While financial technology will never completely remove humans from financial transactions and planning, it can make those activities more efficient through a blend of artificial intelligence (AI), predictive analytics, machine learning and natural language understanding.
And do it one nudge at a time, Personetics’ marketing expert Eran Livneh said.
Mr. Livneh said Personetics’ founders David Sosna and David Govrin previously started a company that employed artificial intelligence to address fraud and compliance. After selling it they thought about what other problems they could solve with their knowledge of how to access and analyze transaction data in real time. Those thoughts turned to consumer benefits and grew into Personetics.
“What we have developed today is a cognitive brain for the bank,” Mr. Livneh said. “It looks at an individual’s transaction stream in real time.”
That cognitive brain is the result of a blend of artificial intelligence, predictive analytics, natural language understanding and machine learning.
The more data a user generates, the better the cognitive brain can predict behavioural trends and make relevant suggestions. Armed with a vocabulary of common terms and queries, one that grows as different requests are made, the system can recognize terms and provide accurate answers.
The platform’s analysis of transactions generates bits of relevant information that assists the user in that moment, he added. Perhaps that phone bill you’re paying is 30 per cent higher than usual. Personetics can alert you. If you are in danger of running short before payday you are warned so you can draw money from elsewhere. Conversely, if you run a surplus, you can be prompted to place funds in a savings account. That’s a much more comprehensive offering than most chatbots, Mr. Livneh said.
“If you are there and are waiting for the customer to ask questions, you will catch less than 10 per cent of a bank’s population in any given year,” Mr. Livneh explained.
By taking a more proactive approach with customers, banks are having a more regular interaction with their clientele, he said. RBC, for example, instituted a mobile automated savings program that looks at the user’s cash flow and moves money to a special account when possible.
“The nice thing about it is you set it and forget it,” Mr. Livneh said. “They let you know when they are moving money into savings from checking. For some people, the amounts are very significant.”
Personetics is applying nudge theory, Mr. Livneh added. Some people don’t have time to actively manage their finances while many are intimidated. Give them a subtle prompt at the right time that doesn’t take too much effort and they’ll opt in.
While the 80/20 rule of common questions to more rare ones applies, don’t look for AI models to be able to address 100 per cent of customer requests anytime soon, Mr. Livneh said.
“AI is part of an augmenting strategy where it works in tandem with humans,” he explained. “The machine can do much better insinuations where you are looking at transactions in bulk, but in a more complicated context where you want to know how to better manage your money it’s better to talk to a person.”
When it comes to deciding how to combine humans and AI in a call centre, the best way to do it is to identify at what point it makes sense to make the switch from a bot to a person, Mr. Livneh explained. Systems can identify topics, conversational tone, and even if the customer suddenly begins to type in caps.
An interesting topic in AI is how good can machines come at understanding context, which is one of the technology’s biggest challenges. An advantage companies like Personetics have over IBM, whose Watson aims to understand everything, is they are limiting the context in which their models operate to financial matters.
“What we are trying to do is say if the conversation is around a set of predefined topics around how you manage your bank account, for example, how can we get it to be 90 per cent like people,” Mr. Livneh said.
In late October Personetics applied its Cognitive Banking applications to America’s $1.4 billion-plus student loan debt. Using the nudge method and predictive analytics, Personetics’ technology can prompt borrowers to pay additional amounts when their cash flow allows. An additional $50 per week toward a $40,000 student loan at five per cent interest shaves 42 months and more than $4,000 of interest off the tab.
Learn more about Personetics:
Personetics Act – Automated savings, debt payoff, and other money management services
Personetics Engage – Personalized insights and guidance for the financial customer journey