This page provides an explanation of both a demo site showing data extraction records and the AI Agent tool available natively in the Pragatix Chat

1. Product Document Processing management site

This website is used to demo and manage the data extraction process done on the Product catalog.

The website is organized into two main sections:

A. Product Catalog Page

• The default landing page displays a table of the company’s products.

• The table includes a filter option labeled “Has Processed Documents”. Selecting Yes will display only products whose documents have undergone data extraction.

• As a start, the system was configured to process all parts of the Cheese category. We processed PDF, Word, and Excel files, skipped images, and scanned the PDF.
  

  

  
  

• The product table displays basic information about each item and provides two action buttons: “View Documents” and “View Properties”.
  

  

  
  

  • View Documents

    • Opens the product’s dedicated documents page.

    • This page lists all documents associated with the selected product.

    • Each entry includes a document link.

    • If properties have been extracted from a document, an additional option is available to view those extracted properties.
      

      

      
      

  • View Properties

    • Opens the product’s properties page.

    • This page displays all properties related to the product, whether sourced directly from the catalog or extracted from documents.

    • Results can be filtered using the following options:

      • All: Shows every property, including catalog entries not found in documents.

      • Differences: Highlights inconsistencies between catalog data and document data.

      • Additions: Displays values found in documents that are missing from the catalog.

      • Differences and Additions: Combines both filters to show discrepancies and missing values together.

B. Product Ingredients page

This page compares the ingredients listed in the catalog with those extracted from product documents.

• Users can filter results by category (e.g., cheese, chocolate, etc.).

• Products can also be filtered by ingredient availability, making it easy to identify which catalog entries align with document data.

2. AI Agent Product Demosheet Tool

The purpose of this tool is to demonstrate the capabilities enabled by the product extraction system. It generates two types of product datasheets:

• Official Datasheet: Identical to the datasheet produced by the standard “Create Demosheet” tool.

• Unofficial Datasheet: Extends the official version by including missing ingredients, sourced directly from the document extraction process.

How to Use

• Using the Peterson Data analysis chat, enter a prompt that includes the term “demosheet” such as: “Create a demosheet for product 00679.”
  

Below are some examples of product IDs that represent cheese products with missing ingredients identified through document processing. These items are handy for trying out the Product Demosheet Tool, since they highlight how extracted data can supplement catalog information:

• 00115

• 00119

• 00383

A complete list of such products is available on the Product Document Processing website under the Ingredients tab, when filtering by Category = Cheese and Filter By = Additions