The growth of machine learning can help financial services companies automate extremely outdated technology and make them more competitive with digitally savvy upstarts, Hyperscience COO Charlie Newark-French said.
Using proprietary artificial intelligence and machine learning platforms, Hyperscience’s automation software sorts through different document types to deliver actionable data which reduces (but does not eliminate) manual work and improves the customer experience. It can increase processing speeds by a factor of 10 and improve accuracy by 67 per cent.
Many financial services providers are using tech stacks that are a quarter century old. Plenty of improvements have happened over that time, and those companies know these cobbled together systems are inflexible and easily broken. An entire sector of business process miners earn a living telling companies what their back office processes actually are. Yet front office systems are much more digitally advanced than their back office counterparts and it will take some effort to catch up.
“It’s kind of crazy a business doesn’t own its own processes and someone has to come in and tell them their processes,” Mr. Newark-French said.
Machine learning is how the divide will be erased.
“Our aim as a company is to use automation, to use machine learning, to speed up those processes,” he added.
Customers care about speed, reducing costs and accuracy, Mr. Newark-French said. Machine learning can improve all three. Financial services firms commonly use manual laborers to manually enter and check data. If the numbers match from two people it’s fine. If they don’t it goes to a third person. After a few hours in this repetitive process human error rates soar. Contrast that with machines, who don’t get tired or bored.
Embedded within those three consumer cares is the desire for transparency, Mr. Newark-French observed. Let them know the status of their application or, for example, immediately inform them when documentation is needed instead of delaying notification and you improve satisfaction.
“Those sorts of things allow you to speed up and have greater interaction with the customer as well,” he said.
Hyperscience improves the integration of data by efficiently processing it however it is received, whether that be via PDF, mobile device or even fax. It turns human readable data into machine readable data and allows a company to begin their automation process from their present state and progress to a goal without having to adjust technologies to even begin.
“We want to make that change possible and go along on the journey with them,” Mr. Newark-French said.
It’s a more nimble approach than robotic process automation (RPA), which can be more inflexible, Mr. Newark-French explained. If you operate an RPA-based system, and you wish to change the method for making a single decision you have to reprogram the entire process. Contrast that with Hyperscience’s approach where software passes a task to a human when the machine cannot address it. The human completes it and the system learns from that and begins its automation process.
“You have the benefit of automation and the benefit of change, and that’s the future for large enterprises in our view,” Mr. Newark-French said.
The financial services industry subscribes to the upside down Internet, Mr. Newark-French said. While in most cases people manually produce content which is spread electronically via the Internet, in financial services the opposite is true. Machines produce complex documents that people still have to read and interpret.
Luckily, computing power has progressed to the point where human readable languages can drive software development, Mr. Newark-French said. It can read previously unreadable data and better understand its context. That is important in an industry where those documents are extremely complex, where a mortgage can have 800 pages of supporting paperwork.
“It takes context to understand what those numbers mean and our approach to machine learning allows us to understand what those numbers mean,” Mr. Newark-French said.
The nature of data security is also changing, Mr. Newark-French said. Data anonymization will play a key role. Combine that with a greater number of companies wanting on-premise solutions and it’s an interesting mix. Much progress is being made in this space, which will allow companies to only send out necessary data while redacting the rest.
That can be a challenge even with big financial services companies, which can be a mix of 15 smaller companies or separate divisions who rarely speak with each other. When Hyperscience works with two or more divisions within a company, they can help begin that seamless data sharing process that is both good for them and the customers.
While a decade ago the prediction was all data would someday be stored on the cloud, that is no longer the future, Mr. Newark-French said. It’s easier to operate private clouds and control where data is being processed. Like many things the answer is somewhere in the middle.
“I think the future at this point is very clearly some things are in true, shared-cloud, multi-tenant systems but the more complex stuff and the more PII, compliant regulatory data issues, people want them somewhere they control,” Mr. Newark-French said.
