In this session of the Predictable AI Placement series, Jay Stewart, Solution Consultant Fellow at Pega, takes a deep dive into the Doc Agent—the second of the five AI placement patterns—and shows how it uses GenAI Connect to process and extract value from documents at runtime.

Using a compliance and audit use case, Jay demonstrates how the Doc Agent is triggered when an auditor uploads supporting documentation during the gather audit details step. The agent automatically analyzes the attached documents and generates a structured summary that is written directly back to the case for review.

You’ll see how the Doc Agent:

• Uses a GenAI Connect rule to summarize uploaded audit documents
• Writes AI‑generated summaries into case properties
• Supports human‑in‑the‑loop review to validate and refine results
• Reduces manual effort while maintaining accuracy and control

The session also explores a broader intelligent document processing scenario, showing how the same pattern can be used to extract multiple structured fields from documents such as bills of lading, even when the document format is unfamiliar. In this example, GenAI Connect extracts both scalar values and list data and maps them directly to an existing case data model.

Jay walks through both design‑time and runtime aspects, including:

• Configuring document processing rules in the AI Designer
• Using attachment fields as source inputs
• Defining extraction prompts and target properties
• Routing documents dynamically based on document type

This deep dive builds on the Application Agent covered earlier in the series and sets the foundation for the next placement pattern—the Step Agent—continuing the end‑to‑end walkthrough of how AI placement patterns work together to support agentic workflows in Pega.