Bendigo and Adelaide Bank is finding hardship in customer calls with AI

Bendigo and Adelaide Bank is building a capability to recognise signs of customer hardship during a live call, enabling it to deal with any issues in the moment.

Megan Papadopoulos, centre, speaks at the symposium.

The capability is using a near two-year-old deployment of Amazon Connect in its contact centre operations as a foundation upon which to introduce AI into its customer service operations.

General manager of customer contact Megan Papadopoulos told an AWS financial services symposium last month that the contact centre is one of the top AI users in the bank.

“Amazon Connect has given us the ability in the contact centre to be at the leading edge of the use of AI within the bank,” she said.

“Between us and the tech team, [we] would be the two teams that are using AI in production in the most ways.”

In the contact centre, AI is being used to augment the work and ability of human agents, and to reduce customer, organisational and regulatory risk.

Hardship is one area where the bank is putting AI capabilities in Amazon Connect to production use.

“We're using AI to improve risk outcomes for our customers,” Papadopoulos said.

“We are building out little things and putting them together to create more complex controls.

“One would be around hardship. The regulator's really keen that we do not miss hardship, and when someone indicates hardship, you can't ask them to tell you again, and they don't have to use the [specific] word.”

Papadopoulos said that because human agents “sometimes missed” hardship triggers in calls, the bank started using Amazon Connect to “pick up [those] missed hardship triggers”.

“We then moved to a Gen AI search, which picks up hardship as an intent much more successfully than the keywords, and more successfully than some of our staff,” she said.

“We’re doing that as a detective control at the moment. So a couple minutes after the call picks [hardship triggers] up, a team leader is notified, they'll listen to the call, coach the staff member, [and] fix the item for the customer.”

The next step is to flag the presence of hardship triggers with the human agent in real time.

“The next step is that we'll be doing that in real time, where it'll pop [up] to the staff member and say, ‘Hey, that sounded like the customer might be in hardship’.

"The staff member can say yes or no, and then deal with it appropriately in the moment, and that'll be a great experience.”

Papadopoulos said that other use cases beyond hardship are also set to be explored.

“The next use cases are around recognising vulnerability and customer complaints, so … we are augmenting staff to provide a much better experience for the customer.”

Papadopoulos added that the work reflects a broader vision in the bank to take a “data-led approach to improving our customer experiences.”

Ry Crozier attended the AWS financial services symposium in Sydney as a guest of AWS.