Workflow AI - Intent Detection is a feature within the Workflow AI tool that helps identify the purpose or goal behind a contact's message or query. It analyses the content of messages to determine what the person intends to do or ask, such as booking an appointment, asking for support, or enquiring about products. This enables automation workflows to respond more accurately and trigger appropriate actions based on the user's intent.

📌Note: Intent Detection is a premium action. Using this action will incur additional charges per execution.

How It Works

Begin by creating a new workflow or editing an existing one within the CRM. Then, add a suitable trigger that captures the contact message or interaction you want to analyse.

Insert Intent Detection Action

Navigate to the actions list and select the "Workflow AI - Intent Detection" action. This action will evaluate the message's content.

Action Name

Once selected, you can update the action name if required. This lets you assign a clear, descriptive name that gives context to the action's purpose.

Input Text

Specify the message source or input field that the intent detection will analyse.

If your workflow trigger is "Customer Replied" or you have preceding actions that capture a response, you can use the Message Body custom value.

Save the action to set conditions, trigger subsequent actions, or route the conversation accordingly. Positive intents typically include replies indicating interest or approval, while negative intents include responses showing disinterest or rejection.

Use the testing features to verify that the system correctly identifies intents and tweak the settings for better accuracy.

Practical Examples & Use Cases

  1. Lead Qualification: Automatically identify if a message indicates genuine interest or questions, enabling faster follow-up.
  2. Support Routing: Detect support-related queries to route contacts directly to a support team or chatbot, reducing wait times.
  3. Product Enquiries: Detect questions about products or services, enabling the system to respond with relevant information or offers.
  4. Customer Feedback: Identify positive or negative feedback to trigger appropriate outreach or escalation.

Validations & Constraints

  • The feature requires a clear input message to analyse.
  • Correct classification depends on the quality of training data and prompt setup.
  • It recognises predefined or custom intents; you should define or train these for optimal results by adding relevant filters to the trigger, such as the "Contains Phrase" filter for the Customer Replied Trigger.
  • Misclassification can occur if the message is ambiguous or outside the scope of the training data.
  • Ensure that the intent detection is tested thoroughly within your workflow to avoid misrouting.