Hlavní práce67196_alao00.pdf, 2.3 MB Stáhnout

(1)

University of Economics, Prague

Faculty of Informatics and Statistics

Robotic Process Automation tools comparison

MASTER’S THESIS

Study programme: Applied Informatics (ISM)

Field of study: Information Systems Management

Author: Orkhan Alakbarli Supervisor: Ota Novotný, Ph.D.

Prague, April 2019

(2)

2

Declaration

I hereby declare that I am the sole author of the thesis entitled “Robotic Process Automation tools comparison“. I duly marked out all quotations. The used literature and sources are stated in the attached list of references.

Prague 20/04/2019 ...

Orkhan Alakbarli

(3)

3

Acknowledgements

I would like to thank the advisor of my master thesis Ph.D. Ota Novotný for the effective methodology, pedagogical and expert help and other precious advice he gave me during the writing of my master thesis. Patient guidance, useful critiques, and assistance given by him were greatly appreciated.

(4)

4

Abstract

The primary goal of this work is to compare the leading platforms on the RPA market for the pre-selected business process. The research on this paper provides the description of what the RPA actually is, its pros and cons, as well as the situations when RPA is ideal to use. Further, the paper introduces the most known RPA tools available on the market. The theory is

supported by the practical part where the selected tools are compared to each other. In this part, the process of testing the tools against the pre-selected business process takes place. As a result, all the necessary data needed for comparison is gathered.

Keywords

Business Process Automation, Robotic Process Automation, RPA, pros and cons of RPA, when to apply RPA, ROI of RPA, RPA platforms, real cases of RPA implementation, RPA tools, UiPath, Automation Anywhere.

(5)

5

1. Introduction

Modern technologies are rapidly changing the activities of companies, providing yet unavailable opportunities in terms of the scale and speed of business growth. The key changes that have occurred in the last decade were associated with the introduction of modern IT systems into companies, primarily ERP, CRM, which initiated the transformation of many business processes in organizations through their partial automation and led to improved business efficiency. However, the main trend of recent years, along with the use of innovative financial instruments, is the optimization of personnel work, the key element of which is to connect the capabilities of robotic process automation and intelligent automation to the work of employees.

At a closer look, the term “Robotic Process Automation” (RPA) refers to specialized software that simulates human interaction with information systems in order to perform business processes.

Robotized systems are able to efficiently perform frequently repeated, algorithmized, and labor-intensive work. RPA is often mistakenly perceived as the usual automation with its high cost, or as just another business process management tool (BPM). In addition to the usual effects of traditional automation or BPM, such as speeding up processes, reducing errors and monitoring process performance, RPA provides a number of significant benefits when implemented in a company. First of all, this is a significant saving in personnel costs - for processes with frequent manual operations. Moreover, as RPA works in real-time,

customer satisfaction also increases. Also, the implementation of RPA improves the integrity of processes and data, improves monitoring and tracking capabilities - robots create the most complete logs of processes and monitor the quality of data.

Another important advantage is a reduction in the cost of implementing strategic programs due to the flexibility of RPA systems and the possibility of making rapid changes within them.

(7)

7

Finally, robotic automation provides the ability to quickly create, pilot, and bring to the market product and service innovations, which allows companies to have a serious competitive advantage.

The implementation of RPA-systems in organizations usually occurs in several stages:

 initial analysis of existing processes and assessment of the potential of robotization;

 definition of technical requirements;

 robot prototype creation and setup;

 robotic labor introduction and management.

The economic effect of the project for the implementation of RPA-systems is calculated as the difference between the current cost of the processes being automated, i.e. operational costs - labor costs, rental, and training costs, and the target cost of the process, including investment in the creation of the robot and its subsequent operating costs. According to industry experts, the average time to create a robot is 3 months, and the average payback period for projects on robotization of business processes is about 9 months.

The aggregate savings potential from the implementation of RPA in the business process averages 20–35%. The more detailed calculation of ROI of the RPA projects will be introduced in the later chapters.

The scope of robotic automation of processes is not limited, it proves its effectiveness in both front-office and back-office processes of companies. Here is just a small area of application of RPA: accounting, personnel accounting, budgeting and management accounting,

procurement, logistics.

As an example let us dwell on the possibilities of RPA in banking processes in more details.

The industry of finance today is increasingly intertwined with modern technologies; banks cease to be banks in the sense that we are accustomed to seeing them. Now they are more and more like IT companies with banking licenses. Competitive struggle wins the one who is ready to quickly deliver the service to the client, to simplify the service and communication of the client with the interface of the application of the bank as much as possible, to provide banking services 24 hours 7 days a week. What the bank client sees in the interface of the

(8)

8

mobile application is just the tip of the iceberg, the starting point of the process. On the bank side, virtually any operation leads to complex integrations and IT solutions; even the

elementary transfer from one account to another requires the interaction of several automated systems. RPA can expand the scope of automation in banking processes, for example, in processing applications for payments and transfers, processing transactions, reviewing loan applications, or generating consolidated financial statements. The main value of RPA for banks is that such automation works 24/7, does not change the current architecture of IT systems, and robots can easily be configured by business users.

What results can one expect from the Robotic Process Automation? To evaluate, let us list the most obvious differences between a robot and a human:

 Without a doubt, man is not a robot. Biological functions dictate the need for sleep, rest and lunch break. The employee is also socially guaranteed limited working hours,

holidays, and weekends. The software robot can perform its functions continuously - 24/7 in accordance with the operations it has been programmed for. And this is its undoubted advantage for ensuring efficiency and continuity of business processes in order to more quickly process requests and improve customer service.

 You must admit that people make mistakes on a regular basis. No one has ever taken the human factor off balance: fatigue, mental state, state of health have (and will) in one way or another influenced the quality of human work. And if you add to this multiple

production distracting factors, such as working in open space offices, phone calls, urgent meetings, email and social networks checks, the question “where do errors come from?”

will not be so rhetorical. Psychological research also further confirms the sad fact that the probability of error increases significantly when a person performs monotonous,

mechanical and uninteresting work. And here the software robot, which completely lacks these situational characteristics, has a chance to once again bring its value: it is absolutely indifferent to changes in mood or weather, time of the day or the need for rest after a busy week of work. The robot will not be distracted by checking numerous messengers or phone calls but will concentrate on doing its work, which, strictly speaking, he has been programmed. Thus, another advantage of software robots is the absence or minimization of errors that cannot be avoided when the human factor comes into play.

 When the executable process requires modification, it is enough for the robot to change the rules of operation (modify the script), but employees need to be retrained.

(9)

9

 Robots log 100% of their actions. This is especially important for compliance with various legislative and industrial requirements.

 Depending on the type of process being robotized, by performance one robot replaces from three to eight people.

 The robot needs neither a medical insurance nor a workplace in the office.

How does it look in real life? Does this mean that the employer takes the workplace from the employee and gives his job to a robot? This is an extreme case, although such a scenario is possible. In practice, there are two most common approaches to software robotization:

 Placing the robot on the employee's computer. In this case, the robot does not replace the employee by 100%, but only performs some tasks of the same type, acting as a digital assistant. As a rule, the tasks to be robotized arise from time to time and do not require the full-time allocation of an employee. The solution could be the installation of a robotic module on the employee’s computer in addition to the installed applications. When a task for a robot appears, the user simply starts a program that automatically performs the task.

It is necessary to understand that at this time the computer is occupied by a robot and the user cannot perform other tasks on it. But while the robot is doing its job, the employee can make a phone call to the client or discuss with the colleagues a joint project.

 Dedicated virtual workplace for the robot. In this scenario, a virtual environment is created in which only robots operate. This environment can contain hundreds of different robots that perform their tasks non-stop. The scheme is applicable for the mass use of software robots for a large number of tasks.

There are techniques for optimal selection of the most suitable RPA software. Most often, in addition to software vendors, companies have to get an advice from the consultants. When robots are deployed, consultants help companies transform using RPA solutions, software vendors carry out new functionality development and technical support. The key criteria for choosing such an automation system are functionality, the complexity of configuration, scalability, infrastructure, and ease of use by the end user. The topic of an optimal RPA platform selection will be also described with more details in the later chapters of this paper.

(10)

10

It should be noted that RPA is a new step in the automation of business processes, which is able to offer companies flexible and cheap solutions to optimize manual and time-consuming operations. Today RPA is a universal tool for increasing the efficiency of the company's internal processes. The main goal of modern companies is to connect robotized process automation, intelligent automation, and artificial intelligence to the work of employees as a reaction to rapidly changing market conditions. Today not the one who has a lot of resources wins, but the one who is able to act quickly and with maximum efficiency.

(11)

11

2. Literature Review

For the theoretical part of this paper, various sources of information available on the internet were used. The goal of this chapter is to list these sources and provide their general

description.

1. Robotic Process Automation for dummies.

This book is a well-structured guide into the world of RPA for people who had no previous experience of this technology before. The book was written by the experts in RPA from NICE. It defines what the Robotic Process Automation is, how it works and how can you use it for your organization, taking into consideration the different types of robotic solutions for specific automation needs. It also covers the best practices for inviting the right people and establishing a centre of excellence to get the maximum use of RPA introduction.

2. Robotic Process Automation and Risk Mitigation.

The concept of a “triple win” from RPA implementation is described in this book. It provides a detailed description of numerous case studies and the lessons learned, empirical evidence of business benefits and the risks associated with Robotic Process Automation including the framework for mitigating these risks.

3. Robotic Process Automation (RPA) Tutorial: What is, Tools & Example.

This article provides some basic but perfect for getting the first insights information about RPA. At first, it defines what the Robotic Process Automation is and what are the reasons for considering the implementation of RPA technology in the organizations, while at the same time providing the example of an RPA for the “Invoice Processing” business process. Then it compares RPA with the Test Automation analyzing what is common and different between them. Further, the article describes the RPA implementation methodology, best practices, general use and possibilities of RPA application in various industries. The topics like RPA tools, benefits and disadvantages of RPA, and myths associated with RPA technology also find their place in the article.

(12)

12 4. DMS Solutions blog.

The blog contains many interesting articles related to Robotic Process Automation that were used for writing this paper. The articles include the description of the common mistakes that occur during the RPA implementation process, technological traps, ROI from RPA projects, types of RPA and etc.

5. What is RPA? A revolution in business process automation.

The article identifies the positive and negative sides of RPA, while at the same time defining the tips for effective automation of business processes using this technology.

6. Robotic Process Automation.

The paper analyses the modern trends in the business processes automation. It also provides the economic effect estimates gained from the RPA implementation as well as the stages of the implementation itself.

7. The 12 providers that matter most and how they stack up.

The report uses 28-criteria to evaluate the robotic process automation providers, identify the most significant ones and score them. This material serves as a useful guide to the RPA market and helps to choose the right platform for business process automation.

8. UiPath.

UiPath is one of the leaders of the RPA market. Their website helps to understand the strong sides of their platform as well as tips on how to leverage the use of that instrument.

9. Automation Anywhere.

Automation Anywhere is another well-known brand that was applied for business process automation in the practical part of this thesis. Their website helps to get familiar with the platform and understand its specifics.

10. Gartner.

The website provides insights to various topics including the Robotic Process Automation.

The reviews posted on this website helped to compare two RPA tools mentioned above.

(13)

13

3. Robotic Process Automation

3.1. RPA vs. traditional approaches to increasing productivity

Modern innovative technologies radically change the business landscape. In such conditions, organizations that aim to preserve their competitive advantages on the market and accelerate growth inevitably have to quickly adapt and change the ways and methods of their activities, in accordance with which they have worked until now.

The traditional approaches to improving operational efficiency, currently used in business, have already reached their highest level of maturity and now provide less potential

benefits. Therefore, it is becoming more and more obvious that the business should look for new advanced approaches aimed at improving operational efficiency and reducing operating expenses. One of these breakthrough innovative approaches, which allow businesses to quickly reduce operating costs and significantly increase the productivity of operating activities, is Robotic Process Automation.

This unique technology allows you to automate repetitive, highly structured and rule-based processes and tasks by configuring a software robot to perform rule-based operations.

Software robots are able to take over most of the routine office work, doing it much faster, more accurate and more efficient than a human and thus freeing up human resources to accomplish more creative tasks — tasks with higher added value. At the same time, robots are easily integrated, work alongside already existing systems and applications, without changing the organization's IT landscape.

Why is robotization of business processes becoming a strategic approach to improving operational efficiency in most of the leading companies in the world?

 RPA is an innovative technology that allows companies to automate processes that have never before been considered as amenable to automation. In addition, such automation is possible in a very short time and with minimal investment costs.

 RPA is an experience-proven approach that is now widely used by many leading companies in order to create a full-featured digital back office.

(14)

14

 RPA is a breakthrough innovative approach that allows businesses to dramatically increase the efficiency of operations. (Boulton, 2019)

What results can be achieved by introducing traditional approaches to improving operational performance? There is no single answer to this question since everything will depend on the operating model and organizational features of each company. However, generalized figures reflecting the maximum effect that can be achieved with the help of various approaches to improving operational efficiency with the results reflected in the company's balance sheet are presented below.



20-30% of efficiency - using the approach of centralization and integrated business automation.



15-30% of efficiency - when using this approach to optimize business processes, such as process re-engineering (including Lean & Six Sigma).



40-60% of efficiency - using the approach of restructuring and business optimization.



30-50% efficiency - using the outsourcing and offshoring processes and functions.

(Lacity and Willcocks, 2017)

In my humble opinion, the one should not be afraid that RPA technology is just a temporary trend. The tendencies of its development indicate directly the opposite. In the near future, with the development and improvement of optical pattern recognition technologies, machine learning and artificial intelligence, the processing of unstructured data sets by software robots and, accordingly, the expansion of the spectrum of the tasks they process, will become better and better. This movement in the direction of intellectual robotization (Advanced Robotics) will strengthen the effectiveness of the robotization technology of business processes in the coming years.

I will write more about the future opportunities of the RPA technology in the further chapters.

(15)

15

3.2. The scope of RPA

In the introduction part, we figured out what Robotic Process Automation is. Now let us determine which processes are possible and should be robotized, which are possible, but not advisable, and which processes cannot be robotized. For a business process to succumb to software robotization, it must meet three main criteria:

Be definable. Meaning that the scope of the business process can be clearly defined and distinguished among the flow of tasks. For example, for accounting, you can select the process of payroll.

To be repeated. The business process should be executed on a regular basis (daily, weekly, monthly, quarterly, etc.) and activated according to a schedule or as a result of an event (for example, a letter arriving in a mailbox or appearing in a file folder). In the wage example, the process arises according to a schedule determined by law.

Be rule-based. The business process should be described by clear rules that exclude fuzzy logic. In our wage example, its accrual implies following clear accounting rules and does not require deciding whether the task is ethical from a moral point of view or whether it is worth or not worth paying the salary to an employee who is legally employed and worked according to the established work schedule.

The good news: any branch of the economy lends itself equally well to robotization. But some businesses are particularly prone to software robotization, as they use a large number of man-hours for tasks that can be assigned to software robots. The examples are:

 banks and financial organizations

 insurance companies

 medicine

Here are some classic examples of banking business processes recommended for automation through Robotic Process Automation:

 setting up customer information in banking systems,

 updating customer information,

 conducting financial monitoring,

(16)

16

 cleaning and reconnecting data,

 consolidating financial statements,

 registering encumbrances,

 processing incoming/outgoing payments,

 credit administration. (Guru99.com, 2019)

Now since we have described above the areas in which robotization is expedient, it is time to describe the features of processes which it makes no sense to start the implementation of Robotic Process Automation for. These are:

 Processes in which there is no manual sampling, processing, and re-entry of data. If in your IT environment all applications exchange data via an API or an integration bus without the participation of business users, you do not need to use robots to emulate users’ performance (and in general, you are very lucky with the IT department).

 Processes requiring automation entirely from start to finish. Robotization is not a replacement for automation, but complements it, speeding up the processing of individual manual steps and eliminating inevitable human errors.

 Processes in which decisions are made based on fuzzy criteria (beautiful/ugly, morally/immoral). At this stage, such processes will not work to robotize, although the forecast is such that in 5-10 years the robots will be able to solve these tasks using the tools of artificial intelligence and machine learning.

 Processes with limited duration. If you are for example being audited before M & A (Mergers & Acquisitions), then the data collected will be needed once, when making a decision to purchase a business. However, if M & A takes place, the data transfer when combining business systems can take a considerable period of time (up to 1-2 years), and robotization will help speed up this process significantly.

 Not very large-scale business processes whose processing rules change too often. In this case, the reprogramming of robots can be a separate costly process, negating the economic effect of the implementation of RPA. (Kaelble, 2018), (DMS Solutions Co., 2019), (IoT Agenda, 2019)

(17)

17

3.3. Myths and reality of RPA technology

The topic of Robotic Process Automation is today one of the most debated topics in the high- tech world along with Machine Learning, Optical Character Recognition, and Artificial Intelligence. Not so long ago, analysts predicted the rapid growth of the market for RPA- technology in the coming years, and we are witnessing how their predictions begin to come true. A lot of organizations around the world have already implemented a technology of robotization, and many of them managed to feel its significant advantages in improving the efficiency of business processes.

However, let us be frank, some RPA-projects have failed and did not meet expectations, and some organizations, in general, were quite restrained and suspicious of the prospects for its implementation. The answer to the question of why this was happening is partly due to the fact that the technology of robotization of business processes (as, indeed, every new

phenomenon) is fanned by many myths and biased interpretations. In this chapter, I will try to consider some of the most common of them and, if possible, reasonably prove their

inaccuracy.

Myth #1. The competencies required for an RPA-project pilot will be sufficient for further full-scale deployment of RPA technology in a production environment.

It is unlikely that anyone will be able to challenge the fact that a well-constructed approach at the initial, pilot stage of the RPA project (Proof of Concept, PoC) is one of the important guarantees of its success. The correct choice of a business process for robotization, positive expectations from investing in it (Return of Investment, RoI) are all integral parts of a successful project on robotization of business processes. However, it is also true that PoC is significantly limited by the scale of implementation, data, infrastructure, the number of possible scenarios and exceptions in the process, which require a special approach. As a rule, it is limited to one narrow process that has its own specific functions, and you can never know for sure how such a process will work in another environment. From this point of view, the pilot project, although it makes it possible to be convinced of achieving a positive RoI, but unfortunately, does not give a complete understanding of the future difficulties that can be encountered in the full-scale deployment of the process in a production environment.

(18)

18

For many key questions, both analytical and purely technical, it is simply impossible to give an exhaustive answer at the design stage of a pilot project. For example, which business process will show the best result in robotization and how to build the right RPA strategy and robotization roadmap? What will be the key tasks of the RPA Competence Centre and the main requirements for creating an effective and stable RPA infrastructure? What method of robotization is better to choose and what coding standards, principles, and methods of programming to use to create a software robot? How to ensure a 100% guarantee of the integrity and accuracy of the code and reconcile the difference in environments -

development, test, and production? How to build an effective infrastructure for the further large-scale deployment of robotization and what improved RPA results can be achieved with the help of machine learning technologies, artificial intelligence, etc.?

Myth #2. RPA technology does not require the involvement of IT professionals.

Since we are talking about the robotization of business processes, there may be a false idea that the technology of robotization can be introduced by business analysts without the involvement of IT specialists. However, if we take a look at the RPA technology and the difficulties that the development team may encounter, it will be quite obvious that knowledge and competence in business process management (BPM) alone will not be enough.

To solve complex tasks of robotization of business processes, you should thoroughly know and confidently use programming languages and coding principles. Especially when it comes to the application of cognitive technologies in RPA - elements of machine learning and artificial intelligence, which often necessitates the processing of unstructured data sets by software (the use of the mentioned technologies makes this processing possible).

In addition, business process management and robotization of business processes require fundamentally different skills from specialists. The mappings of processes in BPM and RPA also have a significantly different target orientation. If in BPM mapping the main thing is a correctly built sequence of actions in the process, then for RPA there are correctly

constructed algorithms. For RPA the trigger mechanism is especially important - launching a robot at a specific event or on a schedule (for example, checking mail every two minutes or monitoring the appearance of a document on ShareDrive), while for BPM it is important to focus on the description of input and output data. In BPM, the process goes through the

(19)

19

responsible executives, and in RPA, it goes through the systems, applications, and algorithms that the developers are developing to run the software robot. If in BPM, for steps in a process that require decision-making, unambiguous 'yes' or 'no' answers are sufficient, then RPA to obtain the same unambiguous answers should provide a number of rules and scenarios for which this decision will be made, and such scenarios may be a few dozen.

In addition, RPA developers often encounter such sometimes unpredictable technical

difficulties as, an improperly built development and testing infrastructure or an insufficiently powerful virtual machine that has a software robot and the corresponding systems and

applications with which the robot works, that as a result, it gives a low efficiency of the robot (processing of the process or data per unit of time). And this, in turn, jeopardizes the

effectiveness of the entire project for the robotization of the business process.

Such challenges, no doubt, can be managed only by a professional team of RPA-developers who have the relevant technical skills necessary for:

 building algorithms for software robots,

 building effective infrastructure, developing and implementing cognitive capabilities.

A few other fairly common myths about the robotization of business processes concern the applicability of the RPA technology and its future. They arise mainly due to the lack of familiarity with the technology itself and the latest trends in its development, which prevents many organizations from seeing RPA as a truly revolutionary tool for change.

Myth #3. Business robots have very narrow application.

It is quite common to have a biased view of the technology of robotization of business processes as a technology with a very limited range of applications, which cannot be

implemented in many business sectors. In fact, RPA technology is applicable in any industry where there are manual, repetitive, and executed on a specific schedule (or when a certain event occurs) processes. Thanks to robotization, their flow can be significantly improved by simultaneously reducing the cost of maintaining them. The technology of robotization is applicable in organizations of a wide range of professional orientation and in many sectors - banking and auditing, in the areas of health care and insurance, manufacturing and retail,

(20)

20

logistics and energy, real estate and construction, pharmaceuticals, the media, and many others.

Myth #4. RPA technology is too expensive and unprofitable investment.

Another common myth about RPA, which often becomes an obstacle to positive changes in many business structures, is that robotization is too expensive technology, so you should not invest in it. At the same time, such components as the cost of the software provider of the RPA platform, the cost of development and the cost of supporting software robots are taken into account in the total cost of the RPA project. However, it has been repeatedly proved from practical experience that the advantages of robotization quite quickly (within 9-12 months from the moment of implementation) justify the investment. This happens due to the fact that the software robot:

 is able to work 24/7;

 performs tasks many times faster than a person;

 can replace at least 3-8 employees who perform the same process manually;

 able to perform a large number of operations and tasks of different levels of complexity.

It is also worth considering that the process of reducing overall costs in an organization when implementing robotization will be directly influenced by such factors as:

 reduction of direct and indirect costs associated with the maintenance of staff (salary, taxes, bonuses, sick leave, workplace organization, insurance, training, etc.);

 avoiding the cost of modernizing IT and developing new functionality;

 reducing operational risks and improving overall operating performance.

Myth #5. Robotization of business processes is a temporary trend.

Despite the popularity of RPA technology in the world, some people are still inclined to look at the robotization of business processes as a temporary phenomenon to a certain extent.

However, if we take a look at the history of the development of technology and its current state, then without extra effort we will see that this is far from the case. At present, this is a new, but rather mature and established technology, which further actively evolves and develops, enhanced by the capabilities of other most modern technologies - optical character recognition, computer vision, human speech recognition (voice recognition), data analysis,

(21)

21

machine learning, and artificial intelligence. Therefore, there is no real reason to predict its near decline. On the contrary, RPA technology is rapidly developing throughout the progressive world.

Therefore, robotization of business processes, without any doubt, is today one of the most advanced technological solutions for business and one of the most effective tools for creating digital employees in organizations of any scale and professional orientation. And given the recent trends in the rapid development of RPA in the direction of enhancing its cognitive component, it will remain for a long time in the trend of the most sought-after technologies to optimize business processes, save time and resources and improve overall productivity.

(Guru99.com, 2019), (Grandviewresearch.com, 2019)

3.4. RPA technology opportunities and its moral aspects

Already today, platforms from leading RPA vendors (Automation Anywhere, Blue Prism, UiPath, WorkFusion), as well as open source-based solutions easily cope with the robotic repetitive, rule-based tasks and can drastically change the approach to increasing customer productivity. Software robots are increasingly becoming part of the working team and a reliable workforce to perform operational and administrative tasks. But what will follow the robotization of repeated tasks and where will the development of this technology lead us in the near future? Today, the key indicator of the development of this technology is the ability of software robots to perform actions in the same way as humans do. In other words, it is a combination of robotization of business processes with artificial intelligence and machine learning, which implies the followings:

The ability of robots to obtain information from unstructured data sources. If today robots easily process structured data (tables, database records, e-mail messages), or search for any data by patterns, then soon we will see how robots will use artificial intelligence to extract the necessary information from a complex unstructured data array. For example, a robot will be able to log talks and meetings, not only drawing up a detailed protocol of what has been said but also forming tasks and deadlines following the results of decisions made with assigning these tasks to the right employees, for example, through an electronic document management system.

(22)

22

The ability of robots to make decisions. Today, robots can already make simple decisions based on “if A - then B” algorithms. The advanced versions of this technology also already include elements of machine learning, in which, having encountered a situation not described in the algorithm, the robot offers the person to solve this problem once and later, if this situation is repeated, he is able to implement this solution himself, since it was added to set of available algorithms. With the development of machine learning, the robots of the near future will be able to independently make the right decision in an unfamiliar situation, based on the capabilities of the built-in artificial intelligence.

The ability of robots to conduct a constructive dialogue with humans. Siri and Google Assistant are used by many of us every day, and the use of chatbots in call centers has become a daily reality. However, with the development of artificial intelligence and machine learning, bots will be able to perform more and more complex tasks, providing constructive answers to both trivial and non-trivial questions from customers, contractors, and colleagues.

The ability of the robot to conduct such a dialogue with skill will significantly expand the area of application of the technology of robotization, which will also take place in the very near future.

Thus, in 5-10 years, we will witness business process scenarios in which robots will independently obtain source data, including, if necessary, directly communicating with the client, counterparty, colleague in a chat or voice. After receiving the data for processing, the robot will be able to independently decide on which process to start processing them, and if the data processing script is not known to the robot, it will be able to contact its human colleague for clarification and, after receiving the answer, remember how to further work out such scenarios.

Let us consider a very likely scenario for a client to contact an insurance company about an accident in 2020. So, the client gets into the small (damaged bumper), but an unpleasant accident. He calls the insurance company (if it is more convenient for him, he writes through the messenger) and informs the girl with a pleasant voice (which, in fact, is only the voice interface of the insurance company chatbot) about the incident. Based on the phone number, customer voice and simple test question, the chatbot authorizes the customer and continues to communicate with him already having all the information about him (customer history in CRM, terms of the current insurance contract, etc.). Following the chatbot's instructions, the

(23)

23

client immediately takes a few photos of the damage received in the accident and sends the photo to the robot along with the data of the second participant of the accident (vehicle registration number, insurance policy number and the name of the insurer). Based on the data received, the robot decides whether the further involvement of the insurance agent is

necessary for a detailed assessment of the damage, and if it is not necessary, it informs the client that his request has been sent for processing. Having passed a photo-based algorithm from a place of an accident based on artificial intelligence, the robot evaluates a list of necessary works and spare parts for carrying out car repairs. Also, the robot makes a request to the warehouse system of the auto dealer service center to control the availability of the necessary spare parts and materials, to specify the price, as well as the cost of work. By the way, the request from the service center is also accepted by the robot, since it is much simpler than providing the remote insurance partner with remote access to its warehouse system. So, the robot received the necessary data and it automatically forms the client's case. If the total amount of compensation exceeds the threshold value, the robot makes a request for payment approval to the relevant department of the insurance company and sends it for approval by launching the process in the electronic document management system of the insurance company. If the payment is less than the threshold value, then by the time the client arrives home, he will already have a notice in the e-mail about the insurance company's decision about compensation with the proposed options for the time of the visit to the auto dealer service center, where the spare parts and materials will be reserved for the client.

Looks like a fantasy? Nevertheless, all the individual technologies for the implementation of this scenario have already been developed and applied in business.

With prospects everything is more or less clear, but is it possible that robots will take most of our jobs away from us? According to some experts, until 2025, robots, both industrial and software, will perform work of 110 to 140 million full-time employees. The figure is more than impressive. So, robots take over the world? Yes and no. On the one hand, robotization provides a tangible economic effect, which no one can afford to ignore in today's

dynamically changing world. On the other hand, robotization brings social negativity.

Whatever is said about this technology that people freed from routine processes may be engaged in more creative tasks, or tasks that bring more added value, it must be admitted that many employees engaged in manual data processing are unlikely to become successful

(24)

24

managers or creative artists. Nevertheless, even with the invasion of robots, there will still be a demand for those who will set tasks for them, program their logic, make decisions that put robots at a dead end or communicate live with clients. Will there be people who will lose their jobs due to the fact that his workplace was occupied by a robot? Yes! Is it possible to reverse the process of robotization? Definitely not!

As history shows, the question is not whether an innovation will appear, the only question is how quickly it will take its place. Earlier in the production sector, the industrial robots have already driven people out of some processes. Now we are witnessing a similar stage in

business processes, and this movement is irreversible. In the future, we will most likely create labor codes regulating the use of robots, rather than prohibiting them. (Kaelble, 2018)

Robotic Process Automation in its effect can be compared with such phenomena as offshore programming, cloud computing, and outsourcing, with the appearance of which many jobs have gone to the so-called Third World countries. And those who realized this trend in time and managed to “ride the wave” both on the side of customers and on the side of service providers got their benefits. Those who remained in the old paradigm lost their effectiveness and were ousted from the market. It remains for us to admit that robots in some tasks are much more efficient than humans. So each of us should either develop competence inaccessible for robots, or move towards more creative roles. (DMS Solutions Co., 2019)

3.5. Typical mistakes in RPA projects

At the start of the project on robotization of business processes, as in any other projects in business automation, there are pitfalls that are not obvious to the customer, who encounters this technology for the first time. This chapter will be devoted to the most common mistakes that project teams and their sponsors make in the initial stages of RPA projects. The errors described below, of course, do not form an exhaustive list, as well as not all errors may be relevant for a specific customer.

(25)

25

Mistake #1. Robotic Process Automation capabilities overestimation

Do not consider the robotization of business processes as a solution to all the problems. This technology has a well-defined niche in which it is most effective. Where full-fledged IT automation is possible and economically justified, one should not try to use robots. Data exchange through buses and their processing by specialized IT systems will always be more efficient than the user's manual actions, even if performed by robots 10 times faster and without errors.

You should not always replace the employee with a robot from the beginning to the end of the business process. Such an approach can significantly increase the complexity of the robotization project and reduce the RoI (Return on Investment) indicators. The correct approach is to automate the most simple and routine parts of the process while preserving the participation of the person (employee) in non-trivial situations. In this case, the project will pass quickly, will be economically viable, and at the same time will give a tangible effect. At a time when the company reaches a certain level of maturity in the field of robotization, it will be much easier to transfer the remaining areas of work in the business process to robots, and the probability of a project “failure” will be minimal. (Forbes.com, 2019)

Mistake #2. Proof of Concept overestimation

Before making a decision about starting a project on the robotization of business processes, the customer usually tests the concept itself, that is, confirms the efficiency of the technology (Proof of Concept, PoC). At this stage, a situation often arises when the simplest process is given to the PoC. Such a choice may be due to many objective factors. Here are some of them:

 The complexity of organizing a full-fledged access of the contractor to the customer’s application interface, the preparation of a test environment, coordination with the security service, the organization of the contractor’s access to the site or remote access.

 Terms and costs. PoC by definition should not last for months and require significant investments.

However, the successful Proof of Concept may mislead the customer. It must be clearly understood that the robotization of a simple business process does not provide answers to all

(26)

26

the challenges that may arise during the large-scale deployment of technology. The correct approach could be a three-stage project:

 Proof of Concept on a small business process with an understanding of how this technology works in a particular environment.

 A pilot for robotization of one or two full-fledged business processes that affect all technological aspects of robotization. If the customer is going to robotize processes affecting SAP, applications with a WEB interface, MS Windows applications, OCR (optical character recognition) technologies - all this is desirable to test in the pilot in order not to stumble upon platform restrictions after buying a large number of licenses.

 Scaling a project to the entire pool of processes planned for robotization.

It should be noted that the presence of all three stages is not a requirement. If a customer needs to automate many processes of the same type, then a business process correctly selected for PoC can give him all the answers. There may be another situation where the performance of the technology itself is not in doubt (for example, in the banking and insurance sectors), and you can start the project from the stage of piloting.

Mistake #3. The incorrect composition of the project team

It is extremely important from the very beginning to competently form a working group of the project on robotization of business processes. If you look at the RPA technology superficially, it may seem that it is enough to use the business unit of the owner of the robotized process and IT to successfully complete the project. However, this is a blunder often made by many project teams. The correct approach should be to look at the robot, not as a program, but as a digital employee of the company. In this case, we will see what needs to be added to the business and IT:

 IT security. The robot will need accounts to work in the company's IT systems, which means that it needs to be assigned the access rights. What rights would it be: a single access model for all robots, or would each robot get its own set of accesses? The question of issuing a digital signature to the robot remains open. To sign within the company - this can be solved by internal order. For signing outside the company - this issue should be settled by industry or state legislation.

 Supervising structural divisions. Robotized business processes often affect areas that are subject to industry standards. And although the transfer of certain tasks from a person to a

(27)

27

robot only reduces the risks of nonconformity, the controlling structural divisions should be actively involved from the very beginning of the project to assess compliance with industry standards and issue appropriate recommendations.

 HR. The use of robots frees up the company's human resources, which means that it is necessary to involve the HR department, which should take into account which

employees will be released, plan their retraining (if necessary) and redistribute to tasks with higher added value.

Mistake #4. Robotization of business processes without optimization

Although robotization often simply speeds up user actions and increases productivity, it would be a mistake to transfer the business process to the robot (even at the PoC or pilot stage) without first revising it. The project team must understand that the GIGO rule (Garbage In, Garbage Out) is 100% applicable to the robotization of business processes.

Mistake #5. Underestimation of personnel qualification requirements

If the company decided to develop its own center of competence, it would be a big mistake to believe that you can quickly reorient specialists of a different profile to the implementation of RPA. The express online training will possibly train specialists to perform a small Proof of Concept. However, for large-scale implementation in-house, you need to train specialists at advanced training, then give them the opportunity to work in real projects on robotization of business processes for 3-6 months under the guidance of experienced specialists, and only after such training can such specialists be allowed to enter the industrial project on

robotization.

You also should not blindly trust the marketing materials of the developers of the Robotic Process Automation platforms, telling that to robotize a business process, it is enough just to perform the necessary actions on the process, and the RPA system recorder will record them and create a working script that will just have to be loaded into the robot. That does not work.

For full-fledged robotization, both analysts and developers are needed. And although

developers do not need to be fluent in any programming language, nevertheless, to robotize a more or less complex business process will require programming skills and an algorithmic mindset.

(28)

28

Also, an important issue is the existence of a unified Framework used in robotization. If robotization is carried out randomly, then the customer is guaranteed to face enormous problems when scaling a project, as well as when servicing already robotic processes.

In general, my recommendation for customers planning to build their RPA competence centre is to involve a qualified partner and use it not only in implementation but also in preparing the competence centre in the field of RPA, both technically and methodologically.

The above issues during the implementation of Robotic Process Automation, of course, are not an exhaustive list. Nevertheless, these are the most frequently encountered and obvious mistakes that can be easily avoided by competent project planning and the choice of a qualified partner. (Lacity and Willcocks, 2017)

3.6. Background vs UI Automation

Isn't it fascinating to see how a software robot emulates user actions, in particular, how it performs operations at the user interface level (UI) in the same way that a person does?

Undoubtedly. However, is that the most effective way to take advantage of software robots? I don’t think there is a definitive answer to this question, especially when it is addressed to developers who are used to using the background automation programming method in software development.

What do these methods of robotization of business processes mean in general terms:

 The background automation programming method - your software robot is able to process operations / obtain the necessary data using the application programming interface (API) of target applications;

 UI automation programming method - your software robot is able to perform operations, read and process data arrays of any target applications without actual programmatic access to them, and perform such operations at the user interface level.

(29)

29 Data processing at a program level

Advantages

 Effective use of software robot capabilities: operations at the software level will be much faster (5-10 times) compared to the same operations that would be performed by the robot at the user interface level;

 Lower chance of a software robot’s error/failure, since the robot will perform operations using the API, and in the case of minor changes to the user interface, this will not affect its performance.

Disadvantages

 A limited or even absence of visualization for business users, as the software robot performs operations at the software level and, accordingly, some operations/process steps are not visible to them. In addition, the user will need some time to understand the

reasons for stopping of a software robot and translate the programming logic into a human understandable language.

 Additional time and complexity: the RPA developer will certainly need more time to understand the logic of the business process (complex calculations and data verification principles) in order to correctly program the algorithms.

Key aspects

 If the software robot requires interaction with some applications embedded in your target applications, or if you need to transfer the results to a specific application template in which the robot worked, you should apply programming at the user interface level to simplify the transfer;

 The user-interface programming method is useful when you have difficulty accessing your target applications through an API or API not available as such;

 Some platforms on RPA enterprise level do not support the programming method on the background, so the UI method is the only possible one in this case.

Accordingly, if the question arises, what method to choose to create a software robot in the context of your RPA project, the choice will depend entirely on your processes and the characteristics of your environment, as well as on the skills of RPA developers. (Lacity and Willcocks, 2017)

(30)

30

3.7. Technological traps in RPA projects

There are quite a few strategic recommendations on how to structure programs for robotizing business processes in order to achieve advanced results. In particular, how to build the structure of RPA functioning, form and develop competencies in robotization, define and establish roles and obligations within RPA projects, strengthen the change management function when deploying projects on robotization of business processes, apply Lean and Agile approaches in RPA and etc. However, it is not often that you hear about technological pitfalls in RPA projects that any project team can get into and approaches that would help to navigate the RPA project in the right direction. That is why it is very important to identify the key technological issues that your development team may encounter during the RPA

deployment. The following list, of course, is far from complete, since many questions will appear in the process of work - depending on the specific environment in which you will work, as well as the technologies and platforms that you will use in the RPA project. The most common technological issues that should be paid attention to can be reduced to the following five positions:

Standards and principles of coding RPA. Regardless of which RPA platform you will use or how big your development team is, it is very important to initially agree on standards and coding principles to ensure the accuracy of the code of your software robot. Here I will not make a discovery in the programming world since this is a fairly standard approach that has been used for a long time in the world practice of process automation and engineering.

Considering the fact that the creation of a software robot is also part of automation,

respectively, all the basic principles that are used for classical automation should also be used for process robotization, namely, the principle of “Don't Repeat Yourself” (DRY), general naming conventions, naming variables, commenting, portability, etc.

RPA programming methods. There are two key methods for creating software robots: a user interface (UI) programming method and a program method (background). I have already described these two methods in the previous chapter. Before creating the structure and map of the software robot process, you should decide which programming method will be used for a particular process. The choice of the most appropriate method will depend on the RPA platform and the capabilities of the used third-party applications, on the processes and environmental features, as well as on the end-user’s vision of how the software will process data: at the software level or at the user interface level.

(31)

31

The continuity of the RPA process. What could be worse than losing all your code exactly when you need to transfer a robotic process from a test environment to production? Indeed, it is very annoying to realize that the generated code is lost due to the inefficient configuration of data backup. The RPA development environment must be configured for 100% availability and integrity of the source code data to ensure continuity of processes throughout the RPA project. Literally speaking, when your project team develops a software robot, all the source code that was placed in the repository should be stored in an additional data repository, which is reserved for certain periods of time (for example, every hour).

The RPA code control system. A version control system (VCS), in particular, SVN or Git, is the basis of any software development project, including the development of software robots. The absence of this system, properly integrated into your project and RPA

development environment, can lead to code non-integrity and, accordingly, poor quality of results, and also require additional efforts from the main RPA developer to bring together the intermediate results of the developers' work into a single whole. Therefore, before starting development work, it is imperative to ensure that a version control system is properly installed in your RPA control center when setting up your RPA development environment.

The principles of the RPA transition code. It will not be surprising if the client tells you at a certain stage that your environments - development, test, and production - are different. To solve this problem and to effectively manage environments, you should use configuration files, where all necessary differences will be preserved.

The listed technological moments are only the tip of the iceberg that any RPA developer may face. To be sure that your result of the robotization of business processes is manageable, performed on time and efficient, your project team must have people with the appropriate technical skills to ensure the full realization of all RPA capabilities. (Lacity and Willcocks, 2017)

3.8. Return on Investment in RPA projects

Robotic Process Automation is one of the most discussed and promising technologies in the field of business process automation today. Someone doubts the effectiveness of software robotization, and someone is ready, without any hesitation and calculations, to begin full-

(32)

32

scale deployment of this new technology on the market. And so it makes sense to approach the question carefully and try to assess the potentially achievable effect, which in the corporate world is designated by the term RoI (Return on Investment). And I would like to dedicate the current chapter to this topic.

Let us, for example, analyze in detail one of the most common mistakes - an attempt to compare one robot with one employee. It seems that all marketing materials are just talking about this: software robots are your digital employees. However, it is not entirely correct to consider this. Often, one can also hear from the heads of divisions where hired labor is used:

“It is cheaper for me to keep an employee at a salary close to the minimum than to spend money on the introduction of some robots”.

What are the costs incurred by the company for the office employee?

1. Direct costs

 Salary;

 Taxes on salaries, a single social contribution, etc.;

 Bonuses, rewards;

 Sick pay;

 Paid holiday.

2. Indirect costs

 Office rent, security;

 Communal payments;

 Workplace (furniture, telephony, computer, software, IT infrastructure).

3. Additional (optional) costs

 Social obligations of the employer (insurance, corporate events);

 Team buildings, training, certifications.

Thus, it is obvious that it will be incorrect to directly compare the employee’s salary and the cost of the robot.

(33)

33

But even if we were able to accurately calculate the cost per employee, can we simply multiply this amount by the number of employees released? The answer to this question directly depends on the business process. There is a situation where one or several employees perform only one business process in a company, and their employment is 100%, namely: 8 hours per day for 20 working days per month. There are other situations where one employee can be involved in 20% in one process, 40% in another, etc. In this case, we can say that robotization of one process releases in a company, for example, 0.4 FTE (Full Time

Equivalent - full-time employee). Robotization of another process releases 0.3 FTE, and etc.

In order for the economic effect of robotization to become tangible for the company, and not measured by the free time of employees, whose work was taken over by the robots, it is necessary to redistribute business robots tasks among employees. And only after that the company can get a fully released employee, which allows, for example, transferring him to another department, instead of hiring a new employee there.

Now I will try to identify the costs incurred by the company in the implementation and operation of robots.

1. Direct costs

 The cost of acquiring licenses. As a rule, software for robots is licensed for 1 year.

 The cost of service for the robotization of the business process, which includes all stages - from process analysis to putting robots into commercial operation.

This stage can be performed both by an external provider of robotization services and by the company.

 The cost of keeping robots up to date. Business processes in each company are constantly changing, and robots need to be adjusted to the changed conditions.

2. Indirect costs

 Robots also need a workplace, but a virtual one. As a rule, this is a virtual machine in the data center, on which the operating system and applications that the robot will interact with must be installed.

 Energy and maintenance costs of a virtual machine from the IT department.

(34)

34

When calculating RoI, it is important to understand that you will not dwell on the robotization of one or two processes, and robots will work in your company for years,

processing dozens, if not hundreds of business processes. Accordingly, the cost of the service of robotization needs to be correctly calculated for several years, during which the business process is guaranteed to be in demand. License costs also need to be attributed to several robotized processes, since the robot can work 24 hours a day, performs operations faster than a human, and, accordingly, will carry out several business processes in turn. It should also be noted that in 99% of cases it is incorrect to assume that one robot = one process.

At the initial stage, the RoI indicator is strongly influenced by the vendor licensing policy of robots. So, some vendors sell robots in packages of 10 pcs. This is quite a normal package for a large corporation, but a mid-level company can start its project with a much smaller number of robots (2–3 licenses) and increase their number gradually as business robotization

algorithms become available. In this case, there is no point in paying for robots that are idle while waiting for algorithms to be prepared for them. Accordingly, it is important to correctly calculate the number and volume of processes planned for robotization for the period of time for which RoI is calculated. It is also necessary to remember that there are also open source solutions that allow you to create robots without investing in licenses at all.

Factors indirectly influencing RoI also include:

 Increase the speed of processing processes. If, for example, an insurance company can, with the help of robots, reduce the processing time for requests for refunds from weeks to days, then it will gain a concrete and tangible advantage over competitors on the market.

 Reducing the number of errors. As a rule, employers try to avoid overtime for

employees, as analysis shows that at this time employees make many mistakes, which ultimately leads to operational risks and additional costs of correcting them.

 Robotization can be cheaper and faster than work on full-fledged IT integration. It is always appropriate to compare what will be simpler, cheaper and faster: to fully automate the business process, for example, by introducing new functionality into existing systems, or enter robots into the process as performers.

Thus, it always makes sense to carefully calculate your RoI from project implementation.

And the word “your” here is the key. Do not rely on the methods offered by vendors. They

(35)

35

are always universal, which means they do not take into account the specifics of the customer company and are designed so as not to notice the factors that lower the client's RoI and only show positive factors. In addition, with more or less equal prices for a robot in different markets, the cost of a man-hour can vary significantly. But it is very difficult to calculate RoI on your own for unfamiliar technology. It is necessary to understand not only the general principles of technology but also the subtle points, the awareness of which comes only with the experience of implementing the first projects.

Consider robotization as a means of optimizing costs. Attract a qualified partner, consider together the effect, choose the RPA platform most suitable for your environment, and with a high probability, you will achieve your goals to optimize your costs and improve overall operating performance. (Lavrov and Petyuk, 2017) and (DMS Solutions Co., 2019)

3.9. RPA outsourcing

Probably, not one organization that made a decision to re-engineer its business processes by introducing the technology of robotization of business processes faced a dilemma: to build its own competence center for robotization of business processes or, perhaps, to transfer the function of development, change management and software support to RPA specialized organizations? At first glance, it may seem that it would be more economically advantageous to have the organization automate business processes on its own, relying on its own human resources - business analysts and IT specialists of the company, without the involvement of third parties.

However, is this option rational in the long run? Answering this question, the following components should be considered, in which organizations need to invest in order to successfully deploy projects for robotization of business processes:

Specialists. First of all, companies will need to invest in the search, hiring, and maintenance of relevant professionals. Experience shows that it will take at least six months to prepare a qualified RPA engineer, and a mature RPA analyst will need more than 2 years. Taking into account the fact that the technology is new for the market, it is rather difficult to find

specialists of the corresponding profile on the labor market, and their maintenance will cost organizations much more than other specialists from related fields.