Inside Commonwealth Bank’s quiet AI revolution

The banking major’s early adoption of the cutting edge technology is about to be unleashed on millions of customers.

Commonwealth Bank chief Matt Comyn speaks at the bank's technology briefing in Sydney on Thursday.

Commonwealth Bank was an early adopter on artificial intelligence and today it is among the most advanced in the “real world” use of the powerful technology among Australian businesses.

In a little under three years, AI has moved from a novelty used inside CBA’s South Eveleigh campus to last year being rolled out internally to speed up processes for staff.

Earlier this year, CBA extended its partnership with Microsoft through a wider adoption of AI across the bank, including to help tackle fraud. The results of this and along with work from thousands of the bank’s own tech engineers, AI is about to be unleashed on CBA’s 10 million customers.

From a standing start this progression shows just how quickly AI will find its way through the economy when business starts adopting the technology at scale. CBA chief executive Matt Comyn talks of 100 years of discovery coming in the next 15 years and that represents a big shift for the way today’s businesses will operate.

Commonwealth Bank is about to switch on AI technology for its customers. Picture: NCA NewsWire/Daniel Pockett

Now the CBA boss has taken a big step forward in his AI journey, by tapping the technology to power the bank’s messaging services and live chat with customers.

CBA is one of only a few banks to have done this globally, with US and Canadian banks leading the way. From the outside it may not seem like revolution just yet, but there’s a powerful engine now sitting behind daily queries which is only going to get smarter.

CBA currently fields around 50,000 inquiries a day around the clock via messages. This is already a five-fold increase on usage over recent years and this is expected to increase sharply in coming years.

AI allows a more sophisticated chat experience and can answer more complicated and broader range of questions.

“What generative AI allows you to do is, instead of asking for a fixed input and producing a fixed output, you can actually vary the output depending on the customer,” Comyn tells The Australian.

The technology CBA is designing is built around the idea that different people ask questions about essentially the same thing in many different ways.

“This is where the statistical patterns come in,” Comyn says. “Depending on the context, more information from the customer, the tone, a whole range of different things, it enables you to effectively predict the best answer for that particular instance.”

The savings for CBA are real – although at this stage are hard to quantify.

The smart chatbots mean the bank doesn’t have to add on hundreds more call centre staff – particularly during peak holiday periods when Australians tend to lose their credit cards on international trips. The process for a card replacement and emergency solution then becomes fully automated.

At the same time AI is now quietly sweeping millions of transactions a day seeking to stamp out fraud and block suspect payments before they happen, representing more savings for the bank.

Ultimately, though, those tens of thousands of queries a day can all been answered satisfactorily and in real time without a person involved, and that becomes a major productivity boost, freeing up call centre staff for more complex questions.

In the US players like Bank of America have pioneered smart chatbot Erica, which has now passed 2.5 billion interactions. As well as answering questions around branch or account numbers and finding transactions, Erica is giving the bank’s customers insights into spending behaviour and active monitoring of subscriptions.

In Australia other big banks are using AI tech, but it is still anchored in the back office. ANZ has tapped the tech to help it write vast amounts of code for its new ANZ Plus platform and National Australia Bank is building the customer brain which runs of hundreds of machine learning models. The NAB program pushes out customer interactions, including meeting savings goals, spending habits as well as taking feedback on the bank’s performance.

With the right guardrails in place the technology gives the potential for more individual service for customers, ultimately leading to a better experience, Comyn says.

“As it gets better and better you will get your much higher quality actions that you’re able to take to customers in near real time.”

Originally published as Inside Commonwealth Bank’s quiet AI revolution