What is Document intelligence?
Document intelligence is the use of OCR, layout analysis and language models to read documents such as invoices, contracts and forms and turn them into structured, validated data that a system can act on.
In more detail
It covers the same ground as intelligent document processing (IDP), and the two terms are mostly used interchangeably. Microsoft, for example, brands its own service Azure AI Document Intelligence. The pipeline has stages: classify the document, extract fields and tables, validate against rules and reference data, then route low-confidence items to a person. Newer language-model extraction handles layouts it has never seen without a template, while classic OCR on its own only reads characters. Accuracy depends on document quality and on validation. Checking a total against its line items catches more errors than a better model does. We measure field-level accuracy and the share of documents needing review, using the client's own documents and not a vendor sample.
Example
A logistics firm sends bills of lading, carrier invoices and customs forms through one pipeline. Each is classified, its fields extracted and checked against the shipment record. A mismatch on weight or consignee goes to a clerk with the page image beside the extracted values.