The Dialogue Tree tool in Agent Copilot lets you design structured, guided conversations. You use this tool to define how an agent asks questions, captures inputs, and passes information to subsequent steps in a workflow.

A Dialogue Tree helps:

• Collect consistent information from users

• Reduce free-form conversations

• Enable reliable downstream automation

This document explains each input field available when configuring a dialogue tree.

Dialogue Tree Section

The Dialogue Tree section defines the high-level identity and purpose of the dialogue tree.

• Name: Enter a short, descriptive name for the dialogue tree. This field should not contain any spaces.

  Example: Customer_Issue
  

• Description: Write a brief explanation of what the dialogue tree does.

  Example: Collects user issue details and routes the conversation to the appropriate resolution path.

Setup Function Section

The Setup Function section defines how the dialogue tree behaves and what inputs it expects.

Create Dialogue Tree

To define the conversational flow for your use case, select Create Dialogue Tree in the Setup Function section. This action opens the dialogue tree builder, where you configure the triggers and control how the conversation starts, progresses, and ends.

When you select Create Dialogue Tree, Agent Copilot redirects you to a visual builder that represents the dialogue flow as connected nodes. This builder provides a clear, end-to-end view of how the dialogue tree executes at runtime.

Dialogue Tree Builder Overview

The dialogue tree builder consists of three core components: Start, Trigger Filters, and End Dialogue Tree. These components define when the dialogue tree is activated and how it exits.

• Start Node

  The Start node represents the entry point of the dialogue tree. Every dialogue tree begins at this node and proceeds based on the conditions you define. You cannot remove the Start node, and it does not require configuration.
  

• Trigger Filters

  The Trigger Filters node determines whether the dialogue tree should run for a given interaction. Trigger filters evaluate conditions based on existing bot variables, such as detected intent or previously captured values.

  When you select the Trigger Filters node, the Edit conditions for this Dialogue Tree panel opens. In this panel, you define one or more conditions that must be met before the bot proceeds with the dialogue tree.

  

  Each condition consists of:

  • A variable (for example, Detected Intent)

  • An operator (such as Containing)

  • A value to match against (for example, Purchase Product)

  You can add multiple conditions or group conditions together to create more advanced logic. This allows you to ensure that the dialogue tree runs only for relevant scenarios and does not interrupt unrelated conversations.

• End Dialogue Tree Node

  Every dialogue tree must end with an End Dialogue Tree node. This node explicitly marks the completion of the dialogue tree flow and signals that the dialogue tree has finished execution. If the End Dialogue Tree node is not added to the canvas and connected to the flow, you will not be able to save the dialogue tree.

  Ensure that all conversation paths in the canvas ultimately connect to the End Dialogue Tree node before attempting to save. This validation helps maintain well-defined dialogue flows and prevents incomplete or unintended conversation behavior.

Use the plus (+) icon between nodes to add additional nodes to the dialogue tree. This allows you to incrementally build the flow, insert logic, and control how information is collected and processed before the dialogue tree ends.

Click the Save button to save your Dialouge Tree conversational flow and trigerrs.

Dialogue Tree Input Fields

Each dialogue tree input consists of the following fields:

• Name: It is the unique identifier for an input variable. Agent Copilot uses the name internally and relies on it when mapping variables between steps in a workflow. To ensure consistency and maintainability, use lowercase letters, replace spaces with underscores, and choose names that clearly describe the information being captured. For example, customer_intent clearly indicates that the variable stores a description of the user’s issue.

• Description: It provides a short explanation of what information the input captures. It helps authors understand the intent of the input and improves the long-term maintainability of the dialogue tree. When writing a description, focus on what the information represents rather than how the agent collects it, and avoid repeating the input name.

• Type: It defines the data type of the input value. Agent Copilot uses this information to store and process the input correctly and to ensure compatibility with downstream steps in the workflow. Currently, the available option is String, which is used for textual inputs such as descriptions, names, or categories. Use String for free-text or categorical inputs unless another type is explicitly required.

• Required: This indicates whether the input must be provided for the dialogue tree to proceed. This setting controls validation during execution and ensures that critical information is always collected.

  When set to Yes, the input is mandatory; when set to No, the input is optional. Mark an input as required only if downstream steps depend on it, and keep optional inputs truly optional to avoid unnecessary friction in the conversation.

• + Variable Input: Select this button to add additional inputs to the dialogue tree. This option defines all the information the dialogue tree can capture and supports more complex or multi-step conversations. When adding inputs, include one input for each distinct piece of information and avoid overloading a single input with multiple meanings.

Set Input Variables Mapping Section

The Set Input Variables Mapping section connects outputs from earlier steps to inputs of the dialogue tree.

• Current Bot Output refers to a variable that is produced by a previous step in the workflow. Agent Copilot uses this variable to supply existing information to the dialogue tree, which helps maintain continuity in the conversation and prevents the agent from asking the user for the same information again. For example, a value such as {detected_intent} can be passed directly into the dialogue tree if the user’s intent has already been identified earlier in the workflow.

• Next Bot Input is the dialogue tree input that receives the mapped value from the current bot output. This mapping automatically populates the dialogue tree input and enables seamless data flow between steps without additional user prompts. For example, mapping {detected_intent} to {customer_intent} ensures that the dialogue tree starts with the correct context.

• Add Another Mapping allows you to create additional variable mappings within the same dialogue tree. This capability supports multiple inputs and enables the reuse of information throughout the conversation. When adding mappings, ensure that you map only logically equivalent variables and take care to avoid unintentionally overwriting existing inputs.

Click the Save button to save the dialogue tree configuration and makes it available for use. The Cancel button discards any unsaved changes and exits the configuration view.

The Dialogue Tree tool enables you to define structured conversational inputs, control which information is required or optional, and seamlessly map data across workflow steps. By carefully naming inputs, writing clear and purposeful descriptions, and using variable mappings effectively, you can build dialogue trees that are easy to maintain and deliver consistent, reliable agent experiences across conversations.