Many enterprises see the automation of business processes as a valuable tool for productivity improvement and, therefore, an essential step for company development. However, such a solution brings a whole host of new questions. How do you find out where the resources are leaking? How do you address those leaks? Which activities should be prioritized? Who is involved? And, finally, how should you measure the results and evaluate your final goal?
One method is to build and implement a standalone RPA solution – and then never move beyond that since it concerns a single operational issue. However, when dealing with operational automation on a larger scale, building a Center of Excellence (CoE) should be considered. A CoE is a centralized entity that uses the skills of people from various disciplines and provides shared resources and facilities. CoE’s provide support and training, research, best practices, and leadership within a focus area.
But let’s start at the beginning. We invite you to a discussion related to automation journeys, held by two companies from the perspectives of an end-user and a service provider.
Guy Guldentops: former Head of CoE RPA at BNP Paribas Fortis, currently Head of Backoffice Analytics & AI
Guy Guldentops, the former Head of CoE for RPA at BNP Paribas Fortis, will show us how to map an ideal internal automation journey. Guy has worked on the optimization of banking processes for 20+ years. Over the course of only three years, he orchestrated the assembly of BNP Paribas Fortis CoE for RPA – starting from three proofs of concept (POCs).
More than 250 robots for 16 different stakeholders in the company were built and put into production. Thanks to this, a lot of repetitive work was automated, and employees could spend more time on more valuable tasks.
Iurii Shubin: Head of Intelligent Automation CoE at Intetics
We’ll also take a look at the other side of the story: software engineering. Iurii Shubin, Head of Intelligent Automation CoE at Intetics, will share his insights on how to reach operational excellence and revert losses into additional resources and incomes – all from a software engineer’s point of view.
The team of 30 technical specialists saves over 100K hours of manual work for SMEs, enhances productivity equivalent to 600K USD per year, integrates robot employees into the Client’s team, and scales its capacity as required.
Examples of Routine Tasks
Let’s make some concrete examples of routine tasks.
The banking sector is one of the most regulated industries out there. Millions of customers – and, thus, even more operations – drive this industry towards business excellence. Resource leaks appear in nearly every business department’s workflow. Manual repeatable operations include clicks, data processing and file management, emailing, and similar tasks. These tasks are crucial – without them, obstacles would arise that would postpone business development significantly, even for global leaders. However, the resources consumed by these activities could be considerably decreased and directed to much more valuable operations.
As with the financial sector, a myriad of other industries deal with operational challenges that drain the resources of back-offices and front-offices alike. Task automation can address these resource leaks – but there are several approaches to achieving this. We’ll take a look at these approaches in Part 2.
Let’s Start Talking About Technological Approaches: RPA vs. IA vs. Hyper-Automation
It’s important to split the whole scope of work into two types: Manual and Intelligent.
Manual: This part should be routine – it doesn’t require any decision-making and instead relies upon pre-set, simple rules. The manual part is an operational scope of work that is done by hand.
Intelligent: This part requires decision-making that is driven by human intelligence. Here, we can see all guideline exceptions from the manual part and all other high-level processes, which require analytical thinking (at the very least) to obtain successful results.
Now, what about Robotic Process Automation – what does this do? This technological approach delivers software that deals with a job’s manual scope of work. All remaining exclusions may be processed manually – or, perhaps, with the next level of the automation approach: Intelligent Automation functionality.
By combining these two approaches, you get a human-like intelligent IT system, which can handle business processes almost identically to a human. What’s the main difference? The intelligent IT system, unlike a human, is highly resistant to manual mistakes. Furthermore, the combination of RPA and AI technology can be strengthened even further with features like deep learning or machine learning – thus, refining hyper-automation mode for process optimization.
“What you do – you are replacing the repetitive parts of [business] processes done by human beings today with a software robot (and free up some time for your employees). But you don’t do the development of your own banking applications or banking platform. That’s why it’s important to have a clear understanding that the IT department and the business are on the same page, because otherwise, what I see in a lot of organizations, there’s a never-ending battle between IT and Business (while they need each other).” – Guy Guldentops
Regardless of the technical approach you choose and how you would prefer to implement it, it is crucial to make sure that the business entity and the IT department are on the same side of the automation battlefield. Both sectors must aim to build a business solution together – not just a software solution.
How to Determine Which Tasks Should Be Automated
Many businesses suffer from wasted resources – the need to optimize their business processes is obvious. But what does the first practical step of automation look like? When starting the automation journey, you will first want to implement a discovery phase. During this stage of the roadmap, you are exploring the feasibility of your idea, uncovering the deeper potential of RPA, and calculating your expected outcomes more accurately.
An interesting fact is that BNP Paribas Fortis’ automation journey began with three proof of concept (PoC) integrations, which were distributed among the organization to garner initial feedback. A PoC is a piece of evidence that demonstrates whether a proposal or design concept is feasible – this evidence is usually derived from a pilot project or experiment.
Almost immediately, it became apparent that RPA would have huge potential. With just three PoCs three years ago, BNP Paribas Fortis now has over 290 robots working on mainly operational tasks.
There are two main ways that you can gain the clearest vision for process automation. The first one is to employ the services of a business process engineer. This is the route that was adopted in the BNP Paribas Fortis CoE. The advantage of taking this path is that an engineer who possesses extensive business experience can analyze processes and define which parts could be modified – and, subsequently, where automation may fit into the process with the best ROI.
The second route is to use process mining tools. These tools easily capture information from an enterprise’s transaction system and give comprehensive, data-driven information on the performance of key processes.
“If you have process mining tools in place – better use them[…] Implementing such tools across the organization is also quite a heavy task.” – Guy Guldentops
Intetics’ RPA team used process mining tools – the advantage of these tools was that precise process analysis could be retrieved in a short timeframe without the need to involve additional business process engineers. This actually aligns with the main purpose of CoE for RPA: to increase savings instead of multiplying routine operations.
Further RPA use cases could be discovered step-by-step in a normal operation mode by regularly communicating with end-users and process designers and gathering ideas on how RPA might be adopted within the organization.
Where to Start – With the Automation of a Separate Processes or the Deployment of a CoE?
Automating separate processes individually may seem simple and promising at first. This approach may even deliver concrete benefits in a short period of time when considered from an SME’s perspective. But let’s take a look at this approach within a large enterprise. The variety and complexity of processes may be so huge that single process automation delivers a negligibly small positive impact – or it might make things even messier. This has proven that the CoE for RPA & AI building is the best way to start the automation journey for actively growing companies.
How to Get Things Started
At the beginning of Part 1, we gave a brief definition of what a CoE is. Now, we’ll take a closer look at the concept and explain possible models.
As a refresher, a CoE is a form of professional interaction between team members, with the main purpose being to increase general qualifications and competencies, finding and adopting innovative ideas, and reaching expected benefits and outcomes.
Your model of CoE depends on whether you will gather the business team and the IT department together in one place. If so – it is Centralized. If not, it is Distributed. The organizational form of your CoE influences:
- How clearly team members communicate with each other;
- How deeply they are involved in the solution design process;
- How quick the response time is; and
- When you can obtain the initial results.
“If you put a Center of Excellence in place, certainly at the beginning do it in a centralized way. Try to keep all your skills together and don’t distribute them throughout the whole organization. At the beginning, you don’t have a lot of people and competencies are scarce, and you need to put them all together. Secondly, and perhaps the most important of all, there is always a discussion whether you should put a CoE on the Business side or on the IT side. I was always in favor of putting it on the Business side and not on the IT side because, I think, it is also not an IT solution. It is a Business solution. But IT people should be part of!” – Guy Guldentops
Now, when we look at this matter from the perspective of the predictive software engineering company, Intetics, there are new arguments for the centralized model.
“Per our experience, we also recommend our clients to start with a Centralized Center of Excellence model because of a few reasons, like the resources are utilized in an optimal way, skills are utilized efficiently across the whole organization, and it also centralizes Standards and Delivery” – Iurii Shubin
The list of advantages of a centralized CoE is even broader than what Guy Guldentops and Shubin mentioned. When we consider a company that is allocated in one state, a centralized CoE allows them to reuse previously developed solutions and avoid the development of the same processes again and again in different ways. It also gives them the advantage of keeping clear coding guidelines and organizing work in an optimal way.
Obviously, a global organization has distributed offices and dedicated departments. It may seem challenging to keep the business team of end-users, the IT department that supports delivery, and a number of other team members in one place. However, even though it may be a challenge, building a centralized CoE for RPA is well worth it.
Adding Artificial Intelligence to RPA
Adding AI to RPA solutions is becoming more and more mainstream due to the potential for enhanced operational excellence. If the business process result needs to be retrieved from unstructured data, you could not get it from a pure RPA solution – this is where Artificial Intelligence provides a path to deal with the challenge.
Unstructured data may be found in any of your data sources – for example, in images, forums, and scanned documents. The most common purpose of AI functionality in the process automation scope is to retrieve unstructured data, structure it, and enable further processing.
Another typical AI task is to understand the context for further item classification, such as rooting emails to the departments in charge.
RPA platforms may easily be combined with various tools that have cognitive functions aimed at the recognition and analysis of images, texts, and other features. Thus, you may avoid the need to build two separate departments for RPA and AI tasks. This method uses “one operational flow,” which is commonly adopted in BNP Paribas Fortis’ CoE for RPA.
“Implementation of AI functionality within the same CoE allows your automation team to build an end-to-end automation solution, which is actually the absolute summit of the Business processes automation.” – Iurii Shubin
Now, having automated the process, you might think it’s the end, right? It’s not the end at all.
Look – what if the end-users show skepticism or even sabotage the use of the delivered RPA solution? This is a pitiable situation that is easier to prevent than to deal with upon occurrence. A Change Management system needs to be formed at the beginning of your RPA journey.
Change Management is the process that manages all changes to the environment of production operations from the very beginning to completion. A system flow looks like this:
- Identify the change (type, reason, scope, current and future states, concepts, and organizational readiness)
- Look at the details (changes of process, people, behavior, and information, plus cost and risk assessment)
- Determine the approach (conduct stakeholder analysis and determine roles of the change management team)
- Implement the solution (follow an action plan, communication plan, training plan, readiness review, escalation process)
- Monitor the results (via KPI reporting, sensing behaviors, and management review)
This fundamental step will not only make it easier for end-users to commission and put the robots into daily operation, but it will also strengthen and unite the team of everybody involved in the discussions, design, development, and testing from the get-go. If you don’t involve the entire team to the greatest possible degree from the beginning, there is more chance of pushback.
“The most important part to me is to put all the necessary people on board, from the beginning of the use-case till the end (including your internal customer). [If you don’t, they don’t feel like it’s their baby. They feel like they got something they didn’t ask for). If they are involved from the beginning, they can’t say that anymore, because they were part of the team. So, the resistance to changes is less at the end.” – Guy Guldentops
Another advantage to the deployment of a change management system is much better seen from the perspective of Intetics, which builds CoEs for its customers. Considering the sensitive data that is available to every team member in the banking domain, it may be problematic to involve the largest number of people from the get-go. The combination of both approaches will:
- Help decrease resistance and increase buy-in; and
- Onboard a new robotized team member, which will deliver far more than routine administration.
RPA solutions make people more satisfied with their jobs, help replace manual jobs with intellectual ones, deliver resource savings, and generate synergy for new achievements across the whole company.
Regardless of which stage you are at now (looking to deploy a CoE for RPA and AI in-house OR providing a SaaS to your customers), you may have noticed that even the final steps of founding a change management system must be made at the beginning. This is the same with all other points. Building the Center of Excellence is a complex and complicated piece of work, where each step should be thought through more than once – and then strictly followed in the set order to avoid any extra expenses and loss of time. We would be glad to know that you are ready to start applying these ideas to your company, having seen this process and its potential benefits from both sides of ownership.
Whether you decide to follow the best practices of CoE deployment in-house or request consultancy from an experienced team, we are confident that your journey to business operational excellence will bring you countless benefits.