How Accounts Payable Operations integration with Document Intelligence works
Summarize
Summary of How Accounts Payable Operations integration with Document Intelligence works
ServiceNow’s Accounts Payable Operations integration with Document Intelligence automates the extraction and ingestion of invoice data from email attachments into Accounts Payable Operations. This integration streamlines invoice processing by creating invoice cases, extracting data, validating required fields, and managing invoice states to facilitate efficient accounts payable workflows.
Show less
Invoice Ingestion Process Flow
The ingestion flow varies based on the confidence level of data extraction by Document Intelligence:
- High Extraction Confidence: An invoice case is created and invoice data is extracted and stored temporarily. The data is transformed and imported to create a draft invoice linked to the case. The system verifies that all required fields and at least one invoice line exist. If fields or lines are missing, Accounts Payable Specialists must manually complete and submit the invoice. Currency codes are resolved using purchase order data or user session locale if needed. Upon submission, the invoice state changes to Received, triggering further processing.
- Low Extraction Confidence: An invoice case is created but data is not automatically extracted. The Accounts Payable Specialist must review the invoice in Document Intelligence, manually enter required values, and submit. The subsequent validation and processing steps mirror those of the high confidence flow.
- Transformation Failure: If the transformation process fails completely, an error message prompts users to contact an administrator for resolution.
Field Validation and Data Transformation
The integration ensures all required invoice and invoice line fields are populated before creating or submitting invoices. Data transformation logic converts dates, currencies, unit prices, and decimals into formats compatible with downstream systems. Currency codes are prioritized from purchase orders or user locale settings, with manual correction required if these are missing or unrecognized.
Localization
While Accounts Payable Operations supports multiple languages, the current Document Intelligence model is trained only for English invoice extraction. To process invoices in other supported languages, customers must train their own Document Intelligence models.
Practical Benefits for ServiceNow Customers
- Automates invoice data capture from emails, reducing manual data entry and errors.
- Ensures data completeness and integrity through required field validation and currency handling.
- Provides clear workflows for handling invoices with varying extraction confidence, improving processing efficiency.
- Supports customization for multi-language invoice processing through model training.
- Enables proactive error handling by notifying administrators on transformation failures.
Document Intelligence uses automated workflows to extract invoice data from email attachments and populate invoice records in Accounts Payable Operations.
Invoice ingestion process flow when extraction confidence is high
- An invoice case is created with category Invoice automation and subcategory Invoice processing in Accounts Payable Operations.
- If Document Intelligence determines that the extraction confidence is higher than the threshold value, it extracts the invoice data and stores the data in temporary staging tables.
- The Document Intelligence transformation process imports the invoice data from the staging tables, creates an invoice in the Draft state, and associates the invoice with the invoice processing case.
- The application verifies the following:
- The required fields are populated on the invoice.
- If the required fields aren’t populated on the invoice, the Accounts Payable Specialist is required to open the invoice, manually enter the required values in the required fields, and submit the invoice.
- If the required fields are populated on the invoice, the invoice is auto-submitted.
- At least one invoice line record exists for the invoice. If not, the Accounts Payable Specialist must add an invoice line record and submit the invoice.
- If a unique currency code isn’t identified during currency transformation, the process uses the currency mentioned in the purchase order. If the currency is missing in the purchase order, the transformation process uses the session currency based on the user's locale. In this case, the Accounts Payable Specialist is required to open the invoice processing case, and enter the correct currency code in the invoice fields and submit the invoice. For more information, see Invoice data transformation logic.
For more information about the required fields for invoice and invoice lines, see Required fields for creating an invoice.
After the invoice is submitted, its state changes to Received. The invoice processing begins from this point. For more information, see Work on an invoice processing case.
- The required fields are populated on the invoice.
Invoice ingestion process flow when extraction confidence is low
- An invoice case is created with category Invoice automation and subcategory Invoice processing in Accounts Payable Operations.
- If Document Intelligence determines that the extraction confidence is lower than the threshold value, it doesn’t extract the invoice data and as a consequence an invoice isn’t created.
- In this scenario, the Accounts Payable Specialist must do the following:
- Open the invoice processing case created in step 1.
The case displays the following error message: Invoice has a data extraction error. Review invoice with automation to continue processing.
- Select Review invoice, which opens Document Intelligence and displays the invoice fields.
- Enter the required values into the invoice fields, and select Submit.
- Open the invoice processing case created in step 1.
- The Document Intelligence transformation process imports the invoice data from the staging tables, creates an invoice in the Draft state, and associates the invoice with the invoice processing case.
- The application verifies the following:
- The required fields are populated on the invoice.
- If the required fields aren’t populated on the invoice, the Accounts Payable Specialist is required to open the invoice, manually enter the required values in the required fields, and submit the invoice.
- If the required fields are populated on the invoice, the invoice is auto-submitted.
- At least one invoice line record exists for the invoice. If not, the Accounts Payable Specialist must add an invoice line record and submit the invoice.
- A unique currency code exists on the invoice. If a unique currency code isn’t identified during currency transformation, the process uses the currency mentioned in the purchase order. If the currency is missing in the purchase order, then the transformation process uses the session currency based on the user's locale. In this case, the Accounts Payable Specialist is required to open the invoice processing case and enter the correct currency code in the invoice fields and submit the invoice. For more information, see Invoice data transformation logic.
For more information about the required fields for invoice and invoice lines, see Required fields for creating an invoice.
After the invoice is submitted, its state changes to Received. The invoice processing begins from this point. For more information, see Work on an invoice processing case.
- The required fields are populated on the invoice.
Invoice ingestion process flow when the entire transformation process fails
When the Document Intelligence transformation process fails entirely, the following error message is displayed on the invoice processing case:
Invoice automation is experiencing issues. Please contact the administrator.
You must collaborate with the administrator to resolve this issue.
Localization information
Accounts Payable Operations supports multiple languages. The current DocIntel model is trained to extract invoices in the English language only. If you want to process the invoice in multi-lingual languages supported by DocIntel, you must train the DocIntel model on your own.
For more information on Document Intelligence, see Document Intelligence.