Part 2 of a Two-Part Series
By John San Filippo
At Finopotamus, we listen to our readers. Reader Mike Lantrip of SIU Credit Union asked us to write about robotic process automation (RPA), so that is exactly what we did. Specifically, we tracked down and interviewed these RPA experts:
John Best, CEO, Best Innovation Group (BIG)
Claudio Garcia, Director of Growth and Digital Transformation Consulting, Baker Tilly
Joseariel Gomez, Founder and CEO, Shastic
Lini Susan John, Head of Marketing, Zuci Systems
Phil Schmoyer, Director, Baker Tilly
Janaha Vivek, Senior Marketing Specialist, Zuci Systems
We present our findings here in this two-part series. Topics covered in Part 1 include:
The cultural aspect of RPA
Evaluating RPA Software
RPA vs APIs
Enter Artificial Intelligence and Machine Learning
It’s a common belief that RPA can only handle simple, repetitive tasks. Strictly speaking, that’s true when RPA is utilized in a standalone environment. However, when RPA is coupled with artificial intelligence (AI) and machine learning (ML), the resulting tool can handle much more complex tasks and situations.
“When we start to integrate the AI and ML tools into the automation, we're certainly into a much more productive type of automation,” said Garcia, adding that this sort of technology can address a wide range of “advanced” use cases.
“Let’s say an underwriter approves a loan with conditions, like we need the last two pay stubs or proof of insurance or something,” said Gomez. “Typically, it’ll sit there until some processor gets to it, reads it, and starts chasing the borrower for three days. We can use machine learning to monitor the current state of the loan and determine the next best action in this case. Or, we need to request information. What information do we need? Well, we need the last two pay stubs or the tax return. And then RPA can instantly request that from the borrower, bypassing the processor through a faster channel, like text messaging, for example.”
“We call ourselves an intelligent automation company because we use machine learning and artificial intelligence as a combination with RPA-based workflow automation to provide a much more advanced solution,” said Vivek.
“When some of our clients came to us, it's not like they wanted a machine learning system to be implemented,” added John. “They were totally unaware of this. They just presented the problem to us and we addressed it from a business perspective.” In short, it came down to using the right tools for the right job.
What If It Breaks?
RPA relies on conditions never changing. For example, if it looks for a field called “Loan Amount” and does something with that data, it will stop working if the field name is changed to simply “Amount” in a later release of the software. This is another area where AI can help.
“I believe RPA could be a big key to unlocking very intelligent machine learning services,” said Best. “Like if you had to change the script because a field went from submit to login, or from login to submit, it could fix itself and it could even find other things.”
According to Garcia, the issue is still manageable even without AI. “If a credit union is using one of these leading RPA platforms that we talked about,” he said, “these platforms are very, very good and very, very sophisticated in terms of being able to deal with patches and updates across the various systems that are being integrated.”
Replaced by a Robot?
Any new automation technology tends to make employees wonder whether they’ll eventually be replaced by technology. Finopotamus asked the experts how credit unions should address such concerns.
“The answer depends on what function we’re automating and what that employee can do with their increased bandwidth to repurpose that time to more value-added activities,” said Schmoyer. “I had a client that automated some of their loan intake processes. In doing so, they were worried that some of their loan officers would get discouraged or worry for their jobs. But what they found is that, as opposed to basically pushing paper, the loan officers were able to have more engaged conversations with their clients, able to do more cross selling, and able to identify other opportunities to strengthen the partnership with their customers.”
According to Best, credit union employees are in an especially good position. “What I've seen of credit unions is that they generally aren't looking at attrition. What they're looking to do is double their size without doubling their staff,” he said. “If you're a good employee, trust me, when they're done automating whatever it is, they will just move you along. You'll just have something else to do.”
Adding to this point, Gomez suggested that technology workers should be looking to what skillsets are in need now and will be in greater need in the future. “There are just new kinds of work that needs to get done,” he stated. “I would start moving into those fields, like obviously software. There are not enough software engineers in the world. There are even fewer who specialize in things like financial services. And even fewer who specialize in RPA for financial services.”
Measuring Success
Your credit union has reached a point where it has a substantial amount of time and money invested in RPA-based automation. At what point can you declare the initiative a success?
“Things we're looking for in any implementation are a reduction of the employee's time spent on a given task, the cost savings that come from greater efficiency, and the increase in accuracy,” said John. She said that customers typically report accuracy levels of 96% or greater.
According to Gomez, it depends on the client and the nature of the project. “We've been able to cut down about 30% of phone calls and emails through process improvement for one client,” he said. He added that each client will measure success differently, but agreed with John that it all comes down to time, cost and accuracy.
Garcia said that credit unions can’t overlook the cultural impact of a successful RPA implementation. “One success metric is cultural,” he noted. “By surveying and talking to their own employees as they get deeper and deeper with the technology, they're going to discover to what extent their staff is happy and excited and enthusiastic about that technology.”
Best cautioned that even with the success of the initial project, better automation is an ongoing pursuit that requires strategy. “Many credit unions are going in and buying an RPA tool because they have an immediate deed. They do the math and determine that RPA can fix it,” said Best. “Then they go, woo, we did it. And they start looking around for other things to do. That's not a strategy. A strategy includes the ability to measure your success against what you believe.”