🎯 Learning objectives

  • Explain what the Document Understanding Process template is and why it exists
  • Access and create a project from the template in UiPath Studio
  • Differentiate between the attended and unattended execution modes
  • Navigate the key workflow files and understand their role in the pipeline

🧩 What is the Document Understanding Process?

The Document Understanding Process is a ready-made UiPath Studio template that provides a complete, production-ready framework for building document processing automations. Instead of building from scratch, you start from a fully structured project that already implements best practices for the entire DU pipeline — OCR, classification, extraction, human validation, and export.

✅ Why Use the Template?

  • Saves significant development time — the pipeline structure is already built
  • Built-in error handling, logging, and retry logic
  • Supports both attended and unattended execution with minimal code changes
  • Follows UiPath best practices for long-running workflows and human-in-the-loop validation
  • Compatible with UiPath Studio version 21.4 or higher
📖 Official Documentation
For full reference, refer to the official UiPath docs: Document Understanding Process — User Guide

🚀 How to Access the Template

1
Studio Open UiPath Studio

Open UiPath Studio (version 21.4 or higher).

2
Studio Go to the Templates Tab

Navigate to the Templates tab in the Studio Backstage view (the Start screen when no project is open).

3
Search Search for the Template

Search for Document Understanding Process in the search box.

4
Create Create the Project

Click Use Template, define your file name and directory, then click Create. Studio will scaffold the full project structure.

💡 Tip
You can also find the template in UiPath Marketplace or via Studio Web if you prefer a cloud-based starting point.

⚙️ Project Execution Modes

A clean project created from this template includes two main deployment pathways, each suited to a different execution context:

🖥️ Main_Attended.xaml

Best suited for local desktop execution where developers or users validate anomalies directly on their screens in real time.

  • Human validation happens inline — the robot pauses and the user reviews results immediately
  • Uses the Validation Station and Classification Station activities
  • Ideal for: proof-of-concepts, lower-volume processes, and scenarios where a human is always present

☁️ Main_ActionCenter.xaml

Optimized for high-volume, long-running unattended server processes that push exceptions to the cloud via UiPath Action Center.

  • The robot runs fully unattended and creates Action Center tasks for any low-confidence results
  • Human validators review tasks in Action Center at their own pace — asynchronous validation
  • Ideal for: production deployments, large document batches, 24/7 processing

📋 Attended vs. Unattended — Quick Comparison

AspectMain_Attended.xamlMain_ActionCenter.xaml
Validation styleInline, real-timeAsync via Action Center
Robot typeAttendedUnattended (server)
Human interactionImmediate, on the same machineCloud-based, anytime
Best forLow-medium volume, local validationHigh-volume, production

🔄 Workflow Overview

Both execution modes share the same core DU pipeline. The template pre-wires the following steps — you configure and extend them for your specific documents:

🗂️ Key Workflow Files in the Template

FilePurpose
Main_Attended.xamlEntry point for attended execution — triggers inline validation stations
Main_ActionCenter.xamlEntry point for unattended execution — pushes exceptions to Action Center
Framework/Core workflow logic: initialize, process document, handle exceptions
Config.xlsxCentral configuration: file paths, confidence thresholds, Orchestrator asset names
DocumentProcessing.xamlThe DU pipeline: digitize → classify → extract → validate → export

🔄 Core Pipeline Steps (Shared by Both Modes)

  1. Load Taxonomy — load document types and fields from your DU project or local taxonomy file
  2. Digitize Document — run OCR to convert the document to machine-readable text
  3. Classify Document Scope — identify the document type
  4. Data Extraction Scope — extract field values using configured extractors
  5. Validation — human review of low-confidence results (inline or via Action Center)
  6. Train Classifiers / Extractors Scope — send validated corrections back as feedback
  7. Export — write validated data to Excel, database, or downstream system
📖 Deep Dive
For detailed configuration steps, troubleshooting, and extractor setup within the template, refer to the official guide: Document Understanding Process — User Guide

✅ Knowledge check

1. Which execution mode is best suited for a high-volume, 24/7 unattended document processing scenario?

2. Where do you find the Document Understanding Process template in UiPath Studio?