Businesses have numerous tasks with no added-value. They represent a significant cost and can generate more or less shocking errors. Hence the tendency to use “robots”. Robotic process automation (RPA) is the term used to describe software tools that partially or fully automate human activities.
On paper, the RPA principle is simple and quite appealing. An RPA software « automates repetitive processes based on rules, usually executed by people sitting at their desks, » explains David Schatsky, general manager of Deloitte.
Robotic process automation is a promising technology. It allows you to configure software – or a « robot » – to emulate and integrate the actions taken by your employees. The best-known application is the chatbot. More and more and businesses use it or are planning to do so soon. A Gartner study estimated that these virtual consultants would power 85% of all customer service interactions by 2020.
But by interacting with these applications like a person would, these « robots » can accomplish even more tasks: open attachments, complete electronic forms, save and re-enter data, extract structured and semi-structured data from documents …
Plus, they never sleep, never make mistakes and cost a lot less than employees! If we study these « bots » more closely we can identify four advantages :
Automation software robots are programmed to follow specific rules and tasks in relation to Optical Character Recognition (OCR) software or Application Programming Interface (API). They never get tired and never make mistakes. They are complacent and coherent.
Once « educated » these robots run reliably. Everything they do is supervised to ensure all existing standards are met. You can « configure » them so that they are compliant with the rules. For example, collecting only certain personal data in order to follow the General Data Protection Regulation (GDPR)
RPA can reduce treatment costs by up to 80%. In less than 12 months, most businesses can already see a positive return on their investment and the potential for additional cumulative cost reductions can reach up to 20% over time. In Japan, Fukoku Mutual Life Insurance, one of the leading insurance companies, reports an increase by about 30% while achieving significant savings (approximately $1.3 million dollars per year) just by using IBM Watson Explorer AI.
Employees are the first to appreciate the advantages of RPA because it offloads activities that are time-consuming and that offer no added value. They can then concentrate their efforts on projects or tasks related to other skills. According to IBM, up to 80% of common and simple questions can be handled by such a program.
But be careful not to be too optimistic as there are limitations. Such a program can not do everything; bots require constant management and maintenance throughout their lifetime! Since RPA solutions rely on screenshots to visualize the automation process, you are quickly faced with a number of challenges. In particular, it is not always easy to create an optimal sequence of actions involving multiple.
AI to the rescue?
To meet this challenge, RPA providers integrate a layer of artificial intelligence and more specifically “cognitive” capabilities. This integration has led to the development of Cognitive Robotic Process Automation (CRPA) software robots that integrate different cognitive abilities, including natural language processing, machine learning, and speech recognition.
Today there exists various RPA/CRPA solutions (Automation Anywhere, Blue Prism, Nice Systems, Work Fusion, UiPath, Kryon Systems, Softomotive, Ipsoft, etc.) that claim to be able to automate perception and judgment tasks that have been, up till now, reserved for humans.
Automating regular operation
Financial service companies are interested in automating their regular operations. For example, a Swedish bank purchased software from CPRA (Ipsoft) to improve its customer service. The program, called Amelia, can speak 20 different languages and understands the semantics of a language. If it can not resolve the problem, it transmits the request to a human operator and observes the interaction to improve its knowledge in order to deal with similar cases that may arise.
Another application: insurance. Collecting, validating and updating data can be time-consuming for businesses. Thanks to RPA, fraud audits, policy renewals, insurance premium calculations, and data collection can be automated. In France, AXA uses this type of solution to improve its subscription process.
Although promising, the CRPA is still in its early stages. The union of RPA and AI is very recent. At this moment, the RPA offers better results when the application interfaces are static, processes are not changing and the data format remains stable. However, this combination is becoming increasingly rare in today’s dynamic environment.
In conclusion, the integration of this technology is promising. But in order for its integration to be successful, it must rely on two key elements to support and allow evolution beyond the RPA. First, it is essential to define a strong and precise API strategy. Secondly, There is a need to develop a proven user-centered design process, in particular, modularization and customization.