Skip to main content

​
Overview

Creating your first AI agent in SubVerse is a straightforward process. This guide will walk you through every step, from choosing the agent type to configuring settings and deploying your agent.

​
Step 1: Access the Agents Section

  1. Log in to your SubVerse Dashboard
  2. Navigate to Agents from the left sidebar
  3. Click the + Add Voice Agent button in the top-right corner

​
Step 2: Choose Your Agent Type

SubVerse offers two types of agents to suit different use cases:

​
Standard Voice Agent

Best for:
  • Simple, predictable conversations
  • FAQ handling
  • Structured information delivery
  • Quick deployment needs
Characteristics:
  • Predefined responses and conversation paths
  • Faster setup and configuration
  • Lower complexity
  • Ideal for straightforward customer interactions

​
Conversational Flow

Best for:
  • Complex customer service scenarios
  • Dynamic, context-aware interactions
  • Sales conversations
  • Technical support
Characteristics:
  • Advanced AI with natural language understanding
  • Handles multi-turn conversations
  • Adapts to customer intent in real-time
  • Requires more detailed configuration
If you’re new to AI agents, start with a Standard Voice Agent. You can always create a more advanced Conversational Flow agent once you’re comfortable with the platform.

​
Step 3: Select a Template (Optional)

SubVerse provides pre-built templates to accelerate your setup.
Templates come with pre-configured prompts, suggested voices, and recommended settings. You can customize every aspect after selection.

​
Step 4: Configure Basic Settings

​
Greeting Message

The first message your agent speaks when the call connects. Best Practices:
  • Keep it concise (under 15 seconds)
  • Include your company name
  • State the purpose clearly
  • Be welcoming and professional

​
Greeting Type

Choose how the conversation starts:
  • Prompt: Agent modifies the greeting message to match the agent’s identity
  • Say: Agent says the greeting message exactly as it is

​
Who Speaks First

  • SubVerseAI Agent: Agent initiates the conversation
  • Customer: Customer speaks first
For outbound calls, always have the agent speak first. For inbound calls, consider your use case—customer service typically has the agent greet first.

​
System Prompt

This is the most critical configuration—it defines your agent’s behavior, personality, and capabilities. Components of a Good System Prompt:
  1. Identity: Who is the agent?
  2. Role: What is their purpose?
  3. Personality: How should they communicate?
  4. Capabilities: What can they do?
  5. Limitations: What should they avoid?
  6. Instructions: Specific behavioral guidelines
Use the “Refine using AI” button to get AI assistance in creating or improving your system prompt.

​
Step 5: Save Your Agent

  1. Review all your settings
  2. Click the Save button in the top-right corner
  3. Your agent is now created and ready for configuration

​
Next Steps

After creating your agent, you’ll want to:

Configure Advanced Settings

Fine-tune LLM, voice, speech detection, and call settings

Test Your Agent

Make test calls and monitor results in call history

Set Up Integrations

Connect knowledge bases and functions

Monitor Performance

Track calls and analyze agent performance

​
Quick Tips for Success

​
Start Simple

Begin with basic functionality and add complexity gradually. A simple, working agent is better than a complex, problematic one.

​
Test Frequently

Make test calls after each configuration change to ensure everything works as expected.

​
Iterate Based on Feedback

Listen to call recordings, review transcripts, and continuously improve your agent based on real interactions.

​
Use Templates Wisely

Templates are great starting points, but customize them to match your specific business needs and brand voice.

​
Monitor Performance

Regularly check your agent’s performance metrics and make adjustments to improve success rates.
Before deploying your agent to production, conduct thorough testing with various scenarios and edge cases. Poor agent performance can negatively impact customer experience.

​
Common Mistakes to Avoid

Overly Long Greetings
  • Keep greetings under 15 seconds
  • Get to the point quickly
  • Save detailed information for later in the conversation
Vague System Prompts
  • Be specific about what the agent should do
  • Include clear examples of desired behavior
  • Define boundaries and limitations
Skipping Testing
  • Always test before going live
  • Try different conversation paths
  • Test with various accents and speaking styles
Ignoring Voice Selection
  • Choose a voice that matches your brand
  • Consider your target audience
  • Test different voices with sample customers

​
Getting Help

If you need assistance:
Join our community to share experiences, get tips, and learn from other SubVerse users who have successfully deployed AI agents.