalized help and advanced troubleshooting.
By following this guide, you can build, configure, and optimize your AI agents for natural, effective, and engaging interactions—all powered by our innovative speech-to-speech model. Enjoy creating smarter, more intuitive AI assistants tailored to your business needs!