Azure AI Document Intelligence: Extract Data from Documents

Azure AI Document Intelligence

Azure AI Document Intelligence (formerly Form Recognizer) uses AI to extract text, key-value pairs, tables, and structures from documents. It handles invoices, receipts, IDs, tax forms, contracts, and custom document types.

Key Capabilities

• Prebuilt Models — Ready-to-use models for common document types
• Custom Models — Train on your own document formats
• Document Classification — Automatically sort documents by type
• Add-on Features — Handwriting, barcodes, formulas, font styling

Prerequisites

• Azure subscription
• Create a Document Intelligence resource in the Azure Portal

Step 1: Try the Studio

Document Intelligence Studio lets you test models visually.

1. Go to documentintelligence.ai.azure.com
2. Choose a prebuilt model (e.g., Invoice)
3. Upload a sample document or use the provided samples
4. Click Analyze and review extracted fields

Step 2: Use Prebuilt Models

Invoice Model

from azure.ai.documentintelligence import DocumentIntelligenceClient
from azure.core.credentials import AzureKeyCredential

client = DocumentIntelligenceClient(
    endpoint="https://my-doc-intel.cognitiveservices.azure.com/",
    credential=AzureKeyCredential("your-key")
)

with open("invoice.pdf", "rb") as f:
    poller = client.begin_analyze_document(
        model_id="prebuilt-invoice",
        body=f
    )
result = poller.result()

for invoice in result.documents:
    print(f"Vendor: {invoice.fields['VendorName'].content}")
    print(f"Invoice #: {invoice.fields['InvoiceId'].content}")
    print(f"Total: {invoice.fields['InvoiceTotal'].content}")
    print(f"Due Date: {invoice.fields['DueDate'].content}")

    # Line items
    for item in invoice.fields.get("Items", {}).get("valueArray", []):
        desc = item["valueObject"]["Description"]["content"]
        amount = item["valueObject"]["Amount"]["content"]
        print(f"  - {desc}: {amount}")

Available Prebuilt Models

Model	Use Case
prebuilt-invoice	Invoices and bills
prebuilt-receipt	Receipts from retail/restaurants
prebuilt-idDocument	Passports, driver's licenses
prebuilt-tax.us.w2	US W-2 tax forms
prebuilt-healthInsuranceCard.us	Insurance cards
prebuilt-contract	Contracts and agreements
prebuilt-layout	General document structure
prebuilt-read	OCR text extraction

Step 3: Build Custom Models

For documents unique to your business:

1. Gather Training Data

• Minimum 5 sample documents (recommend 15+)
• Ensure variety in layouts and content

2. Label Your Documents

1. Go to Document Intelligence Studio
2. Create a new Custom extraction model project
3. Upload your training documents
4. Label the fields you want to extract (drag to select text regions)

3. Train the Model

poller = client.begin_build_document_model(
    build_mode="template",  # or "neural" for varied layouts
    model_id="my-purchase-order",
    blob_container_url="https://mystorage.blob.core.windows.net/training-data"
)
model = poller.result()
print(f"Model ID: {model.model_id}")
print(f"Fields: {[f for f in model.doc_types['my-purchase-order'].field_schema]}")

Template vs Neural

Feature	Template	Neural
Layout	Fixed/similar	Varied
Training data	5+ docs	15+ docs
Speed	Fast	Slower
Best for	Structured forms	Semi-structured docs

Step 4: Automate with Logic Apps / Power Automate

Power Automate Example

Trigger: When a file is added to SharePoint "Invoices" folder
→ Extract invoice fields using Document Intelligence
→ If amount > $10,000: Send approval to manager
→ Create row in Excel with extracted data
→ Archive the processed document

.NET Integration Example

using Azure.AI.DocumentIntelligence;
using Azure;

var client = new DocumentIntelligenceClient(
    new Uri("https://my-doc-intel.cognitiveservices.azure.com/"),
    new AzureKeyCredential("your-key"));

var content = new AnalyzeDocumentContent
{
    UrlSource = new Uri("https://example.com/invoice.pdf")
};

var operation = await client.AnalyzeDocumentAsync(
    WaitUntil.Completed, "prebuilt-invoice", content);

var result = operation.Value;
foreach (var doc in result.Documents)
{
    Console.WriteLine($"Vendor: {doc.Fields["VendorName"].Content}");
    Console.WriteLine($"Total: {doc.Fields["InvoiceTotal"].Content}");
}

Resources

• Document Intelligence Documentation
• Studio
• Quickstart
• Microsoft Learn Training

Video: Search "Azure AI Document Intelligence" on Microsoft Azure YouTube for the latest walkthroughs.