NAB readies to scale AI-based customer interaction

National Australia Bank is gearing up for a major expansion of its AI-driven customer engagement armoury after maturing its so-called “customer brain”.

Customer brain is the result of a muilti-year effort by the bank to develop ways to use AI and other technologies to process customers signals, anticipate their needs and automate digital interactions with them.

A key objective is to interpret the signals using AI models and personalise a response back to the customer.

The types of applications for the system range from supporting simple marketing campaigns and recognising debt repayment milestones to entire dispute resolution management workflows.

The pool of AI models that customer brain draws on has grown to 3500 and its adoption across the bank’s divisions has increased to 90 percent.

Speaking at a Pega Innovate event in Sydney, NAB’s acting executive for customer decisioning and data science Lisa Marchant said the customer brain is able to suggest "over 400 next-best actions" to customers today.

"A big focus for us is how we actually scale. We're wanting to get into the thousands,” Marchant said.

“The other thing that we're really focused on in terms of our strategy is really thinking about when we're choosing between multiple actions for a customer that the right decision is made.

“We're really focused on a balance of both value and quality,” she said.

Behind the scenes, the bank also appears set to introduce more prompt-based application design with Marchant revealing that it had started using the Pega GenAI Blueprint platform with a view to using it as a potential means to achieve scale.

Blueprint is Pega’s offering in the low-code, generative AI application and automation design market.

Broadly speaking, platforms like Pega Blueprint aim to use AI to automate the process of analysing older software code, and interrogating business documentation and policies, to recreate their workflows in more modern applications.

In NAB’s case, for instance, this could mean importing customer dispute resolution or debt repayment plan workflows from legacy applications and redeploying them as modern conversational AI agents to deal with customers in the place of humans.

The sophistication of agents that NAB adopts could see it use them to provide basic financial guidance, interpreting customer needs and alerting them to options to deal with their circumstances that might not previously have thought available.

At its Sydney event, Pega demonstrated this concept using a fictional customer of a hypothetical lending institute seeking to delay paying a debt.

The conversational agentic interaction allowed the hypothetical lending provider to automate the process of identifying that the customer had an alternative option and suggest it.

It also provided another example of fictional customer seeking a debt repayment plan and showed that the interaction could be complete entirely as a voice interaction.