Description of the demand

ConwayCS is a seller/lessor of shipping containers and logistics services in Latvia. The company occupies 30% of the Latvian market for the sale and rental of shipping containers.

Due to the growing number of incoming documents processed, there is a need in invoice processing automation of the work of specialists involved in entering documents into electronic document management systems. When planning, the process of invoice management turned out to be the most indicative from the point of view of the automation effect. Despite the fact that an invoice is a structured document, a large amount of data needs to be extracted from it, which in turn slows down the processing of the incoming correspondence flow and carries the risks of errors when filling in.

The essence of the idea

To reduce the burden on specialists involved in processing incoming documents, automate the invoice scanning and data capture using artificial intelligence technologies as part of Graip.AI and further invoice data extraction to the Xero accounting platform used in the company.

It is necessary to extract data and automatically fill in the required fields in the document card with them:

  • Document number, date;
  • Shipper;
  • Consignee;
  • Seller:
  • Registration number;
  • Tax number;
  • Buyer:
  • Registration number;
  • Tax number;
  • Product ID;
  • Amount without VAT;
  • The amount;
  • Quantity.

Implementation option

Graip.AI is a tool for automatic recognition and classification of documents. Machine learning-based mechanisms allow the solution to independently determine the type, number and date of the document, as well as other details.

After scanning the incoming invoice, the solution extracts its actual data and automatically fills in the required fields with them in the document card of the accounting platform. A task is created in the system to verify the pre-filled details, the user only needs to check their correctness. Manual input of information is excluded and the time of entering documents into the system is reduced.

Business effect

When tested on real documents of the company, the solution showed the quality of the details of incoming invoices scanning and data capture at the level of 90-92%. Therefore, it is expected to achieve the planned effect and reduce the time for processing incoming documents, speeding up processes.

At the moment, the pilot testing stage has been completed and the transfer of the solution to trial operation is being prepared, as well as the expansion of the tested processing methodology to other types of incoming documents received by the company, including through e-mail channels.