Custody banks are generally not the first to adopt new technologies — as highly regulated entities with oversight of large amounts of money, they take few risks with the systems they use to safeguard money for institutional clients.
But this year Bank of New York Mellon became one of the first U.S. banks to use robotic process automation — not just in pilots but in everyday operations.
"A bunch of our folks knew about robotics, we heard about it, [and] we said let's give it a try," said Doug Shulman, senior executive vice president and global head of client service delivery at BNY Mellon. "We are leaning heavily into let's automate as much as we can."
In robotic process automation, software "bots" with a rudimentary level of artificial intelligence are configured to interact with applications and perform high-volume, repetitive tasks.
The deployment of bots is part of a broad transformation program at the bank, through which it hopes to become a technology leader. "If you think about smart automation, robotics is a piece, workflow is a piece, and we're combining smart forms, optical character recognition, workflow and robotics to get momentum around automating tasks," Shulman said.
BNY Mellon has a robotic process automation team that works with the bank's business leaders and the transformation team to find places to use robotics. So far, they have come up with eight pilots and moved four of them into production.
One is in trade settlement.
"One of the services we provide as a custodian is to settle trades, make sure they're allocated appropriately," Shulman explained. "Most trades settle automatically and go straight through the client's system to our system. But sometimes trades don't settle automatically — there's an anomaly in the market or [an order] comes in late or however it happens. In the past, we'd have a queue built up of nonsettled trades, and we would have somebody go into the system, compare our database with an external database and review that and clear it."
Using robotic process automation software from Blue Prism, the bank has programmed bots with rules that let them perform research on the orders, resolve discrepancies and clear the trades.
It takes a human five to 10 minutes to reconcile a failed trade. A bot can do it in a quarter of a second.
"It's freeing up people who used to do this," Shulman said. "We're taking people off the mind-numbing tasks that a bot could do, and freeing them up to service clients and do higher-value-added tasks and more analytical tasks."
BNY Mellon is also using bots in its data-reconciliation group. For certain clients, such as pension funds, the bank has to regularly go to the client's website, pull certain records, and compare them against the bank's own records, to make sure all the records match.
Where people used to do that task, now bots do the basic work and humans only handle the exceptions.
In addition to being faster, bots can work at night, saving employees from having to work night shifts. They can handle overflow when a stream of work outpaces the available staffing.
The same control teams in the bank that check human work on a regular basis for accuracy, quality and timeliness keep an eye on the bots' work.
"We pride ourselves on being an incredibly controlled institution," Shulman said. "Anytime we deploy a new technology, we make sure there's a proper risk frame, control frame, infrastructure frame."
The bank plans to use bots more widely. "We're still in early days," Shulman acknowledged.
Along with robotics processing automation, BNY Mellon is also experimenting with machine learning, a more advanced artificial intelligence technology.
"There is the promise to marry up machine learning with robotics, so it's learning more and more as it does things," Shulman said. "As I look into the future of this industry, I think companies that figure out how to meld these technologies together to drive better automation, workflow and integration with clients will move ahead. That's our vision."
Shulman said the bank has already learned a few things from its tests. One is, if processes are designed poorly or are inefficient, the bots will not work well.
"The important thing is you need to understand your processes; you need to have them mapped," he said. "You don't want to automate processes that aren't engineered correctly."
In some cases it is not obvious which rules the bots should follow. Those rules will have to be written before robotic process automation can be deployed, Shulman said.
Artificial intelligence has been around for about 60 years. So what's new and why are banks getting into this now?
According to Marco Bressan, chief data scientist at BBVA, which has also begun using bots to help employees do their jobs, the reason interest in AI and bots is gaining momentum is that the technologies have improved. Perhaps more important, the infrastructure pieces they need to work — including high-capacity data management, data storage, algorithms and basic computing resources — have all recently gotten better, cheaper and faster, making some forms of bots and AI practical that were once only theoretical.
Computer hardware can now easily store the vast amounts of data needed to teach computers to think like humans. "We have only had that recently," Bressan said. Similarly, in-memory processing allows for the processing of large volumes of data in real time. And cloud computing also makes storing reams of data more affordable.
"My kid who is 15 years old can go into a site and train a deep learning neural network, just for fun, to see how Van Gogh would paint a photo of his friends in school," Bressan said.
David Weiss, senior analyst at Aite Group, sees a few forces converging to make bots and AI popular.
One is the fintech startup culture. "We've got a lot of startups that are going nuts with this," Weiss said.
Another is big banks' need to cut costs. "There are cost-constrained firms looking to do things cheaper and easier," Weiss said. "It used to be you'd do labor arbitrage, and use workers in India or the Philippines. A lot of companies are trying to figure out what to do."
A third is a willingness to try new technology. "They got turned on to the bots very organically, and said OK, we'll try it," Weiss said. "There's a lot of that now at Tier 1 firms, and there's lots of firms that have not even put their foot in the water."
Editor at Large Penny Crosman welcomes feedback at firstname.lastname@example.org.