Identify Intent
The Identify Intent stage is designed to automatically understand the intent of input data (like user messages or form entries) using AI, and classify it into pre-defined intents such as Complaint, Request, or Feedback.
This stage helps workflows branch based on user intention, ensuring accurate routing and response automation.
Identify Intent
Select variable or repositories:
The dropdown displays the "Variable" value in the list.
Select variable or fields:
The dropdown displays all variable names from previous stages.
If no variables are available from previous stages, it shows 'No variables found’.
When select a variable in the first dropdown, the second dropdown dynamically displays all variables from previous stages.
Create a prompt:
“Create a prompt” field mandatory, the validation message displays as 'Prompt is required' when the prompt field is left empty.
When you copy a variable from the dropdown, it gets inserted into the prompt in the following format: ${Variable:variableName}
After saving the form, the prompt remains exactly as entered — no formatting loss or data corruption occurs when you reopen the stage.
What Is Intent-1, Intent-2, Intent-3?
These are predefined categories you create to classify input text. You can name them based on what types of intent you expect from the input.
Each intent is a label that defines how the AI should classify the input.
Intent Label
Meaning
Example Classification Case
Intent-1
Complaint
" I am not able to log in to the portal."
Intent-2
Request
"Please add my team member to the CRM system."
Intent-3
Feedback
"I really like the new dashboard UI you launched."
These labels can be customized based on use case.
Example Scenario:
Variable value: I am not able to log in to the portal
Prompt:
Analyze the following content and identify the user's intent.
Message: ${Variable:Message}
Classify the intent into one of the following categories:
Intent-1: Complaint - A problem or issue the user is facing
Intent-2: Request - A request to take an action or provide access
Intent-3: Feedback - A positive or constructive opinion
Output:
Option-1 (Complaint)
Identify Intent Under For Each Stage
Select variable or repositories:
The dropdown list displays the Variable and Repository names selected in the 'For Each' stage.
Select variable or fields:
When a Variable is selected in the first dropdown, the second dropdown dynamically displays the list of available variables from previous stages.
When a Repository is selected in the first dropdown, the second dropdown displays all fields of the selected repository.
If no variables are available from previous stages, the second dropdown displays the message: "No variables found."
Create a prompt:
“Create a prompt” field is mandatory, the validation message displays as 'Prompt is required' when the prompt field is left empty.
When variable values are copied, the prompt inserts them in the following formats:
For Variables: Format: ${Variable:variableName} Example: ${Variable:Message}
For Repository Fields: Format: ${RepositoryName:FieldName} Example: ${Candidate_Details:Experience} -The system uses the internal names of the repository and the field when generating the token.
After saving the form, when the stage is reopened, the entered prompt appears exactly as saved without any formatting changes or data loss.
What is Intent-1, Intent-2, etc.?
These are predefined categories you create to identify input text. You can name them based on what types of intent you expect from the input.
Each intent is a label that defines how the AI should classify the input.
Intent Label
Meaning
Example
Intent-1
Complaint
" I am not able to log in to the portal."
Intent-2
Request
"Please add my team member to the CRM system."
Intent-3
Feedback
"I really like the new dashboard UI you launched."
These labels can be customized based on use case.
Expected Behaviour: Identify Intent Output Storage
When the Identify Intent stage is used inside a For Each loop, it evaluates the decision for each record in the repository.
If the AI responds with “Customer Complaint”, the system automatically selects and stores Intent Option-1 as the result and routes the workflow accordingly.
Example scenario:
If select For Each Repository: IT Ticket
Prompt to AI:
Message: ${repository:RequestTitle}
Analyze the above message and identify it into one of the following Intent:
Intent-1: Complaint - A problem or issue the user is facing
Intent-2: Request - A request to take an action or provide access
Intent-3: Feedback - A positive or constructive opinion
Purpose:
We are looping through each ticket in the IT Ticket repository. For each item, the AI evaluates the Request Title and identify the intent of the request into one of the predefined options.
Output Examples:
Intent Label
Meaning
Example Classification Case
Intent-1
Complaint
" I am not able to log in to the portal."
Intent-2
Request
"Please add my team member to the CRM system."
Intent-3
Feedback
"I really like the new dashboard UI you launched."
Expected Behaviour:
For each ticket, AI will analyze the request title and return the matching Intent (Intent-1, Intent-2, or Intent-3).
This option will be stored in the defined output variable (e.g., RequestType), and can be used for routing or filtering in later stages like update record stage (update the priority)
Identify Intent Under Condition Stage
Select variable or repositories:
The dropdown shows variable values.
Select variable or fields:
It shows all variables from previous stages in the dropdown.
If no variables are available from previous stages, it displays 'No variables found’.
When a variable value selected in the first dropdown, the second dropdown dynamically shows the list of available variables from previous stages.
Create a prompt:
“Create a prompt” field is mandatory, the validation message displays as 'Prompt is required' when the prompt field is left empty.
When you copy a variable from the dropdown, it gets inserted into the prompt in the following format:
${Variable:variableName}
After saving the form, the prompt remains exactly as entered — no formatting loss or data corruption occurs when you reopen the stage.
Expected Behaviour: Identify Intent Output Storage
Scenario:
When the Intent Identify stage is used under a Condition stage, and the condition checks a specific variable value, the AI will evaluate the prompt only if the condition is true.
Condition Logic: If ${Variable:Request} is equal to Intent Option-2 (In the Intent Option-2 store service request value)
Only when this condition is satisfied, the workflow proceeds to the next stage — Identify Intent.
Identify Intent: Prompt: Request Title: ${Variable:RequestTitle }
Based on the above content, Identify intent of request.
Intent-1: Complaint - A problem or issue the user is facing
Intent-2: Request - A request to take an action or provide access
Intent-3: Feedback - A positive or constructive opinion
Output Storage Logic:
Intent-1 → AI returns "Complaint"
Intent-2 → if AI returns "Request"
Intent-3 → AI returns "Feedback"
Example:
"I really like the new dashboard UI you launched."
Expected Output (from Identify Intent Stage)
Intent- 3 (Feedback)
Last updated