ChallengeAs in any financial institution, in Modulbank there are routine business processes, that take a lot of time and effort of the staff. For example, payment claims generation that are sent to the bank by the Federal Bailiffs Service. In order to relieve the employees and to minimize the number of mistakes they make, it was decided to automate the process. The bank's internal team had been working on this task for two months. The customer contacted Sibedge when it became clear that the project could not be finished by own forces. RPA experts were needed to create the robot.
ApproachThe developers carefully thought through the sequence of actions that the robot should perform when processing FBS payment claims. The business process was divided into several subprocesses to make it easier to identify bottlenecks and speed up execution time. Scanned documents received by the bank were to be recognized, analyzed, and entered into several internal systems of the bank. In case of any errors and inconsistencies, the robot should report this to the administrator.
While working on the project, the team encountered a problem of poor quality document processing. They had to conduct a comprehensive study in the field of printed text OCR. The robot was trained on documents scanned in different quality, partially scanned or with errors made when filling in. Optical Character Recognition (OCR) technology was used as a basis. In the course of the research various approaches and settings were selected to find the best option for Russian-language documents scanning.
UiPath platform was also used, along with the instruments to work with web-browsers and Microsoft Office package. A list of requirements to the quality and structure of documents was generated in the process, in compliance with which the robot would provide a guaranteed result with a minimum of errors. It took a month and a half for the experts to develop the project.
ResultSoftware solution was implemented in one of Modulbank’s offices in Novosibirsk. The customer's technical support made minor adjustments to the document scanning process, according to the requirements of the development team. It was important that payment claims would not be scanned upside down or partially. After that, the robot was fully launched.
Practice has shown that the robot processes documents at least four times as fast as a real person, makes almost no mistakes, works without breaks, weekends and vacations, and does not require any salary. The customer was satisfied with the result shown by Sibedge experts. After all, the automation of this business process allowed to free employees from routine work, speed up document flow and reduce the bank's costs.
- 1 - Project Manager
- 1 - RPA expert
- 1 - Analyst