According to Andrew Ng, Agentic Document Extraction now supports field extraction, enabling users to extract specific fields such as vendor name, item list, and prices from images or PDFs of invoices and structured documents (source: Andrew Ng, Twitter, July 10, 2025). This upgrade allows businesses to automate data capture from common business documents, significantly reducing manual entry and boosting operational efficiency. The feature is expected to streamline AI-powered document processing workflows, opening new opportunities for enterprises in finance, logistics, and administration to leverage AI for automated data extraction and reduce costs.

According to Andrew Ng, Agentic Document Extraction has dramatically reduced its median PDF processing time from 135 seconds to just 8 seconds. This AI-driven tool now extracts not only text but also diagrams, charts, and form fields from PDFs, producing outputs optimized for large language models (LLMs). This breakthrough enables faster and more comprehensive data extraction, improving automation and accuracy in industries such as finance, legal, and healthcare where rapid document analysis is critical. The speed and versatility of Agentic Document Extraction present significant business opportunities for enterprises seeking to streamline workflows and leverage AI for document-intensive operations (source: Andrew Ng on Twitter, May 27, 2025).