Generative AI in Pega

Pega GenAI™ Autopilot describes all generative AI features available at design time. One benefit of Autopilot is that it accelerates the low-code application development process. For example, application developers who build an application for a home loan can apply the integrated Autopilot features, which provide suggestions for components of the application, including Case Types, fields in the Data Model, and Personas. Developers can review and modify any AI-powered recommendations before generating the suggestions.

The following video shows AI-assisted development in Pega Infinity:

Video transcript

Application development can often be time-consuming. With the help of Pega GenAI Autopilot features, you are presented with suggestions throughout development to accelerate the process.

For example, you are building a loan management application with the assistance of Autopilot. When you create a Case Type, Autopilot features suggest various Case Type names based on the application name. You can then use Autopilot to generate a suggested workflow to kickstart development. You should then review and modify any suggested Stages, Processes, and Steps. Autopilot can provide suggestions for the Case Type Data Model, including suggesting choices for a picklist field. Next, you can use Autopilot to provide suggestions for the users, or Personas, who interact with the Case Type. Use Autopilot to generate sample records for a data object to make testing easier. When you run the Case Type, you can use Autopilot to fill out the form with sample data. You can then use natural language prompts to generate a recommended Insight, which allows you to analyze your application data.

Pega GenAI Blueprint™ is a powerful tool that helps you rapidly and collaboratively design the foundations of an application that fulfills your business needs.

With Blueprint, you can perform the following tasks in record time:

• Collaborate on the application design without the need for a Pega instance or login because Blueprint operates independently as a SaaS tool separate from existing Pega Studios. Project stakeholders with a Pega login can still participate.
• Describe what you are trying to build in business terms. Blueprint uses that information to propose Case Types, Case Life Cycles, Data Models, data objects, and Personas, which enable quick editing of the designs to fit business requirements.
• Receive powerful and detailed best-practice application templates instantly.
• Export the resulting Blueprint and import the file directly into Pega Platform, jumpstarting a demo or a project build-out by producing a functional application in just minutes.

The design of Blueprint does not replace App Studio or other existing Pega tools. Instead, it augments them by offering a fresh perspective on collaboration to effectively transform ideas for workflow transformation into a functional application in record time.

To create a Blueprint, go to https://pega.com/blueprint. 

The power of generative AI is included directly in Pega Platform, which enables you to customize Pega GenAI Connect Rules to meet the needs of your application. For example, the Pega GenAI Connect Rule can summarize the details of a complex Case for a customer service representative who has been brought in to help resolve the complaint or provide a list of required documents that the customer must submit to resolve the complaint. 

With Pega Infinity '24, you can now configure a Connect Generative AI automation Step in App Studio. Creating a Connect Rule begins with defining a Prompt. The Prompt defines the request for information sent to the Pega GenAI engine. It can be a custom Prompt specific to the Step or an existing Prompt created in Dev Studio. 

Pega GenAI responds to the prompt with one of the three following structures: 

• Structured - Single: The response consists of a single page with short paragraph or a field with a corresponding value. The system can  automatically map the response directly to a field associated with the Data Model.
• Structured - List: The response consists of a bulleted list of fields and corresponding values. The system can then automatically map the values associated with the fields in the response list directly to the fields associated with an embedded data list in the Data Model.
• Unstructured: The response is long or complex, such as JSON code. In this case, it is a best practice to map the response to a field with a Field Type of Text (paragraph).

Finally, the system requires you to identify a target field in which to write the response data. You must add the target field to the UI of the application so that the response information can be displayed to the application's user.

The following figure displays both the Request and Response sections for a Connect Generative AI Rule in App Studio:

Pega GenAI Connect Rules include additional functionality, such as the power to mask sensitive data during the request/response cycle and the ability to change the language of responses based on the locale setting of the current user.

Check your knowledge with the following interaction: