“Robotic Process Automation (RPA) is a low-threshold option to acquaint an organization with digital transformation,” says Nikie van Duurling, Strategic Lead RPA at Deloitte Tax & Legal. “Instead of working with complex technology projects, here very simple tools are used to assist and improve existing processes.”
Mrs Van Duurling explains how tax professionals use technology to take monotonous labour-intensive work out of manual processes and outsource it to technology. When Deloitte digitized its internal expense reporting process, the organization came up against an unexpected challenge. With the introduction of SAP Concur, they went from paper to digital expense reports. This naturally resulted in a stack of digital — instead of physical — receipts. Digitization made the expense report filing process a lot easier, so the quantity increased. In addition, the tax professionals noted that the new mountain of receipts meant that the retention requirement of Inland Revenue and its conditions had to be met. This also increased the backlog in the expense reporting process. A solution had to be found quickly.
Inland Revenue requires process control
“You see organizations switching more and more to a system where you can simply take a photo of a receipt using an app,” Mrs van Duurling points out. But this digitization also means that receipts are sometimes submitted twice, albeit not with malicious intent, but because users make mistakes with the app or submit the receipt again accidentally at different times. “This has to be checked. For the Dutch Inland Revenue, you need a process where you can show that double receipts are checked.”
Setting up a control process
Such a control process had to be set up – a sterling task for RPA, the basic underlying idea of which is: take a repetitive process that a financial worker has to go through again and again, analyse it to an abstract level, and then build a software tool to take it over. Put simply, RPA is the automation of specific processes, i.e. labour-intensive tasks that really ever run only from A to B and nothing more complex than that. When it comes to a robot, we are in essence talking about software that performs a task based on the simple specifications: ‘do this’ and ‘don’t do that’.
More intelligent processes
Mrs van Duurling elaborates: “When you start dealing with ifs and buts, you run into more intelligent processes, where you can use cognitive intelligence to automate further, for instance. Robots do not like ‘sometimes’, ‘now and then’ or ‘perhaps’. They want ‘always’ or ‘never’. ‘Sometimes’ or ‘perhaps’ is an extra scenario to be considered and that will take time. It all boils down to keeping the question as simple as possible, especially in the early stages of RPA.”
Inland revenue agrees with the way of working
The technical story in turn boils down to the process of recognizing duplicate receipts, which are then checked manually by the back office: the expense reporting robot running at the back of the system takes the receipts from SAP Concur, compares images with each other to remove duplicates, performs additional checks using Optical Character Recognition as well as fuzzy matching with what is in the database, and in the end expunges receipts that may be duplicates. Whereas this would be an impossible task for a human, a robot can check individual receipts against the entire digital database of receipts that it reads every day. The system has been presented to Inland Revenue and the tax authorities have agreed to the RPA check.
Bringing tax expert knowledge to bear
Mrs Van Duurling focused on automation in the Tax department and on which solutions could bring disruptive innovation there. More broadly, in 2016 Deloitte examined blockchain, cognitive intelligence and RPA as disruptive technologies that could have an impact on finance. Meanwhile, serious gaming AI, augmented reality and other technologies are being looked at in greater depth. Mrs Van Duurling emerged as Strategic Lead RPA from the plan to experiment with such technology in practice. Deloitte started working with it first and developed an RPA vision only later. The crux of that vision can be summed up in one sentence: “We believe that it is easier to teach a tax expert how technology works than to teach a techie exactly how taxation works.” A developer for a financial process must understand that process through and through in order to see what the requirements are for RPA. That expert knowledge lies with the financial worker, not necessarily with the IT expert.
Implementing a business case for technology
“We wanted to train ourselves and transfer that knowledge, so that robotics is actually a first step towards the low technology threshold,” Mrs Van Duurling explained. “When it comes to technology, many people think: this is a black box where I do not know what happens. And to be sure tax professionals think in terms of exceptions, not of structures. So they would at times tend to look for exceptions that would prevent the system from working: ‘… but this would not work in a fiscal unity situation’. We started to take a far more transformational approach: how can we place a robot with customers, so that they can experience how such a technology project works in a nutshell? They can then learn to understand that you have to make a business case for such a process. In doing so, you assess whether it is worth the investment. You consider the purchase of licences. And you place these robots somewhere in cooperation with IT. You also take due account the maintenance for the software provided, and as a business owner you must clearly implement change management.”
From a simple tool to the entire toolbox
Mrs Van Duurling explained the development of RPA in recent years. “At the outset, RPA focused very much on outsourcing activities to the technical back-end (virtual machines).” That part was implemented mainly with the idea that companies could save money. In the meantime, RPA has become the starting step on the way to digital transformation. It introduces people to the practical aspects of technology, such as developing a business case. “You need to have a good business for such a back-end robot, one that shows how many hours a day you are going to save. As RPA matures, we are moving from a robot running its processes on a separate environment to the infrastructure around it.”
Stuffing robots with processes
This means that organizations will at some point have a platform to manage the software robots and the related licences. That will reduce the investment for rolling out additional robots – also because the processes in cooperation with IT will be clear by then. “As we have also started to run something like this ourselves and are stuffing the robots more with processes but are also splitting them off again, we can switch to assistant robots.” Such an assistant is a process for the employee. The assistant robot carries out a check on the expense reporting process, e.g. the parameters that determine whether the employee is entitled to a travel allowance; whether he or she is eligible for a leasing allowance; whether the manager is correctly linked, etc. “You start very small with a robot that can do a few things, but you can very easily add to it for the future because of its modular construction, so that it can become a full-fledged assistant for such a back-office employee.”
Taking the magic out of technology
She calls RPA a steppingstone at the beginning of the curve to intelligent automation. “You then move up. In other words, you start with RPA and then start adding bits of cognitive intelligence like Natural Language or even artificial intelligence later.” Particular emphasis was given at Deloitte to the business case for employees taking their first steps on the way to actual digital transformation. “We said: go and find a business case first and then you can help build that robot and become an RPA developer alongside your tax-related activities.”
A blueprint for RPA development
Deloitte uses the model that the Tax department developed for RPA for other parts of the organization also. Mrs Van Duurling: “It is a blueprint. We have done this at another Deloitte department, but also abroad. We went to other Deloitte offices to share the methodology, because you notice that they too want to work with and understand the technology better. It is sometimes held back as if it were magic, whereas you want to get that magic out — so that colleagues will actually dare to try it and acquire the knowledge themselves.
By Henk-Jan Buist Previously published in Executive Finance
Want to know more about RPA in finance? IMA organizes the webinar Supercharge your financial operations with RPA en AI on April 20