Covid Provides RPA a Chance to Shine, but Skeptics Remain

San Francisco-based Bank of the West, a regional subsidiary of BNP Paribas, recently found itself facing a significant increase in requests for payment deferments from its customers, as the economic impact of Covid-19 hit the US.

The wave of requests was more than the bank’s current workforce could handle. But rather than go through the process of recruiting and training new personnel, it hired digital workers—a new-ish buzzword for bots—to accelerate the processing of these urgent customer requests.

“The bots were developed in a matter of days, and they started handling the large influx of such transactions,” says Shafi Yusuf, head of transformation and digitization at Bank of the West, BNP Paribas USA.

Robotic process automation (RPA) has been around for decades, and is often confined to non-complex, repetitive tasks. During the coronavirus outbreak, RPA has proven more useful on the retail side of finance, but as is true with many technologies, once they’re battle hardened on the retail side of the market, they often seep into the wholesale side.

Bots have received mixed reviews, with some users describing it as a great enabler of efficiency that helps to achieve resource reductions, while others warn of its fragility and advise firms not to replace a core platform with robots. 

Senior bank executives have said that part of the problem is the word “robot,” which evokes associations with helpful androids, creating unrealistic expectations. The pandemic, though, could shift the perception of digital workers in a significant way. Some feel it has already, while others remain more skeptical. 

“I would say Covid-19 has changed the work culture in a matter of months, which would once have taken years to change, in terms of how things are done,” Yusuf says. 

He says once a robot is built and deployed, it only requires observation and monitoring. There is no need for human resource management, meaning it costs as little as a tenth of a human worker’s salary.

“Bots, once developed and deployed, work unattended almost 24 hours a day with just some monitoring for upkeep, in case other systems cause any outages,” Yusuf says. “They are usually a very low-maintenance digital workforce; they don’t need any coffee to go on and on producing much higher output.”

Santander German operation has been working with RPA since 2012. To improve the service, this work was centralized within the bank’s OPS Digilab consumer operations services department at the end of 2018. Department head Michael Methner says there will be an acceleration of RPA in the banking space due to the pandemic.

“At the beginning of [the Covid-19 pandemic], it was very important to be creative and invent processes in a new way,” he says. “We identified a lot of processes that could be more digital and more automated.”

In April, Santander Germany announced a partnership with Roboyo, a company that consults on and implements RPA. Via the partnership, the bank is in the early stages of automating changes in customer data using a customized front end, with a back end automated with RPA technology vendor Blue Prism. 

Methner says the project has improved service quality.

“The front end is doing all the checks, which normally need to be done manually in several screens and systems. After all the validations, the ‘OK’, is given and the RPA is running through the systems to change the data of the customer.” He adds the organization now handles more requests with higher quality and in less time.

Jon Theuerkauf, chief customer officer at RPA specialist Blue Prism, says his company’s banking industry clients have been able to better weather the impact of the pandemic because they have a platform of digital workers.

“The digital workers were the things that kept running for them in the midst of the human workforce being destabilized, having to be sent to homes, having to work full-time in remote locations,” he says.

One of Blue Prism’s clients is a large financial institution with large asset management and investment banking arms. It recently added 70 digital workers to help with cash management processes within the organizations. The firm, which already had over 1,000 digital workers, was able to build 70 more in three days.

“In a different time, they would have probably taken four or five weeks to build something like that,” Theuerkauf says.

RPA: Not a Cure-All

RPA has limitations, and banks need to be aware of where it should not be used during the pandemic, Bank of the West’s Yusuf says. One example is optical character recognition (OCR), a technology that converts images of typed or handwritten text into machine-readable text.

Yusuf says that in cases where there is complexity in variations in the documents, bots should not be used. His firm has trialed such technology, and Yusuf says the success rate has been much lower. 

“RPA and adjacent automation tools are very reliable to handle structured and semi-structured data,” he says. “However, if the information is unreliable and varies based on randomness in nature, then perhaps other tools like artificial intelligence and machine learning algorithms could be usable technologies.

Ed Wicks, head of trading at Legal & General Investment Management, says it is dangerous to use RPA for tasks with limited or zero-tolerance for errors. While RPA can produce clean results, it will only accomplish that for very specific tasks.

“Rules-based algorithms would be a safer choice,” for more complex needs. “Other tasks that are not supported by the availability of reliable data should also be avoided,” he says. 

Wicks says when the outcome of some repetitive tasks can be statistically optimized with the support of data, RPA will have greater benefits. “The automation in the selection of an execution channel or the macro-management of low-touch trades would benefit from greater use of RPA,” he says.

And there are hidden costs associated with RPA. Ulrik Modigh, head of product and operations at Nordea Asset Management, has previously said that his firm’s experiences with RPA delivered “limited success.” The Nordic heavyweight applied the technology to a number of use cases but reaped little return compared to the cost of implementing the pilots and fixing technical issues.

“I think the gains that we saw were too small, and there was also a lot of investment that we had to make to keep the process up and running,” Modigh said.

It’s a sentiment that has been said previously by industry technologists. Says one senior manager at a vendor company that specializes in AI, ”RPA is a glorified workflow tool. I use RPA because I was too lazy to set up the workflow myself. And it’s not even necessarily cheap, because I have to upgrade that robot every time. It’s not the answer.”

So the jury is still out, but it’s clear that if more people are going to work remotely in a post-Covid world, RPA is primed to take on greater importance. And as the retail side of finance works out the kinks, the institutional side will likely take greater notice.