Support Multiple File Types for Product Extraction AI Agent

Oracle CPQ 26C enhances the Product Extraction AI Agent by enabling users to submit and process multiple document formats within a single extraction experience. Users can upload purchase orders and related request documents in common business formats, including Microsoft Word documents, Microsoft Excel spreadsheets, image-based files, and email content.

The Product Extraction AI Agent processes both structured and semi-structured content and can extract information from scanned documents using Optical Character Recognition (OCR) technology when necessary. Extracted product information is returned in a structured format that can be reviewed and used as part of quote creation workflows.

The extracted output can include key product information such as the product identifier or product name, product details, product type, and requested quantity. When available, the extraction can also capture additional commercial information, including unit price and currency, providing users with a more complete view of the extracted product data for quote preparation and review.

The following example shows uploading a Microsoft Excel file in the AI-Assited Add Products Agent.

Example - Upload Microsoft Excel File in Product Extraction AI Agent

This enhancement reduces manual data entry from customer documents and accelerates the transition from request intake to quote preparation. Support for multiple input formats provides a consistent identification experience across a variety of customer-submitted content, helping improve efficiency in high-volume quoting scenarios.

Steps to enable and configure

You don't need to do anything to enable this feature.

Tips and considerations

  • Extraction accuracy depends on source document quality. Low-resolution, blurry, or heavily distorted files may reduce extraction quality.
  • Review confidence scores and source references as part of your validation process before creating or updating quotes.
  • Email content containing embedded images, forwarded messages, or complex formatting may produce different extraction results depending on how the content is rendered.
  • Although spreadsheets with varying column layouts are supported, using a consistent spreadsheet structure can improve extraction accuracy and streamline bulk line-item processing.