Machine Learning Is Vital to this App-Based BankBy Samuel Greengard | Posted 2016-06-09 Email Print
Atom Bank, an app-based financial institution in the United Kingdom, uses machine learning to redefine the way customers and the bank interact and transact.
Hardly a day goes by without word of advancements in artificial intelligence (AI) and machine learning. The technology is rippling through a growing array of industries and transforming the way people live and work.
At Atom Bank, an upstart financial services firm in the United Kingdom (UK), machine learning is the fundamental underpinning for the business. "We are an app-based bank," says Chief Operating Officer Stewart Bromley. "There are no branches or massive call centers. There's also no Web functionality."
The app-only approach is revolutionary, and, if successful, it could prove highly disruptive to the financial industry and beyond.
"It's our belief that people in the UK who want to bank digitally are underserved," Bromley explains. "There is no other firm that allows them to do all their banking entirely within an app.
"In many cases, they must spend time and money traveling to a branch and pay for things they don't actually use. The idea for Atom Bank is to deliver an outstanding digital experience, but also provide greater value."
To achieve those goals, the company turned to WDS, a Xerox Company, to fundamentally redesign and rewire tasks and connect to customers in an automated but personal way. Atom's mobile app focuses on advanced self-service functionality, including the ability to manage tasks and answer questions directly from the app.
Moreover, as the technology fields questions, examines behavior and deals with issues, it will progressively learn the most efficient way to deliver information and solutions. The WDS Virtual Agent relies on analytics to capture the context of each customer inquiry and deliver relevant responses and solutions.
APIs Drive the Mobile App System
The system, which is driven through a series of APIs, automatically recognizes the mobile device a customer is using and contextually adjusts answers and information to match the specific phone and its features. Along the way, it studies how customers use the phone, and it continually adapts—including its information and responses—to what works best.
"We chose not to use a Website for service delivery because of security considerations," Bromley says. "Phones offer a much higher level of security. Because customers use the app, there is no phishing, and we are able to authenticate with biometric credentials."
The company's business model also eliminates a large call center, which is standard among banks. "We operate a contact center, but it is really more of a glorified help desk," Bromley explains. "We aim to attract customers that desire a self-service system, so a human is there in case the system can't resolve the question or issue. But it's more of a last resort."
The support desk operates 24x7. The WDS Agent IQ software can determine which agents are achieving optimal results and then feed that learning about best practices to the rest of the team. This will help ensure that customer queries are handled consistently and in the best way possible.
Atom Bank has tested the system internally since October 2015, and it is now in the process of rolling out the app commercially. The firm relied on a DevOps approach to build the functionality, and it plans to take an agile approach to updates and releases.
"We will continue to build and advance features and functionality," Bromley reports. "We believe this approach represents the future of banking."