8 minute read

Moving forward

MOVING

FORWARD

UK lender TSB and Australia’s Sandstone Technology have both seen artificial intelligence come to the fore during one of the most volatile homebuying periods in history. The bank’s Mike Gamble and Sandstone’s Ross Watts swap stories

There’s no place like home, but millions of us were desperate to change ours when the world was told to stay indoors.

Eight months after the first UK lockdown in 2020, demand for mortgages was at its highest in 13 years. Breakneck property sales were still going strong the following autumn, with spiralling house prices defying the rules of normal economic activity, as a set of extraordinary circumstances combined.

Near-zero interest rates, low inflation and fiscal and monetary policies designed to support the economy and the property sector in particular, a demographic shift as working from home became the norm for many, and household savings accumulated from months of restricted activity, all conspired to create the boom.

On the other side of the world, a similar picture emerged. In Australia, where, if anything, COVID restrictions were even more severe, house prices in Sydney rose by 15.1 per cent in the first five months of 2021 alone; Melbourne, which endured the world’s longest lockdown, saw a 9.4 per cent increase over the year. By the end of 2021, there were more home loan applications in the system than at any point in the last decade.

The public’s sudden desire to up sticks and move left many financial institutions in both countries overwhelmed; such hot property markets put direct and collateral stress on banking systems and forced lenders to look for ways to stay ahead of the game. TSB Bank was one that massively benefitted as the UK packed its belongings into removal vans because the lender was digitally well-prepared for the challenge.

Remembered by many in the UK from its 1980s advertising slogan, ‘the bank that likes to say yes’, TSB had been forced to rebuild both its reputation and its IT infrastructure following a disastrous tech migration to new owner, Spanish bank Sabadell, in 2018, when millions of customers were locked out of accounts for weeks. And, as average house prices increased by 10 per cent over the year, TSB kept pace with buyers and said ‘yes’ to the tune of a record £9.2billion of gross mortgage lending.

By placing an emphasis on digital, it had built what is widely acknowledged as one of the most modern banking systems of any high street lender – although, critically, as Mike Gamble, director of analysis and design at TSB, stresses: “The human element of banking is really key to us and how we connect the digital world to the human world is absolutely front and centre for TSB.”

Ross Watts, chief customer officer of Sandstone Technology, whose LendFast platform processes about 20 per cent of all home loans in Australia, couldn’t agree more. There, processing staff, in addition to dealing with the sheer volume of business, were also under pressure to stay on the right side of compliance after a series of scandals that had seen major lenders fined for sending billions of dollars to operations overseas without appropriate governance – errors that, Watts argues, could have been avoided by using AI.

In both TSB’s and Sandstone Technology’s cases, AI has been the answer to meeting the needs of customers during a pandemic, in a fast, compliant – and human – way. It could take care of the hard stuff in the background by making processes more efficient, while freeing up people to deal with issues that require personal interaction – to the benefit of the banks and customers alike.

The jewel in TSB’s digital crown is its AI-driven TSB Smart Agent, developed with IBM and launched in the mobile app within just five days at the start of the pandemic to give customers immediate access to measures such as repayment holidays for mortgages and loans.

The live chat service – which uses IBM Watson’s natural language processing and is powered by LivePerson – answered more than 40,000 customer requests between March 25 and May 1, 2020, alone, using a combination of virtual assistant and employees.

“Our AI Watson brain within TSB Smart Agent is very good at identifying when customers are in financial difficulty, for example, or they’re vulnerable,” Gamble explains. “And we just flip them straightaway from talking to a TSB Smart Agent and connect them in real time to a human being. We see that as the most powerful development in how we are learning how to use this channel and technology going forward.”

About 70 per cent of TSB’s five million customers are fully engaged with digital and the vast majority (90 per cent) execute their own transactions such as paying bills or moving money, says Gamble.

“But, for me, the biggest learning is that it cannot be just standalone AI. When you want to speak to somebody about buying a home or restructuring your debt, or you’re in financial difficulty… that’s when the interaction is really important,” he says. “In retail banking, we’ve had to move fast to catch up with that.”

“We recognise we can’t be experts in everything,” says Gamble.

For Sandstone, it’s AI’s role in back office processes that has really came to the fore over the past two years.

“It’s almost impossible to handle the amount of data that is going through legacy systems, from a transactional perspective, using anything other than an AI-based solution,” says Watts. “Not only are we able to speed up the time per application, in terms of days to approval, we’re also able to increase the number of applications that are handled, and reduce the amount of work and error.”

Sandstone’s AI-enabled Digital Intelligent Verification Assistant (DiVA) platform, for example, claims to reduce manual touch time in the assessment process by up to 80 per cent. As such, AI has become a critical part of the home-buying process in the fiercely competitive Australian property market.

“AI is integral to our home-lending offering. It’s almost impossible to think how asked for she knew, she knew how it worked, she knew how to support them, she pretty much knew every customer, as they walked in the door as well. I was just blown away with how professional this woman was,” he says.

But then he sat with her behind the counter and realised that, on the other side of the screen, were hundreds of Post-It notes

It’s almost impossible to handle the amount of data that is going through legacy systems using anything other than an AI-basesd

solution Ross Watts, Sandstone Technology

For me, the biggest learning is it cannot be just standalone AI

Mike Gamble, TSB

TSB Smart Agent is on a continuous evolutionary path of improvement – another advantage of using AI and machine learning-based systems, adds Gamble.

“We’ve seen more than 1.2 million conversations since launching with our customers. We teach the bot some basic principles of how banking works and how to support our customers, and then keep building on it. We get the analytics out and we’re tweaking it every day, As we keep learning, we keep re-educating it.”

Gamble gives an example: “I never knew that a common term for ‘I’m moving home’ in Scotland is ‘flitting’. But our TSB Smart Agent now recognises that and will immediately help the customer record their new address.”

Key to evolving, particularly at speed, has been working closely with partners such as IBM Watson and LivePerson, as well as with Adobe on introducing digital forms and electronic signatures. much longer the home loan process would be for customers, for those people in processing centres, for the brokers who are referring loans, if we didn’t have AI intelligently sorting and managing those documents, in terms of identifying the key phrases, putting them the right way round, making sure that the dates are correct and valid, and then sending out notifications to accelerate that process,” says Watts. “And then AI is also helping lenders in so many ways to meet the obligations they have – to their customers, the regulator, and to their boards.

“Those financial institutions that have AI within their processes are able to get to an approval in three or four days. There are other, really big banks over here being criticised because their approval timelines are about 23 to 24 days. That illustrates the quantum of difference,” he adds. AI in the back office can keep everyone in the property buying chain automatically updated – freeing up time and space for those human moments that really matter for customers dealing with one of the most stressful purchases of their lives.

Gamble harks back to watching an in-branch colleague who was ‘absolutely amazing’ at her job: “Everything a customer on the wall, telling her everything she had to remember. AI is modern banking’s automated equivalent of the Post-It note prompt.

“Much of the conversation with our clients around AI is about how we can remove low-value, manual handling processes from the system,” says Watts. “How do we speed up the capture of information that’s going to provide insight to the business, to make decisions more efficiently and with more accuracy? AI plays a really important role in that, in addition to the areas that are customer-facing.

“What you get, as a result, when you think about cost-to-income, is a happier customer at the end of the process, and a happier introducer (the vast majority of the market here in Australia is broker-introduced). And, if you’ve got happy brokers, you’ve got someone choosing you over a competitor. You’ve also got happy colleagues; people in the processing centres don’t have a customer or a broker breathing down their necks for information. That leaves the great, high-five moments. And those are the interactions that can be had between humans, because AI is taking care of the difficult stuff and creating a really efficient process.”

This article is from: