Invoice Processing using ML

We helped a leading insurance firm in Thailand overcome the burden of comprehending unstructured data using the RPA coupled with cognitive capabilities


The client has systems that would create scanned copy of invoice files in different formats (word, PDF, jpg etc.) and business users had to manually copy relevant content from invoice documents and feed it into invoicing processing system. Initially business users planned to mimic and replace the manual extraction and filling process using RPA. KGiSL performed a thorough analysisand during the implementation phase, it was identified that quality and accuracy of data extraction performed by RPA is not at satisfactory level. ​

Strategy & solution

As a common practice, invoice processing is automated using OCR. Business users would mention few co-ordinates from where content from invoice would get extracted. This approach of coordinates-based data extraction fails in most of case since the format of invoice drastically varies between different formats and files.KGiSL proposed a solution which combined RPA capabilities with cognitive aspects using ICR and ML based extraction enabling quality extraction from invoice is made possible. This completely eliminated the use of coordinates-based extraction. With the improved enhancements, any new invoice format could be easily accommodated by the system.


Faster invoice processing​

Eradicates manual invoice data extraction and filling​

Easily adapts when new invoice formats are received

Ready for your own success story?

For a FREE consultation on process discovery for automation or POC request, fill in the below form.

Quick Contact