Azure AI Document Intelligence remains the trusted choice for many document-centric scenarios. Customers continue to rely on the industry leading OCR capability and structure extraction, including table recognition, figures, paragraphs, selection marks, sections, and more output in markdown format for easy integrations with LLMs for ingestion for RAG, field extraction and document chat scenarios. Document Intelligence has the tools to build scalable and flexible IDP solutions with classification and conditional routing for high-accuracy extraction from prebuilt models like invoices, receipts, tax forms, and identification cards. For any custom template, you can label a few samples to train a custom extraction model on any document type. Document Intelligence models have some limitation like supporting only extracting results, limited generalization of custom models across many template variations and limited semantic understanding capabilities. With confidence scores and grounded results, you can build an effective, low latency, consistent extractive document processing solution for most scenarios. Document Intelligence provides the following models:
0 commit comments