How Sage Intacct populates draft vendor invoices

Artificial Intelligence/Machine Learning (AI/ML) draws data from several sources when populating the header, line item, and subtotal information in draft vendor invoice. These sources include the following:

  • AI/ML predictions
  • The vendor document that you uploaded or emailed
  • The transaction definition for the vendor invoice
  • The source purchasing transaction, such as a matched purchase order

You can find specific field sources based on the transaction type.

Header information

AI/ML uses predictions to match the vendor document to the following:

  • The correct vendor information record
  • The source purchasing transaction, which is shown in the Converted from field, when document matching is selected

After the vendor and source purchasing transaction are identified, header fields are populated using data from the vendor information record, the uploaded or emailed vendor document, and the source purchasing transaction.

Line items

With header-level document matching, line items are copied directly from the source purchasing transaction. You can then compare these to the attachment to see if further edits are required.

With line level matching enabled, AI/ML uses predictions to populate line items. When AI/ML detects a near match that's not exact, Sage Intacct alerts you with an exclamation point (!) in the Exception column for that line.

For automated transactions without document matching, line items are predicted by AI/ML.

Subtotals

The Subtotals section of the vendor invoice is populated using the discounts and charges you defined in the transaction definition template that was used to create the vendor invoice.

If a discount or charge on the vendor document varies from what was defined on the transaction definition, you can edit the value on the vendor invoice.

The Subtotals section does not appear in the following circumstances:
  • When the vendor and all items IDs are non-taxable
  • When Subtotals were not enabled in the transaction definition

Date formats

AI uses various parameters to interpret date formats in documents, including the following factors:

  • Vendor country: Determines regional date format preferences. For example, 4/11/2025 could be April 11 or November 4.

  • Historical data: Uses past data to identify common date patterns.

  • Template models: Applies trained models specific to vendors for improved accuracy.

  • Date string structure: Analyzes the format to decide which part represents the day, month, and year.

  • Contextual signals: Considers surrounding text to assign meaning. For example, due date or issue date.

By combining these elements, AI can accurately extract and interpret dates from documents.