Pega Agents are powerful, amazingly helpful partners for data‑driven operations. By letting you chat with your data in natural language, they quickly uncover bottlenecks, spotlight long‑pending items, balance workloads, and recommend optimizations—so teams can act faster and keep processes running at peak efficiency.

Chat with your Data Agent, powered by Pega GenAI, lets users ask natural-language questions about their application data and get answers in a chat window. It combines Connect SQL queries with the Claude-37-sonnet generative AI model to interpret questions and produce data-backed responses.

How it fits into your app

• The capability is delivered as the pxChatWithYourData Tool, which ships with Pega Platform by default.
• To use it, create an Agent and associate the pxChatWithYourData Tool with that Agent. The Agent then invokes the tool at run time whenever data analysis is needed.
• You can enable multi-Agent collaboration and attach the tool to any Agent that should answer data questions.

Typical questions it can answer

• Provide the list of current trends for the Loan Application
• Provide the list of spikes for Loan Application cases
• List of historical cases which require pending resolution.

Always-available data sources (Data Pages)

• D_pzAllCwydEnabledCaseTypes: Classes enabled on the Data Configurator landing page, including Case Types and Data Objects from the current and inherited applications.
• D_pzDisabledClassesForCWYD: Classes explicitly disabled on the Data Configurator landing page; the Agent will not use these at run time.
• D_pzRecentlyVisited: Items (for example, Cases) recently visited by the current user.
• D_pzFollowed: Items followed by the current user.
• D_pzApplicationDomainKnowledge: Default (empty) domain knowledge. To customize, save the pyDomainKnowledgeForCWYD Rule into your application Ruleset and edit it.

Key setup steps

• Ensure the relevant Case Types and Data Objects are enabled in the Data Configurator.
• Associate the pxChatWithYourData Tool with that Agent 
• Ensure the necessary data sources are configured to the Agent to pull the information.

Limitations to keep in mind

• Up to 200 fields per class (per enabled Case Type or Data Object) are processed.
• Deprecated Association Rules are not supported.
• For Index- and Link-automatically added classes, only properties from those classes are fetched—not from their parent classes.
• Index Rules linked to Data Objects through Associations are not fetched.
• Only one level of automatically added classes is supported; Associations for these are not supported.

Notes

• If a user’s question requires data from a disabled class, the Agent will not answer.
• The feature relies on Connect SQL for data retrieval and uses a generative model to compose natural-language answers.
• For broader coverage or better phrasing, customize domain knowledge and ensure all required classes are enabled.

Output:

Example1: Loan Application Cases by Month

Based on the data retrieved, I've analyzed the loan application cases by month. Here's a breakdown of the case count by month:

Monthly Case Distribution

| Month | Year | Number of Cases | |

-------|------|-----------------| |

December | 2025 | 7 | |

November | 2025 | 36 | |

December | 2025 | 7 | |

May | 2025 | 2 | |

April | 2025 | 1 | |

March | 2025 | 5 | |

January | 2025 | 1 | |

December | 2024 | 3 | |

Key Observations:

1. 1. Highest Volume Month: November 2025 has the highest number of loan applications with 36 cases, representing approximately 53% of all applications.
   2. Recent Trend: There's a significant increase in loan applications in the last quarter of 2025, with December and November combined accounting for 43 cases (63% of total volume).
   3. Seasonal Pattern: The data suggests a seasonal pattern with higher application volumes toward the end of the calendar year.
   4. Monthly Distribution: The distribution shows uneven case creation across months, with some months having minimal activity (1-2 cases) while others show significant spikes.
   5. Year-over-Year Comparison: Comparing December 2024 (3 cases) to December 2025 (7 cases) shows a 133% increase, suggesting potential growth in the loan application business.

This monthly breakdown provides insights into application volume patterns that could help with resource planning and identifying seasonal trends in loan application processing.

Example2: Key Trends in Loan Application Data

Based on the analysis of the loan application data, I've identified several notable trends:  

Volume Trends

1. 1. Significant End-of-Year Spike: There's a dramatic increase in loan applications during Q4 2025, with November 2025 showing the highest volume (36 applications) - approximately 53% of all applications.
   2. Year-over-Year Growth: Comparing December 2024 (3 cases) to December 2025 (7 cases) shows a 133% increase, indicating substantial growth in loan application volume.
   3. Seasonal Pattern: The data reveals a clear seasonal pattern with higher application volumes toward the end of the calendar year, which may be related to holiday spending, year-end financial planning, or promotional activities.  

Loan Characteristics Trends

1. 1. Consistent Loan Amount: The majority of applications request $50,000, showing consistency in the loan amount sought by applicants.
   2. Purpose Consistency: "Home Improvement" and "Home renovation" are the predominant loan purposes across applications, suggesting a specific market focus.
   3. Mortgage Status Pattern: Most applicants have existing mortgages, typically reporting "1 active mortgage" with balances around $125,000-$300,000  

Processing Trends

1. 1. Status Distribution: Most applications remain in "New" status, with fewer progressing to "Review-LoanEligibility," "Pending-Approval," or "Loan-Approved" statuses, suggesting potential bottlenecks in the approval pipeline.
   2. Applicant Profile Consistency: Applications show a pattern of similar applicant profiles (primarily "John Doe" with consistent contact information and employment details), which may indicate either a template-based testing environment or a targeted customer segment.
   3. Documentation Patterns: There's a consistent pattern in acknowledgment letter generation, suggesting a standardized initial response process.  

Business Implications

1. 1. Resource Planning Opportunity: The significant volume increase in Q4 2025 suggests a need for increased processing capacity during this period.
   2. Process Efficiency Concerns: The large number of applications remaining in "New" status may indicate processing delays or efficiency issues in the loan review workflow.
   3. Market Focus: The consistency in loan purpose (home improvement) suggests a potential opportunity to develop specialized products or streamlined processes for this specific use case.

Demo: 

Pega GenAI Cookbook - Recipes series

Enjoyed this article? See more similar articles in Pega Cookbook - Recipes 🔥🔥🔥 series.