Last week it also revealed that it will allow its customers to hold and use bitcoin and other cryptocurrencies, and it has also made investments in energy and telecommunications retailers to add deals for its app users.
H20.ai describes itself as “democratising AI,” by providing a platform that organisations can use to develop their own AI-based apps and services. This ethos matches closely CBA CEO Matt Comyn’s objective for the bank to be a leading player in the AI era.
In a media briefing, Mr Comyn said there were broad applications for AI and machine learning technology across its business, in operational processes, fraud protection, risk management and credit assessments to make lending decisions.
“We have run some of these machine learning, AI techniques across credit models already, and you can see the improvements in terms of predictive accuracy,” he said. “I think there is enormous potential, especially in areas like small business.”
It will also be integrated into customer-facing products such as online shopping app Little Birdie, buy now, pay later service Klarna, CommBank Rewards and gift card start-up Karta.
The CBA boss said the investment would allow the bank to compete more effectively against global technology companies, not just other banks.
Machine learning ‘grand masters’
“Data is really at the heart of understanding about delivering great experiences for customers. I think institutions and companies who can do that extremely well will find it is a very important part of their competitive advantage,” Mr Comyn said.
“Clearly over the medium and long term, we see the competitive landscape changing quite significantly. … We are competing with some of the big global technology companies, and we want to make sure we can deliver a better experience, that’s relevant and localised in the Australian market and context.”
H20.ai founder and chief executive Sri Ambati told The Australian Financial Review that CBA has used its cloud-based AI platform for two years, but would now get a dedicated team of experts, including machine learning “grand masters,” to help it develop new AI-based applications and services.
CBA’s chief data and analytics officer Andrew McMullan will join the company’s board and Mr Ambati said its vision was to help the bank become an “AI superpower”.
“They are a really demanding customer, pushing the ante, and some of our best modellers were starting to get really inspired by the work they were doing, so we were really keen to pursue this partnership,” he said.
“Our user base has often become our investor base because we let them experience our product and then have a co-innovation model where they become an inventor. They can create new applications and services that are much more valuable than the underlying technology infrastructure.”
The partnership means CBA is H2O.ai’s exclusive financial services partner in Australia, although it already works with other local clients like Telstra and the NSW government.
As well as CBA’s lead investment, New York-based Pivot Investment Partners contributed to the round, along with Goldman Sachs and UK-based Crane Venture Partners.
Mr Ambati said that, while some tech companies may bristle at the idea of agreeing not to work commercially with the other big local banks, he believed CBA was well ahead of its Big Four rivals in terms of AI sophistication.
He said its partnership with CBA would offer it strong growth potential with other banks globally, as well as helping CBA face the increasing threat from global tech giants.
Much of its newly raised capital will be added to H20.ai’s existing reserves, while it will also look to increase investment in its technology and hire Australian employees in sales and marketing roles.
Mr Comyn fielded questions about the bank’s greater use of AI, in the context of both the privacy of customers’ data and the potential for AI-led decisions to discriminate against some groups in society.
In a comprehensive report in May, the Australian Human Rights Commission made 38 recommendations calling for appropriate safeguards to be place for governments and companies using AI, including laws to ensure customers are notified when AI is used in material decisions relating to them.
“Governance, explainability and transparency” of the models would be crucial, including to meet the demands of regulators, Mr Comyn said.
“Perhaps customers aren’t necessarily as interested in that, but we are, and we believe regulators [are too]. That is an important part of how we are using technology, being able to add that degree of transparency.
“But ultimately, part of being able to deliver personalised, relevant experiences to customers that adds value to them, that is what builds trust.”