Skip to main content

Multi-Turn Router Node

The Multi-Turn Router node creates sophisticated multi-turn conversational interfaces with dynamic routing based on user responses. This advanced chatbot node enables complex decision trees, personalized learning paths, and interactive experiences where the conversation flow adapts based on user input. It combines AI-powered conversation with intelligent routing capabilities.

Multi-Turn Router node

Basic Usage​

Use the Multi-Turn Router node to create interactive multi-turn conversations where the dialogue branches based on user responses, creating personalized and adaptive experiences.

Inputs​

The Multi-Turn Router node accepts the following inputs:

  • The Bot Name: The display name for your conversational bot.
  • Icon Image: Optional custom icon/avatar for the bot.
  • Bot Introduction - First Message: The initial greeting message when users start the conversation.
  • Reference Text: Background information or context for the AI to reference during conversations.
  • Append Instruction: Additional instructions to guide the bot's behavior during the conversation.
  • Overwrite Instruction: Custom instructions that completely replace default AI behavior.
  • First User Message: The initial message shown to users to start the conversation.
  • Placeholder: Placeholder text shown in the user input field.
  • Route Inputs: Dynamic text inputs that define different conversation paths based on user interest or responses.

Outputs​

The Multi-Turn Router node provides multiple outputs:

  • Full Chat Log: Complete conversation history including all turns and responses.
  • Route Outputs: Multiple output handles corresponding to each defined route, allowing different conversation paths to trigger different downstream processes.

Configuration​

Process Type​

Select the AI model to use for conversation processing:

  • GPT-4o-mini: Fast and cost-effective option for most conversational tasks.
  • GPT-4o: Advanced reasoning and more sophisticated conversation capabilities.
  • GPT-3.5: Efficient model for simpler conversations.
  • Claude 2.1: Anthropic's AI model with strong conversational abilities.
  • Other supported models: Additional AI models available in your system.

Bot Control Section​

The Bot Name: Enter the name displayed for your conversational bot (e.g., "Physics Tutor Bot", "Career Advisor").

Icon Image: Upload or select an icon/avatar to visually represent your bot in the interface.

Bot Introduction - First Message: Write the greeting message users see when they start interacting.

Knowledge Base Section​

Reference Text: Provide background information, knowledge base content, or context that the bot should use during conversations. This helps the bot provide accurate and relevant responses.

Instructions Section​

Append Instruction: Add supplementary instructions to guide the bot's behavior without replacing default instructions.

Overwrite Instruction: Completely replace default bot behavior with custom instructions. Use this for specialized conversation patterns.

User Interaction Section​

First User Message: Set the initial message or prompt shown to users to begin the conversation.

Placeholder: Define the placeholder text in the user input field (e.g., "Type your response...", "Share your thoughts...").

Routing Configuration​

Route Definition: Create multiple conversation routes based on user interests or responses:

  • Each route has a text input defining the trigger condition
  • Routes correspond to different paths the conversation can take
  • Example routes: "User is interested in Physics", "User is not interested in Physics"
  • + Add New Route: Click to add additional routing paths

Additional Options​

Full Chat Log: Output handle for accessing complete conversation history.

Temperature Slider: Control the creativity and randomness of AI responses (0.0 = deterministic, 1.0 = creative).

Math Input: Enable mathematical expression input if needed.

Handwriting Input: Enable handwriting recognition for user responses.

Example Workflows​

Physics Learning Path Router​

Scenario: Create an interactive bot that assesses student interest in Physics and routes them to appropriate learning content based on their response.

Multi-Turn Router Example

Steps to Create the Flow:

  1. Set up the Multi-Turn Router node:

    i. Configure Bot Control:

    • Process Type: Select GPT-4o-mini
    • The Bot Name: "Physics Tutor"
    • Icon Image: Upload a friendly robot or teacher avatar
    • Bot Introduction - First Message:
    Hello! {{SYSTEM_DISPLAYNAME}}! 👋 I'm here to assist you with a wide range of topics. 
    From...

    ii. Configure Knowledge Base:

    • Reference Text: Add physics curriculum information, learning objectives, or subject descriptions

    iii. Configure Instructions:

    • Append Instruction:
    Ask 3 questions to determine whether a user is interested in Physics.
    • Overwrite Instruction: (Leave empty or add custom conversation flow instructions)

    iv. Configure User Interaction:

    • First User Message: Set initial prompt or question
    • Placeholder: "Type your answer here..."

    v. Configure Routes:

    • Route 1 Text Input:
    User is interested in Physics
    • Route 2 Text Input:
    User is not interested in Physics

    vi. Set Temperature: Adjust slider to 0.5-0.7 for balanced responses

  2. Connect Route Outputs to appropriate paths:

    i. For "User is interested in Physics" route:

    • Connect to Display Text showing:
    ## User is interested in Physics
    • This leads to physics learning materials, lessons, or advanced content

    ii. For "User is not interested in Physics" route:

    • Connect to Display Text showing:
    ## User is not interested in Physics
    • This might lead to alternative subjects or general content

  3. Add supporting content nodes:

    • Add Image nodes for visual representations
    • Add Text nodes for route-specific messages
    • Connect "Multi-Turn Router" text input for follow-up interactions

Preview:

Conversation Flow:

  1. Bot: "Hello! I'm here to assist you. Let me ask you a few questions about Physics..."
  2. User: "I love learning about forces and energy!"
  3. Bot: "That's great! Can you tell me what interests you most about Physics?"
  4. User: "I'm fascinated by quantum mechanics and relativity."
  5. Bot: "Wonderful! It seems you're very interested in Physics."
  6. System: Routes to "User is interested in Physics" path → Shows advanced physics content

Result: An intelligent conversational bot that engages users in multi-turn dialogue, assesses their interests through natural conversation, and dynamically routes them to appropriate learning paths based on their responses.