Westpac is taking a backseat to adopting artificial intelligence technologies in its environment while it works on rebuilding its IT foundations.
The bank is in the midst of a handful of significant transformation initiatives that are intended to bring the organisation up to speed with the demands of banking customers.
Arguably the largest in this series of overhauls is the creation of a customer service hub, a single application that pulls and centralises data from every system across the bank.
Westpac’s end-state vision for the Oracle-based customer service hub - where, for example, a customer who gets disconnected from a call centre could easily continue that conversation in any other Westpac customer service channel - requires a rebuild of its operations from the middle out.
Earlier this month Westpac revealed it had passed the first goal post in the multi-year project by originating the first lot of customer home loans on the platform.
This mammoth architectural rethink is among a handful of reasons the bank doesn’t want to go full throttle into the deployment of artificial intelligence technologies just yet.
Its $600 million Panorama wealth systems overhaul - now “largely complete” - as well as a dedicated data and API program also have roles to play in its decision to put AI on hold for now.
“If we’re really going to use AI to remove complexity and make it easier for customers, we’ve got to lay it on solid foundations,” Westpac’s general manager of business integration Jayne Opperman told iTnews on the sidelines of the FST Media future of financial services conference.
“Obviously pace and time and getting things to market is very important, but from a technology point of view I think getting the architecture and the data right will allow us to leverage [AI] a lot better in the end.
“We’ve also got a huge amount of transformation underway, so we need to sequence [this] into the change schedule. There is an order of deployment that makes the most sense.”
This doesn’t mean the bank isn’t playing around with AI, however - it’s got several proof-of-concepts running at the moment to work out where the likes of IBM’s Watson and Westpac-backed start-up HyperAnna can provide the most value.
One trial involves credit card statement data; Westpac had identified an issue with customers not recognising merchants on their statement when the business name differs from the retail (trading) name.
Westpac deals with “a lot” of enquiries and disputes about perceived incorrect charges as a result, Opperman said.
The bank is testing out different cognitive software to identify aspects of data that could help address this problem.
”For example, you might remember the location of the store if we were able to show you the map of where the merchant is - you might think ‘Oh yes I did go there, that was me’,” Opperman said.
“Today as an industry we don’t pull through all those different aspects of data to statements,” she said - customers are only given the date and name of the transaction.
The trials are resonating with customers, Opperman said. But “the complex part is about how we introduce those types of technologies safely into the environment”.
“We’ve got a series of thresholds [the technologies] have to meet and we would absolutely hope that if we see real value in them for customers we’ll deploy them into production,” she said.
One of the other things holding Westpac back from putting AI into production is its data holdings.
The bank is currently on a concerted campaign to make the most of the use of the data it holds, whether that’s through cleaning up unstructured, inaccurate and disparate data sets, or implementing APIs to enable better data sharing.
This drive is being backed in large part by the interest Westpac has in start-up Data Republic and the relevant partnerships it has sealed through its Reinventure investment arm.
The bank also has to prepare for a new open banking scheme announced by the government earlier this year.
Under the changes, Australia’s banks will be required to share the data they hold on a customer upon request by that individual. Consultation on the proposed scheme was completed in late September, and a subsequent report is due before the end of the year.
Also on the horizon is the looming national new payments platform (NPP), alongside a growing influx of data from mobile devices.
Westpac has been working to ready itself for these changes by making sure it has “the right types of APIs across the organisation”, and that its data is clean and accurate, Opperman said.
“Traditionally we’ve looked at data through a number of lenses,” she said.
“We’ve got regulatory and external reporting data, and customer data, where we have to get the accuracy absolutely 100 percent right. And then there are other pieces of data that we’re starting to supplement from partners in other industries, and we need to make sure that’s at the appropriate level before we open that up to customers.
“[So] we’ve built an API platform that allows us to really modernise the environment, together with the investment into platforms. We’re moving away from technologies of the last decade when we all built enterprise service buses, and adopting APIs.”
The changes are also being made in the expectation that the finance industry will start turning to different datasets to make decisions.
“What we can see happening in geographies like India for example, where they don’t have the history of seven years of financial data, they are making credit decisions on things like where you went to university and the type of degree you hold,” Opperman said.
“We’re not ready to to do anything like that, but it does give us an insight into what the future might be like.
“I think we need to appreciate some of the differences that come from having mobile devices and real-time data - that will impact the types of data we need to collect.”
iTnews has a new report on the adoption of artificial intelligence within Australian enterprises. Download the report here.