Reasoning Builder

​

Overview

The Reasoning Builder is a powerful visual tool that allows you to create sophisticated conversation flows for your AI agents. By combining objectives and conditions, you can guide your agent’s responses and decision-making process based on user inputs and context.

​

Key Concepts

​

Objectives

Objectives are specific goals or instructions that tell your agent what to do at each stage of the conversation. An objective includes:

• Label: The name of the node in the chat graph
• Model: The AI model to use for this objective
• Instructions: The specific prompt or instructions for the AI to follow
• Suggested Replies: Optional predefined responses that can be shown to users as buttons

​

Conditions

Conditions are decision points that determine the flow of conversation. They help your agent choose the next appropriate objective based on:

• User input
• Conversation context
• Contact information
• Previous interactions

​

Building Your First Flow

Let’s create a simple customer service flow:

1. Start with a Greeting
   • Create an objective node labeled “Hello”
   • Set instructions like “Greet the user and ask how you can help”
   • Choose your preferred AI model
2. Add Decision Points
   • Create conditions to branch the conversation
   • Example conditions:
     • “Payment Methods” - triggers when users ask about payments
     • “Account Issues” - handles account-related queries
     • “MEPrepaid” - manages prepaid utility questions
3. Connect the Nodes
   • Use edges to connect objectives and conditions
   • Label edges to help understand the flow
   • Create multiple paths for different scenarios

​

Best Practices

​

Writing Effective Instructions

• Be specific and clear
• Include context about available actions
• Define the scope of responses
• Consider edge cases

​

Organizing Your Flow

• Keep it modular
• Group related objectives
• Use clear naming conventions
• Test different paths

​

Using Conditions Effectively

• Make conditions mutually exclusive when possible
• Use specific triggers
• Consider fallback paths
• Test edge cases

​

Advanced Features

​

Context Awareness

Your agent can maintain context throughout the conversation by:

• Remembering previous interactions
• Accessing contact information
• Using conversation history

​

Dynamic Responses

Configure your agent to:

• Suggest relevant replies
• Adapt to user preferences
• Handle multiple topics in one flow
• Escalate to human agents when needed

​

Testing Your Flow

Before deploying:

1. Use the Test button to simulate conversations
2. Try different user inputs
3. Verify all paths work as expected
4. Check condition triggers
5. Test fallback scenarios

​

Next Steps

• Learn about Models available for your agents
• Explore Channel Integration to deploy your agent
• Discover how to add Tools to enhance capabilities