How ANZ is tapping AI to build a new bank and take on its bigger rivals

ANZ bank has invested $2.5bn into its new online-only products and CEO Shayne Elliott says his team of software engineers has used AI to write code for apps and other programs.

ANZ chief Shayne Elliott says the bank’s AI foundations are complete. Picture: Arsineh Houspian.

ANZ is using artificial intelligence to write its own code to keep pace with customer demand and eliminate bugs, as chief executive Shayne Elliot reveals the bank has poured $2.5bn into its new online-only offerings.

Mr Elliott said his team of 7000 software engineers had used AI to write more than 7 per cent of the bank’s code for apps and other programs in the past six months – “and that number will only ­increase”.

Mr Elliott is positioning ANZ as a more AI-powered bank, saying technology provided “tactical tools that will become table stakes” and enhance its competitive advantage.

He said the bank had spent $2.5bn on its online-only offering ANZ Plus and its business and institutional product ANZ Transactive over the past five years. But integration into “classic” apps has been slow, with the bank running parallel systems.

“You can’t build a skyscraper on sand and you can’t build it overnight,” Mr Elliott said.

“We have built that stack in Plus and Transactive. The foundations are complete and we are ready to leverage them.

“So unlike peers, who’ll continue struggling with multiple legacy, high-cost and ponderous platforms, our ambition is to have the simplest, contemporary platforms powered by the best partners. Getting that right allows us to better serve customers, help them for longer and substantially lower cost with faster deployment, unlocking the real benefits of some location.” ANZ has bought thousands of licences for Microsoft Copilot – the tech giant’s AI platform – which it is now using instead of its own in-house-developed Z-GPT, a generative AI tool that could tell staff about interest rates and respond to simple language queries.

“Every bank will use out of the box Copilots, and they’ll use them well. These are tactical tools that will be table stakes, and the advantages will be competed away to the benefit of customers, no different than how we all use Microsoft Office today,” Mr Elliott said.

“But the competitive advantage accrues to those who engineer strategic solutions using AI and intelligent automation into their propositions and into critical operations, and that’s only possible for the few who’ve invested in building the contemporary ­technology and data platforms ­required.”

The big banks have been racing to adopt AI and other technology.

Westpac chief executive Peter King kicked off a $3bn technology overhaul earlier this year, which is set to run until 2028.

After more than 16 years, Westpac and its St George bank brand still run on separate systems. Westpac bankers frequently switch between three or four systems for standard processes. Additionally, the industry-wide rollout of the Open Banking scheme cost Westpac 30 per cent more than necessary due to the extra systems in place – making the bank more expensive to run than rivals. It also increases the complexity for customers.

National Australia Bank is developing a “customer brain” built from billions of data points that aims to deliver 1970s-inspired service, analysing customer habits to get a better view of the customer and anticipate their needs.

Among the tools ANZ has deployed is Microsoft-owned software development platform Github.

Mr Elliott said ANZ engineers were using owned Github’s AI Copilot, which not only made them more efficient but more accurate.

“We have a team, some 7000 engineers, mostly writing software that millions of our customers use daily. At peak, every second of the day, more than 200 customers look at their bank accounts, and 300 payments are made, all done through the solutions our team engineers (have developed).

“We’ve systematically been giving modern tools to our engineers to help them write software faster, better, safer, because the better the tools, the better our engineers are.

“We added GitHub Copilot over a year ago, and all our leading engineering teams use it daily. As a result, more than 7 per cent of the code written at ANZ in the last six months was written by AI, and that number will only increase.”