At NAB, Swinburne is responsible for determining how to make customer experience first rate. And he’s looking at AI more and more, particularly in the area of chatbots, to achieve it.
NAB is looking at it on several fronts. “We’re looking at proof of concepts around how you team up with a third party, maybe a software provider, and use some AI to do some deep analysis on customer’s transactional behaviour on our side, and maybe their software on the other side,” he explained. “Then you can start to build a solution to a problem the customer may not even know they have.”
The bank’s biggest challenges revolve around resources and determining how much money gets spent on AI, versus legacy systems versus other digital projects.
“What do you do first? Is it a chatbot? Is it fulfilment? Do you leap ahead with speech?” Swinburne asked.
For BPay, the pursuit of AI is at the beginning stages, Donohoe said. The overarching mission is finding practical and valuable ways of offering services to the customers it serves. BPay is also trying to understand the end-to-end payments experience as it relates to AI and what the company can do in order to help reduce frictions in commerce and use of its products and services.
“We’re working on a venture where we’re exploring the opportunity to lift data from unstructured documents and turn those into the opportunity to use AI to help with bill payments and other types of use cases,” Donohue said. “That’s quite exciting for us because it’s stepping outside the realm of what we would normally do. It will help us service our customers in a different way.”
BPay is also exploring data science and using all sorts of techniques and learning. She recognised a top challenge in AI is getting access to adequate data sources in order for the algorithms to learn.
“One of the things we’ve had to overcome is how you get those sources of data while also making sure we’re respecting privacy and the use of that data,” she continued. “It’s then learning from that and feeding that back into the undertaking.”
For BPay, the desire is how to be different from the competition. “There are lots of people investing in similar areas. And so understanding what everybody else is doing, and seeing if you can leverage those solutions, is something to be thinking about,” Donohue said.
Another big challenge is talent. “We need to have people to conduct the experiments and actually turn them into production-grade solutions,” she said. “Looking at the emerging data pool for data scientists and analysts, I think we’re going to see a shortage. There are light years between a good and poor data scientist.”
Hubbard said UBank, a challenger, digital-only brand, is focused on changing banking for its customers. The company commenced its foray into AI about 18 months ago and has two experiments in production now: One customer facing, and one internal.
One launching in coming months will use big data to do deep analytics and provide unique insights for customers, Hubbard said.
“The lesson we’ve had over that last 18 months is the more we understand what’s possible, the more use cases that we spot, and the more opportunities that we see,” he said. “So it’s been a good start, but we’ve got many more to do.”
In delivering the two use cases, Hubbard said he was amazed just how quickly the team was able to move from idea to production. The first use case was a chatbot, dubbed RoboChat.
One of the interesting things about our chatbot is its focus and its scope. It is very much targeted on helping customers as they apply for a home loan in a digital-only channel. We’re able to go from idea from finishing developing in six weeks, and were in production within eight weeks,” he said.
The tool has involved quickly since going into production and now answers about 80 per cent of customers questions correctly on the first attempt.
“It’s had a dramatic improvement on our home loan conversion rate, particularly the part of the experience which is starting a digital application through to getting to the end of that application,” Hubbard said. “Across the customers engaged with the chatbot, we’ve seen a 15 per cent increase in conversion rate. It has been a great chance for our team to get their first taste of developing using some AI toolsets.”
This sparked the next phase of its journey, which flipped the concept on its head and sees teams taking insights from the chatbots to be used by customer service agents.
“We’ve invested over 1000 internal documents from our various knowledge databases and information sources and using natural language processing, and some machine learning models, teaching it about UBank, and then enabling our customer service agents, through a single search query, to get real-time help as they service our customers,” Hubbard said.
For UBank, a short-term problem is data and access to data. “We’re only a 10-year old company, yet somehow we still have legacy systems and we still have challenges getting our data, particularly our global data, our customer data, from legacy data stores into a place where we can use it for AI,” Hubbard added.
View the original article online at https://www.cmo.com.au/article/print/641236/impact-ai-4-key-brands-cxo-panel/