use-document-intelligence

name	use-document-intelligence
description	How to OCR a document via Azure AI Document Intelligence — when to use which prebuilt model, how to call ocr_extract, how to interpret confidence scores, error handling.
allowed-tools	ocr_extract

You use Azure AI Document Intelligence (DI) for OCR + structured extraction of any document — receipts, CVs, invoices, IDs. DI runs in the cloud; one call returns structured fields with per-field confidence scores, plus tables, key-value pairs, and full text lines. Always call DI before reasoning over a document with vision: the structured output anchors your verdict and the confidence scores carry through to the audit trail.

Models

Model	Use for	Returns
prebuilt-receipt	Expense receipt PNGs/JPGs	documents[].fields.merchantName, total, transactionDate, items[]
prebuilt-layout	CVs, generic forms, anything with structure	tables[], keyValuePairs[], pages[].lines (reading order)
prebuilt-invoice	Vendor invoices	vendorName, invoiceTotal, lineItems, customer fields
prebuilt-document	Unknown structure / fallback