Creating an agent for an Agentic Messaging interface

This video shows you how to create an agent that you can use in an Agentic Messaging channel interface to provide immediate assistance for customers.

U+, a retail bank, wants to create an agent named Self-Service Agent to assist customers with updating their residential or mailing addresses. The agent guides them through the process of updating their address information without needing to connect them to the customer service representatives.To implement the business scenario, go to AI Designer, and then select agent. Click the Create new Agent button to create the Pega Intelligent Virtual Agent™.

In the Agent name field, enter a descriptive name for the agent (for example, Self-Service Agent), and then in the Description field, describe the purpose of the agent. For example, The purpose of this agent is to assist customers in managing their address information. It can add new addresses, update existing ones, or modify addresses according to customer requests.

In the Scope section, determine whether the agent applies to your current application or a Case Type. Because the company wants to help customers update their address, select Case Type.

In the Applies To list, select the Address change Case Type to associate it with the Intelligent Virtual Agent, and then click Submit.

The agent definition opens; you must provide the details for the following fields by referring to the Prompt document:

• Agent instructions
• Guardrails
• Response style and tone

To provide more context for the agent, in the Additional context section, provide Data Pages that the agent always accesses while generating a response by clicking Add Data page. In this case, add the Customer context Data Page because the agent always needs access to customer context data.

In the Add Data Page dialog box, in the Name field, provide a descriptive name for the data source. For example, Customer context data. In the Data page field, enter the name of the Data Page, D_pxDMSContextData, and then click Submit.

Optional: To retrieve only specific values from the Data Page, in the Apply Data Transform field, enter the Data Transform that maps the required fields from the Data Page.

Next, preconfigure questions to enhance user interactions with the agent. In the Starter question section, enter an initial question for the agent to ask on launch. For example, I am an AI assistant specialized in helping with address change. I can help you update your address. Would you like to proceed with changing your address? I will guide you through the process step by step.

In the Quick-Select Question section, you can enter questions that users can select while interacting with the agent. In this example, enter Getting started.

To enable the agent to start and process Cases at run time, on the Case Types tab, select the Case Type that you want to associate with the agent from the list. In this example, select Address change. Agents can begin and process Cases that include scalar values only. In the background, the system creates the Tool Rule that references the selected Case Type.

Click Save to save the changes.

Optional: To enable the agent to source information from a Pega GenAI Knowledge Buddy™, on the Knowledge tab, in the Buddy section, select customer_service_buddy.

You can select one of the Buddies provided by default or your own Buddy. For more information, see Creating a Knowledge Buddy.

Now provide Data Pages as additional knowledge sources for the agent; for example, Asset list, Transaction list, Self-service case types, and Get assignment key.

Click Save to save the changes.

On the Settings tab, in the Model section, select the AI model that you want the agent to use. In this example, use the Claude-35-Sonnet model.

Optional: To integrate the agent with external resources, in the External Access section, select the Enable external access checkbox.

Go back to the Agent's landing page and observe that the Self-Service Agent is in the list of agents.

Save the configurations.

You have reached the end of this video. You learned:

• How to create an agent for an Agentic Messaging interface.