Creating a custom question in the Generative replies block in conversation flows for advanced AI agents

What's my plan?

Add-on

AI agents - Advanced

This feature is in beta state - if you experience any issues or would like to share feedback, please contact your CSM.

The Generative replies block, by default, utilizes the last customer message as a prompt for searching the knowledge source and delivering an answer whenever a generative reply is triggered within a conversation.

However, this default behavior has limitations. Frequently, questions that can be answered with a knowledge source article are found in the middle of a conversation or within a specific Intent, and the last customer message may lack the necessary original context. This results in the AI agent being unable to generate an accurate response.

To enhance the capabilities of the Generative replies block and enable its use within pre-defined Intents, we introduced the Custom question field within the Details section of the Generative replies block in the dialogue builder. This feature allows for the specification of a custom question or prompt, ensuring more accurate and contextually relevant responses, even when the original context is not readily available in the last customer message.

The custom question feature empowers you to incorporate context from previous messages, including intents, specific user inputs, session parameters, and more. This enables the AI agent to provide more precise and contextually relevant generative replies, even when the original context is not readily available in the last customer message.

Custom question examples

In this example, we illustrate how to effectively utilize the custom question feature in the context when a specific intent is triggered:

1. Triggering the Intent Reply: Sizing Enquiries

The Intent Reply "Sizing Enquiries" is triggered when customers inquire about their shoe size.

2. Dialogue Design

Within the dialogue, the AI agent engages with customers in the following manner:

• The AI agent requests the customer's foot length in centimeters.
• It inquires about the preferred location for receiving shoe size information, which can be UK, US, or EU.

3. Session Parameter Collection

Customer inputs, such as foot length in centimeters and chosen location, are collected and stored as session parameters: {{foot_lenght_cm}} and {{location}}.

4. Configuring the Custom Question in the Generative replies Block

To harness the potential of the custom question feature, the dialogue builder sets up the custom question within the Generative replies block. In this instance, the question posed is: "Which shoe size would you recommend based on my {{foot_lenght_cm}} and my {{location}}?"

5. Testing the Custom Question

During testing, the generative reply utilizes the custom question as the prompt instead of relying solely on the last customer message - "US" as shown above. The AI agent uses the custom question to search the knowledge source and deliver an accurate answer to the customer's sizing inquiry based on the session parameters shared.

This example demonstrates how the custom question feature empowers the AI agent to provide context-aware responses, enhancing the user experience in pre-defined Intents.