The customer had to process about 1500 invoices/ day which was sent to them by 15 OEMs each with their own template. The invoices were structured and had pre-determined rules and conditions. On an average, each invoice was 3 pages long and was in the form of a scanned PDF or an image captured using a mobile. The data that was extracted from the invoice was fed into the CRM system after finding an appropriate mapping between the CRM data and invoice item name
The bot is designed to access the invoices from a FTP server and extract data from it. It then accesses an intermediary master data where the mapping between CRM and invoice item name could be done. Once the match is found, bot feeds the extracted data into appropriate fields in the system.