In the early days, documents and images were delivered by means of paper. Following the rise of email and word processors and the full construction of the worldwide web, people began to process massive amounts of emails. However, how to convert structured, semi-structured, and non-structured data in the massive amounts of emails into text information and integrating it into the corporate IT system has become one of the issues that businesses face during digital transformation. Robotic Process Automation (RPA) discussed in this article is a tool that can accelerate enterprise digital transformation.
We will explain this using the order acceptance process of a customer (Figure 1) and whether it can be automated.
Figure 1: Order acceptance process
Before starting automation, the customer was advised to perform a “first round of sorting” of its routine operations in order to systematize the related procedures in line with the characteristics of RPA. Besides ensuring that RPA can accurately carry out the required process, this step aimed to optimize the process by sorting and standardizing the procedures in the process.
Then, the routine process was reviewed and broken down by principal task. The order acceptance process was divided into four major tasks: order extraction, data sorting, system login for data registration, and mail delivery. It was also confirmed whether or not there were special needs in this process that needed to be taken into account.
After confirming the operating procedures and standardizing the forms, standardized routine tasks and data were designed into procedures executable by RPA to smoothly automate the order acceptance process.
However, when designing the automation process, “order extraction” became an important task. Although RPA can be programmed to capture fixed data, the Document Understanding (DU) function provided by UiPath can be used to extract variable data in electronic orders in order to automate the complete process with RPA.
DU aims to help users digitize massive amounts of electronic documents. Based on the characteristics of traditional RPA, the accuracy of extracting electronic documents is enhanced with AI training and retraining. Figure 2 shows the entire process and following paragraph on process breakdown will describe how DU works.
Figure 2: Automation process of Document Understanding (DU)
First, documents are categorized using load taxonomy (Figure 3) before defining the format of the electronic data (e.g., setting PO Date as the date format) required for extraction for use in the subsequent data extraction.
Figure 3: Document categorization using load taxonomy
After defining the format of the data required for extraction, different methods (e.g., optical character recognition (OCR) software) are used digitize electronic files (e.g., PDF, TXT) (Figure 4).
Figure 4: Digitization
After digitization, the data is classified (Figure 5) according to the definitions made in the taxonomy to accurately extract the required data for the subsequent use for different purposes.
Figure 5: Classification
Lastly, the required data is extracted (Figure 6) and exported into the designated file format.
Figure 6: Extraction
In the extraction process, validation (Figure 7) can be performed at the same time for users to repeatedly revise the conditions of the data extracted by RPA. Then, the AI training and retraining function is used to enhance RPA’s ability to identify documents and columns.
Figure 7: Validation
Three major stages of enterprise digital transformation: “digitization, digital optimization, and digital transformation”.
Automation tools play an important role in accelerating all stages. UiPath’s simplified and flexible design process makes it easier to generalize process automation within enterprises. With the assistance of AI that can optimize RPA, it is not only a perfect tool for digitizing physical data, but also an ideal tool for enterprise digitization that enables the sharing, analysis, and reuse of optimized data for enterprises to gain favorable opportunities.