Westpac is anticipating a role for AI at every stage of the business lending process, building on the benefits it has already derived from applying the technology to aspects such as credit decisioning.
(L-R) RDC.ai’s Gordon Campbell and Westpac’s Dr Martin Anderson.
The bank recently offered a behind-the-scenes view of a partnership with RDC.ai, formerly Rich Data Co, that it struck back in 2021, but only talked about in early 2023.
The initial intent was to use RDC.ai’s machine learning technology to improve the lending experience for business customers and bankers, with a particular emphasis on explainable credit decisioning.
At the AWS financial services symposium in Sydney this month, both Westpac and RDC indicated the partnership has broadened to cover generative and agentic AI, and more use cases beyond credit decisioning.
But the bank remained reluctant to reveal too much about how it uses the technology.
“When we look at the tangible outcomes, there are a number of areas I can’t talk about too much because they’re market sensitive,” head of technology and business lending Dr Martin Anderson said.
The technology is serving its initial purpose: influencing the way the bank assesses credit applications from its business customers.
“If we can understand the data, we can be transparent about the reasons we approve or decline applications, or we can be transparent with the way we want to treat customers, either by extending credit or helping them manage them through some hardship,” RDC.ai’s cofounder and chief market and strategy officer Gordon Campbell said of the initial use case.
“Those things will help banks deal with their customers more proactively.”
Anderson said it wasn’t just about ensuring the right decision was made, but also that the outcome was explainable, which would go a long way to “making the regulator comfortable with what we’re doing.”
By all accounts, the technology has worked to the point that broader use of AI in the business bank can now be contemplated.
Anderson predicted that AI, in its various forms, would “have an impact at every single stage of the [lending] process.”
“We will have to leverage it to get the most out of the technology and to get the best outcome for our customers,” he said.
“Whether it is faster decisions, better decisions, protecting the customers, protecting ourselves from fraud and scams and the like, the usage of AI will be even more ubiquitous than it is now.”
Anderson suggested that the whole process sequence would be redesigned for AI.
“People are going to be surprised by how these sequences are going to be reorganised, resequenced, rewired,” he said.
“It’s not just going to be a case of adding in AI-specific points; it’s actually reimagining the process so that you can actually optimise it to leverage AI and all the various capabilities.
“Whether it’s within documentation, KYC [know your customer], annual reviews, credit decisioning, the writing of credit memos, even communications – the whole end-to-end sequence of business lending can be optimised.”
Campbell said that RDC.ai has been working with Westpac and AWS on “how we can bring products to market that safely use agentic capability … that enable the bank to realise outcomes that they wouldn’t otherwise realise.”
He raised the prospect of an “interaction bot” to interact with portfolio governance and cashflow information within the business bank – data that is presently analysed in Excel or PowerPoint.
Anderson indicated there was latent and untapped value in the data holdings of the business bank.
“When you are looking at a dataset that [contains] ingoings and outgoings of an organisation like Westpac, it’s in the order of trillions of dollars of value, and we are mining that,” he said.
“There is so much insight in there, so much information, so much we can[use to] help our customers get a better experience [and] help them make better decisions.
“That’s where we are really leaning into.”
Anderson said that the three-way partnership between Westpac, RDC.ai and AWS acted as an effective accelerator to experiment with AI to solve key business problems.
“There’s an awful lot that we have available to us and a huge backlog of activity that we have that we probably wouldn’t be able to deliver at the pace that we are [without the partnership],” he said.
“Sometimes those experiments can be promoted to production, but sometimes it’s actually … for a hypothesis that can inform a roadmap,” Campbell added. “We’ve done that really successfully with Westpac.”
AI is anticipated to be incorporated into BizEdge, a digital tool Westpac built to reduce paperwork around “applying for a new loan, topping up an existing one, or setting up an overdraft”, as is described in Westpac’s documentation.
“That’s opened up huge areas for AI optimisation,” he said.
AI is also expected to help the bank realise a commitment of “bringing in more and new bankers to help our customers”.
“That brings a natural challenge where you have new bankers coming in and you have the policies, processes and products, and we’re expecting them to be experts in all these things,” Anderson said.
“That’s a massive challenge, but something like AI helps enormously in guiding them through all of these things.
“What we found as well is not only does it really help the bankers and the customers, but it’s made us look at some constructs of documents and say: ‘Actually, these need to be better, because they’re sometimes ambiguous or they’re conflicting’.
“This is actually being called out by using AI tools, which is a bit of an unexpected benefit there.”
Ry Crozier attended the AWS financial services symposium in Sydney as a guest of AWS.
