Ai Engines

AI engines extract and predict data values from your documents using machine learning models trained on your document types and corrected annotations.

An AI engine is designed to extract and predict data from your documents.

Generic Engines

Generic Engines are ready-to-use, pre-built engines available to our customers right away. There are different versions of Generic Engines to choose from. Engines listed below are legacy solutions that we continue to maintain and support for our existing customers who use them.

AP & AR Generic Engine

This engine is specifically designed to extract information from financial documents, mainly for Accounts Payable (AP) and Accounts Receivable (AR) tasks. It comes pre-trained to recognise a set list of fields.

While the engine doesn’t learn directly from your specific documents or annotations, it uses additional prediction methods to improve its accuracy. These fine-tuning improvements only apply to documents with very similar layouts.

List of fields the AP & AR Generic Engine is pre-trained to recognise

Document-level fields
Document type Type of the document
Currency Currency of the document
Language Language of the document
Data fields
Bank Account Bank account number
Sort Code Bank code that identifies bank and branch location
IBAN Bank account number in IBAN format
BIC/SWIFT Bank BIC or SWIFT code
Constant Symbol Statistical code on payment order
Customer Number Number assigned to a customer registered in a supplier’s accounting system
Date Due Invoice payment due date
Issue Date Document date of issue
Tax Point Date The date of taxable event
Document Identifier Document number
Order Number Purchase order identification number
Recipient Address Address of the customer
Recipient Tax Number Tax identification number of the customer
Recipient Company ID Company identification number of the customer
Recipient Name Name of the customer
Recipient VAT Number Customer VAT Number
Recipient Delivery Name Name of the recipient to whom the goods will be delivered
Recipient Delivery Address Address of the recipient where the goods will be delivered
Supplier Address Address of the supplier
Supplier Tax Number Tax identification number of the supplier
Supplier Company ID Business/organization identification number of the supplier
Supplier Name Name of the supplier
Supplier VAT Number VAT identification number of the supplier
Supplier Email Email of the supplier
Supplier’s Order ID Internal order ID in the suppliers system
Specific Symbol Payee ID on the payment order or similar
Terms Payment terms as written on the document (e.g. 45 days upon receipt)
Payment Reference In some countries, used by the supplier to match the payment received against the invoice
Payment Method Payment method defined on a document (e.g. Cheque, Pay order, Before delivery)
Delivery Note ID Delivery note ID defined on the invoice
Place of Supply Place of supply (the name of the city or state where the goods will be supplied)
Amounts
Amount Due Final amount, including tax to be paid after deducting all discounts and advances
Amount Rounding Remainder after rounding up total amount
Total Amount Subtotal of all items, including tax
Amount Paid Amount that has already been paid
Tax Base Total Base amount for tax calculation
Tax Total Total tax amount
Tax details table
Tax Base Sum of tax bases for items with the same tax rate
Tax Rate One of the tax rates in the tax breakdown
Tax Amount Sum of taxes for items with the same tax rate
Tax Total Total amount, including tax for all items with the same tax rate
Tax Code Text on document describing tax code of the tax rate (e.g. GST, CGST, DPH, TVA). If multiple tax rates belong to one tax code on the document, the tax code will be assigned only to the first tax rate
Line items table
Item Code/Id Can be the SKU, EAN, a custom code (string of letters/numbers), or even just the line number
Item Description Line item description. Can be multi-line with details.
Item Quantity Quantity of the item
Item Unit of Measure Unit of measure of the item (kg, container, piece, gallon, …)
Item Rate Tax rate for the line item
Item Tax Tax amount for the line. Rule of thumb: tax = rate * amount_base.
Amount Base Unit price without tax (this is the primary unit price extracted)
Amount Unit price with tax. Rule of thumb: amount = amount_base + tax.
Amount Total Base The total amount to be paid for all the items, excluding tax. Rule of thumb: amount_total_base = amount_base * quantity.
Amount Total The total amount to be paid for all the items, including tax. Rule of thumb: amount_total = amount * quantity.
Other Unrecognised data type

Chinese Invoices

Rossum’s Chinese Invoices engine has been pre-trained to recognise data fields in Governmental Tax Invoices from Mainland China (fapiao).

List of fields the Chinese Invoices Generic Engine is pre-trained to recognise

Basic information
Machine code Machine code
Invoice Code Invoice code
Invoice Number Invoice number
Check Code Check code
Invoice Date Invoice date
Remarks

The field may contain various notes, such as:

Purchase order number

Supplier code

Order number

Check code

Handwritten notes

Note about the total quantity and weight

Address

Date and Import method (by air, sea, etc)

Payment terms

Customer number

Date of shipping

Discount

Contract number

Serial number

Construction name

Construction location

Payment instructions
Seller’s bank name Bank name of the seller
Seller’ account number Account number of the seller
Buyer’s bank name Bank name of the buyer
Buyer’s account number Account number of the buyer
VAT & amounts
Total without tax Base amount for tax calculation
Total tax Total tax amount
Amount due Final amount, including tax to be paid
Seller and buyer details
Seller’s name Name of the seller
Seller’s VAT number VAT identification number of the seller
Seller’s address Address of the seller
Seller’s phone number Phone number of the seller
Buyer’s name Name of the buyer
Buyer’s VAT number Buyer VAT Number
Buyer’s address Address of the buyer
Buyer’s phone number Phone number of the buyer
Line items
Model The model number or code
Quantity Quantity of the item
Unit Unit of measure of the item
Unit price Unit price without tax (this is the primary unit price extracted)
Total amount base The total amount to be paid for all the items, excluding tax
VAT rate Tax rate for the line item
Total tax Tax amount for the line
Description Description of the item

Certificates of Analysis

Certificates of analysis are quality control documents that are common in the food and beverage industry. These documents confirm that the received food products comply with the desired parameters and targets.

List of fields the Certificates of Analysis Generic Engine is pre-trained to recognise

Basic information
Supplier Name Name of laboratory who performed the analysis
Product Code Identifier of the product
Sample ID Sample identifier usually assigned by a laboratory
Purchase Order Purchase order number
Issue Date Issue date of the certificate in D/M/YYYY format
Batch number Batch number / Lot number of the product
Manufacturing Date Manufacturing date in D/M/YYYY format
Parameters 1 table - table of the analysis containing the parameters.
Description Description of a measured parameter
Result Result of the analysis, usually a number
Unit of measurement Unit of measurement. E.g. mg, kg, 100g/ml , g/100 ml, °C, %
Method Method code or description
Parameters 2 table - table of the analysis containing the parameters. This is a secondary table, which often has a different layout.
Description Description of a measured parameter.
Result Result of the analysis, usually a number, or < LOQ.
Unit of measurement Unit of measurement. E.g. mg, kg, 100g/ml , g/100 ml, °C, %.

Aurora Generic Engines

The next generation of Generic Engines can quickly learn custom fields and adapt to new document formats.

Aurora Generic Engine

Aurora Generic Engine is pre-trained to recognise data/fields related to Accounts Payable and Receivable (the same set as AP & AR Generic Engine), suitable to process transactional documents like Invoices, Purchase Orders, Sales Orders, Credit Notes, Debit Notes. Additionally it learns custom fields and new document formats rapidly.

List of fields the Aurora Generic Engine is pre-trained to recognise

Document-level fields
Document type Type of the document
Currency Currency of the document
Language Language of the document
Data fields
Bank Account Bank account number
Sort Code Bank code that identifies bank and branch location
IBAN Bank account number in IBAN format
BIC/SWIFT Bank BIC or SWIFT code
Constant Symbol Statistical code on payment order
Customer Number Number assigned to a customer registered in a supplier’s accounting system
Date Due Invoice payment due date
Issue Date Document date of issue
Tax Point Date The date of taxable event
Document Identifier Document number
Order Number Purchase order identification number
Recipient Address Address of the customer
Recipient Tax Number Tax identification number of the customer
Recipient Company ID Company identification number of the customer
Recipient Name Name of the customer
Recipient VAT Number Customer VAT Number
Recipient Delivery Name Name of the recipient to whom the goods will be delivered
Recipient Delivery Address Address of the recipient where the goods will be delivered
Supplier Address Address of the supplier
Supplier Tax Number Tax identification number of the supplier
Supplier Company ID Business/organization identification number of the supplier
Supplier Name Name of the supplier
Supplier VAT Number VAT identification number of the supplier
Supplier Email Email of the supplier
Supplier’s Order ID Internal order ID in the suppliers system
Specific Symbol Payee ID on the payment order or similar
Terms Payment terms as written on the document (e.g. 45 days upon receipt)
Payment Reference In some countries, used by the supplier to match the payment received against the invoice
Payment Method Payment method defined on a document (e.g. Cheque, Pay order, Before delivery)
Delivery Note ID Delivery note ID defined on the invoice
Place of Supply Place of supply (the name of the city or state where the goods will be supplied)
Amounts
Amount Due Final amount, including tax to be paid after deducting all discounts and advances
Amount Rounding Remainder after rounding up total amount
Total Amount Subtotal of all items, including tax
Amount Paid Amount that has already been paid
Tax Base Total Base amount for tax calculation
Tax Total Total tax amount
Tax details table
Tax Base Sum of tax bases for items with the same tax rate
Tax Rate One of the tax rates in the tax breakdown
Tax Amount Sum of taxes for items with the same tax rate
Tax Total Total amount, including tax for all items with the same tax rate
Tax Code Text on document describing tax code of the tax rate (e.g. GST, CGST, DPH, TVA). If multiple tax rates belong to one tax code on the document, the tax code will be assigned only to the first tax rate
Line items table
Item Code/Id Can be the SKU, EAN, a custom code (string of letters/numbers), or even just the line number
Item Description Line item description. Can be multi-line with details.
Item Quantity Quantity of the item
Item Unit of Measure Unit of measure of the item (kg, container, piece, gallon, …)
Item Rate Tax rate for the line item
Item Tax Tax amount for the line. Rule of thumb: tax = rate * amount_base.
Amount Base Unit price without tax (this is the primary unit price extracted)
Amount Unit price with tax. Rule of thumb: amount = amount_base + tax.
Amount Total Base The total amount to be paid for all the items, excluding tax. Rule of thumb: amount_total_base = amount_base * quantity.
Amount Total The total amount to be paid for all the items, including tax. Rule of thumb: amount_total = amount * quantity.
Other Unrecognised data type

Aurora Instant Learning

This engine is not pre-trained on any specific data. It quickly learns to recognise custom data using user annotations and is recommended for custom use cases.

Dedicated Engines

Dedicated engines are designed for a specific use case and are trained using documents and annotations from a particular customer. Our team regularly re-trains these engines to keep them updated. Unlike Aurora, dedicated engines do not offer instant learning; training results are only visible after our team completes the re-training process, not after just a few annotations. Dedicated Engines are legacy solutions that we continue to maintain and support for our existing customers who use them.