Creating intents from scratch allows you to define custom intents tailored to your specific use cases. This approach gives you full control over intent names, descriptions, priority, languages, and user expressions, ensuring that your Conversational AI model can accurately capture and respond to user messages. Properly defining high-priority intents and including multiple example expressions improves detection accuracy and overall model performance. For further details, see What are Intents?

Creating Intents

1. Initiate New Intent

   1. Click the New Tab icon. Under the Sprinklr Service tab, select Conversational AI within Persona Apps.

   2. On the Conversational AI window, select the application for which you want to create an Intent.

   3. Click ML Models under AI Tools.

      ​

      ​

   4. Click + Add ML Model in the top right corner.

      ​

      ​

2. Create New Model: On the Create New Model window, perform the following steps:

   1. Name: Enter a name for the intent model.

   2. Model Type: Choose the relevant model type.

   3. AI Settings (Optional): Adjust settings as needed:

      1. Exact Match Models: Enable this option if you want intents to be detected only on exact-match phrases. Disable this option if you want the AI to detect intents based on an understanding of the training data.

   4. Select the Pre-Processing Pipelines and choose if it has to be shared with other applications.

      ​

      ​

   5. Click Create.

3. Add Intents:

   1. Click the Caret icon in the top right corner and select + Create Manually.

      ​

      ​

   2. In the Intent Details window, enter:

      1. Intent Name

      2. Description

      3. Priority: Higher-priority intents are selected over others when multiple intents are detected simultaneously. A lower priority number indicates higher priority (e.g., 1 is the highest).

      4. Confidence Threshold: This threshold, which ranges from 0 to 1, helps determine how confident the model should be in its predictions before acting on them.

      5. Hint: Hint will be used to recognize the Entity from the user's speech in VoiceBot.

         ​

         ​

      6. On the left pane, choose the desired Languages for this intent.

         ​

         ​

4. Add User Expressions

   1. Click Next and, for each selected language, add user expressions under Expressions.

   2. Include as many example expressions as possible to accurately convey the messages this intent should capture.

      ​

      ​

   3. Click Save at the bottom.