Azure AI Document Intelligence

Azure AI Document Intelligence Cheat Sheet

Azure AI Document Intelligence, previously known as Azure Form Recognizer, is a cloud-based AI service from Microsoft Azure designed to automate data extraction from various documents. By leveraging machine learning, it can analyze and extract key information such as text, tables, checkboxes, and other structured data from forms, invoices, receipts, and more. The service offers customization options to enhance data extraction accuracy for specific document formats or use cases.

Key Benefits:

• Automates data extraction from documents

• Helps eliminate manual data entry and validation

• Scalable and highly customizable for different document formats

• Supports integration with other Azure services and business processes

Features:

• Document Analysis – Extract actionable data from forms and documents to drive data-driven decisions and integrate it into your data visualization services for analysis.

  • Text Recognition (OCR) – Extracts text from scanned documents, images, and PDFs, including handwritten text in some cases.

  • Table Extraction – Detects and extracts tables from documents, enabling structured analysis.

  • Key-Value Pair Extraction: Automatically extracts relationships between key-value pairs in forms and invoices.
    

• Prebuilt Models – Extract data from various specialized document types.

  • Invoice Processing – Automatically extracts key data points from invoices (e.g., vendor name, invoice number, amount, date).

  • Receipt Processing – Extracts information like merchant name, transaction date, total amount, and items purchased.

  • Business Card Recognition – Extracts contact details such as names, titles, email addresses, and phone numbers from business cards.

  • Identity Document Processing – Recognizes and extracts data from passports, driver’s licenses, and other identity documents.

  • US Personal Tax – Classify and extract data from documents like W2s, W4s, 1040s, 1099s, and others related to personal tax filings.

  • US Mortgage – Extract information from mortgage-related documents, such as 1003, 1004, closing disclosures, and more.

  • US Pay Stubs – Extract employee details including earnings, deductions, net pay, and other payment-related information from pay stubs.

  • US Bank Statements – Extract data such as transaction dates, account numbers, bank information, and more from bank statements.

  • US Checks – Extract information from checks, including amount, date, MICR numbers, payer’s name, and payer’s address.

  • Credit Cards – Extract card details, including card number, cardholder’s name, due date, and associated bank information from credit card documents.

  • US Marriage Certificates – Extract information like the couple’s names, birth dates, marital status, nationality, parent details, and officiant information from marriage certificates.

  • Contracts – Extract information on titles and signatory parties (names, reference names, addresses) from contracts.

  • Business Cards – Extract personal and professional details such as name, job title, company, address, email, and phone number from business cards.

• Custom Models – Train custom models to classify documents and extract text, layout, and key fields from your forms or structured documents.

  • Custom Extraction Model – Build and label a model to extract structured data from your forms and documents.

  • Custom Classification Model – Train a model to categorize documents based on their content automatically.

• Simple Text Extraction – Easily extract and organize data from documents using prebuilt and custom features—no manual labeling needed.

• Customized Results – Receive tailored output based on your document layouts with automatic custom extraction, which can be further refined with human feedback.

• Flexible Deployment – Ingest data from the cloud or at the edge, and apply it to search indexes, business automation workflows, and more.
  

• Built-in Security – Benefit from enterprise-grade security and privacy for your data and any trained models.

• Intelligent Search Indexes – Quickly locate specific information within your documents and forms, like total accounts payable, by integrating AI Document Intelligence with Azure Applied AI Search.

• Automated Business Workflows – Extract text, key-value pairs, tables, and other structures from documents, and streamline claims, invoices, and receipt handling by integrating with back-end systems.

Security:

Azure AI Document Intelligence adheres to high-security standards to ensure data privacy and compliance. Some key aspects of security include:

• Data Encryption: Data is encrypted in transit and at rest using Azure’s standard security protocols.

• Access Control: Integration with Azure Active Directory (Azure AD) ensures that only authorized users and applications can access the service.

• Compliance: Azure complies with various regulatory standards, including GDPR, HIPAA, SOC, and ISO certifications, ensuring the service can be used in regulated industries.

• Data Retention: Azure AI Document Intelligence does not store user data or documents by default. Data is processed and returned with minimal storage retention, but users can configure their retention policies.

• Audit Logs: Activity logs provide insights into who accessed the data and how it was processed, ensuring transparency and accountability.

• User Authentication: Leverages Azure AD for authenticating and authorizing users, with multi-factor authentication (MFA) options for added security.

Pricing:

Azure AI Document Intelligence offers flexible pricing based on the type of documents processed and the features used.

1. Pricing Models

   • Pay-as-you-go: Pricing is based on the number of pages or documents processed.

   • Custom Pricing: Enterprise-level pricing and discounts are available for large-scale deployments or specific business needs.

2. Pricing for Key Services

   • Azure AI Document Intelligence (Pre-built models):

     • Prices are based on the number of pages processed.

     • Typically priced per page, with different rates for invoices, receipts, and identity documents.

   • Custom Form Recognition:

     • Charges apply for training custom models and processing custom forms based on the number of pages or data processed.

   • OCR and Layout Processing:

     • Priced per page processed for document text recognition and layout extraction.

3. Free Tier

   • Free Tier: Azure AI Document Intelligence offers a free tier with a limited number of pages processed per month, which is ideal for small-scale testing or development.

4. Additional Costs

   • Storage and Storage Costs: Charges may apply for storing processed documents in Azure Blob Storage or other services.

   • Integration Costs: Additional costs for integrating with other Azure services like Azure Logic Apps or Power Automate.

5. Example Pricing:

   • Text Recognition (OCR): $1.50 per 1,000 pages.

   • Invoice Processing: $1.00 per invoice processed.

   • Custom Model Training: Prices may vary based on complexity and the volume of training data.

For up-to-date pricing details, always check the official Azure pricing page: Azure AI Document Intelligence Pricing

References:

• Overview of Azure AI Document Intelligence

• Azure AI Document Intelligence Pricing

• Azure Compliance and Security

• Document Intelligence Studio – Microsoft Azure