Using the invoice model

Your business both issues invoices and receives them from partner organization. There might be many different formats on paper or in digitized forms and some will have been scanned poorly at odd angles or from creased paper.

The invoice model in Azure AI Document Intelligence can handle these challenges and uses the features of the read model to extract text from invoice scans. In addition, it extracts specific fields that are commonly used on invoices including:

• Customer name and reference ID
• Purchase order number
• Invoice and due dates
• Details about the vendor, such as name, tax ID, physical address.
• Similar details about the customer.
• Billing and shipping addresses.
• Amounts such as total tax, invoice total, and amount due.

Invoices also feature lines, usually in a table, each of which is one purchased item. For each line, the invoice model identifies details including:

• The description and product code of the product or service invoiced.
• Amounts such as the unit price, the quantity of items, the tax incurred, and the line total.

Using the receipt model

Receipts have similar fields and structures to invoices, but they record amounts paid instead of amounts charged. Azure AI Document Intelligence faces the same challenges of poor scanning or digitization but can reliably identify fields including:

• Merchant details such a name, phone number, and address.
• Amounts such as receipt total, tax, and tip.
• The date and time of the transaction.

As for invoices, receipts often include a table of items, each of which is a product or service purchased. For each of these lines, the model recognizes:

• The name of the item.