Share the load – the future of banking robotics

7 December 2016

Whether it’s improving staff morale or reducing costs and error rates, the increasing role that robotics can play in various elements of one’s operations can have a major impact on a company’s business processes and beyond. Patrick Kingsland hears from industry experts about its growing appeal, implementation challenges and future technologies.

Over the past few years, a number of new technologies have emerged with the potential to fundamentally change how and where we work. From robotic process automation (RPA) to social robotics and cognitive intelligence – not only do they have exciting, futuristic-sounding names, according to George Zarkadakis, digital lead at Willis Towers Watson, business leaders are now actively considering how they should apply them.

“There’s a tipping point,” says Zarkadakis. “What is the optimal moment when a business gets the most value out of automating tasks with minimal risks? The answer is different from company to company and from function to function but, depending on their goals and visions, organisations will need to change in the next few years in order to make the most out of the opportunities that come in this new world of artificial intelligence and RPA.”

Top choice

Of all the technologies available, RPA is one of the most popular, particularly in the area of shared services. When David Wright published a report on RPA in 2015, the title alone – ‘The robots are coming’ – alluded to just how promising things looked for technology on the road ahead. A year later and with another report out called ‘The robots are here – Meet your digital workforce’, things appear to have moved even quicker than that first title suggested.

“We have certainly seen a lot of interest over the past 12 months,” says the robotic process automation lead at Deloitte. “Last year, there were a lot more organisations that didn’t really know about robotics or didn’t particularly understand what it was all about. That has changed significantly. Our latest report shows that 87% are now familiar with the concept. People have clearly been educating themselves and also doing something about it. 23% are investigating and 22% are piloting or implementing RPA according to our survey. I guess when you look at other technologies and compare it, this has gone pretty fast.”

Rise of the machines

For those familiar with RPA – a technology that enables computer software to perform a variety of rule-based tasks in areas including finance, HR and procurement – this isn’t especially surprising. Although widespread adoption is yet to occur, its benefits have been apparent for a while.

“What RPA does is replace a great percentage of manual work that takes place in organisations,” Zarkadakis says. “In areas where people either transfer data from one system to another or have to process data and apply rules in order to transfer the outcome into another system, RPA can take away this manual work. It allows you to get some high cost-efficiencies in routine, labour-intensive processes and then use those savings for more valuable tasks to human workers.”

Accuracy is another clear benefit. Wright offers the example of a recent client that managed to reduce error rates on one particularly data-intensive process from 30 to almost 0% following the introduction of robots.

“That particular process involved pulling data from multiple PDFs and brought a really significant benefit,” he says. “It also reduced some of the overheads around controls. With the manual process, they had to have quite a significant verification of all the transactions. Once the robots were introduced, they quickly got confidence in their ability to deal with the transactions accurately. They even discovered that most of the time, when the team member investigated a potential mistake in what the robot did, the mistake was actually on their part. So, quickly that reduced to 50 or 25%, and actually it’s now just down to dip testing where the controls team just test occasional samples from the robot.”

Real jobs for real people

Of course, for as long as machines have been taking over human jobs, there has been resistance to industrialisation and automation. While that can often be the initial reaction, according to Zarkadakis, testimonials suggest RPA has been met with enthusiasm from businesses leaders and the wider workforce.

“Primarily, I think people feel relieved that the robots can take care of the most tedious, menial and boring work that they do,” Zarkadakis says. “It’s not about replacing the human; it’s about replacing the most boring part of their job, allowing them to be able to handle other things in a more creative and imaginative way.”

For a technology that can have such a major impact, one of the benefits of RPA is its relative ease of implementation. Introducing a robotic process can be done without disrupting a company’s underlying IT infrastructure in just two weeks once the rules and workflow are properly documented.

“It is a technology that you can move quite quickly with,” says Wright. “To move it to proof-of-concept in a test environment – we can help clients do that in under four weeks and without a significant amount of investment. It’s a technology that you can just trial or dip your toe into very quickly.”

Robots in disguise

For bigger companies though, Wright cautions against taking things too quickly. “You can create a very simple process in a day or two. For large enterprises, processes are complex, and there are a lot of exceptions, scenarios and systems to consider,” he says. “I think once you get into that level of complexity, you need to put the appropriate steps in place when designing the process and implementing it. At Deloitte, there is normally a process of discovery, of understanding what the process is like, how many variations there are, what systems are being used and what the quality of data is.

“You also have to consider the definition of what will be automated and this is, again, where some people can run into problems. If you take a purist approach, where you really want to automate 100% of the process, you’re likely to get yourself into trouble. The approach I typically suggest is looking at the core route through the process, the main scenario, maybe the main exceptions, and if that is 60% of the process then that is a real win. If we help a client to automate 60% of a process and that releases time to be spent elsewhere then that’s wonderful. But things can go wrong. We had one process where requests were just made by email in an unstructured format. The robot found that difficult to deal with so we had to go upstream and put a form in, which is good practice for any shared service anyway. It reduces the risk of error and increases standardisation when you structure the data. So sometimes there needs to be tweaks to the rest of the process.”

Of any size

Whether it’s a small or large company implementing RPA, for Zarkadakis one of the critical ingredients of a successful change programme is getting the right people.

“I think the key is to find the best people with the right skills and talents to do this implementation,” he says. “It is not a trivial task. The person who undertakes it will need to be able to put together rules because a robot is not a free-thinking agent, it is something that needs to be instructed in very precise terms. It’s also a job that requires communication across many stakeholders in order to be able to encapsulate what this manual work entails as accurately as possible.”

It may seem strange to ask about what comes next for a technology that is itself still fairly new, but when it comes to automation, Wright says the future looks bright. “From a vendor perspective, I think it’s going in two ways,” he says. “Some vendors want to build in cognitive artificial-intelligence technologies to their overall platform and others see themselves as the orchestrators, the process manager that would sit on top of other niche technologies. In my experience, RPA is quite a broad technology; it doesn’t really have a specific use case – it can do anything so long as it is rule-based, digital data. One exciting example I can see in the area of finance is the ability to actually write reports.

By combining natural language processing technology and data analytics technology, you can have a first cut of a report that says in plain English where you are against your budget or how your profit and loss is doing.

Great potential

While it may take time for some of these technologies to be adopted, according to Zarkadakis, demand is only likely to increase as business leaders recognise its potential to tackle future challenges.

“Without AI, our economies will suffer,” he says. “There are a number of threats right now in the future. One is the talent crisis that comes from demographic changes and exists right across the developed world. This crisis can be solved in many ways but automation can allow us to properly leverage human productivity.

Another challenge is low productivity; the reason incomes have stagnated across the developed world. Again, AI is a technology that can increase productivity and help us tackle a problem that has plagued us for
a decade.”