Sibos 25: AI & rethinking data

Artificial intelligence (AI) is helping banks provide deeper insights to customers from data, according to Sally Robinson, Head of Data and New Propositions at ANZ.

Traditionally in banks, data can be easily siloed between teams — with markets data in markets teams, trade data in trade teams and so on, Robinson told On Air with ANZ Institutional. AI is now breaking down those rules.

“When the AI sits in a nice layer in the middle, then you can retrieve [data] across functions and across teams,” she said on podcast. “And then you can actually query quite interesting things that hadn't perhaps been thought of [before], because people tend to stay in their lane around the type of query they have.”

Drawing from various sources helps link insights for customers into a broader economic picture, Robinson said.

“From a customer perspective, maybe there's a correlation between a commodity price and then the footfall of a customer in a mall,” she said. “Maybe there's something that we should be exploring around the coffee there.

“The AI helps us to get quite brave around queries, and a lot more curious.”

Robinson made the comments on a podcast with Hari Janakiraman, ANZ’s Head of Industry & Innovation, ahead of the Sibos financial services conference in September. You can listen to the podcast — the second in a two-part conversation — below.

[video]

Profoundly

Janakiraman said there were prospective use cases for AI in data over the immediate, near and long terms.

“In the near term, we are seeing AI tools already in place to help treasuries with cashflow forecasting, predictive analysis, portfolio management,” he said.  

“Those tools are already being deployed and they're being adopted at various speeds by various customers, depending upon their level of maturity, their technology, adoption rate, etcetera.

“That is profoundly changing…how quickly our customers can react to that type of data and make decisions.”

The next use case centres around agentic AI, according to Janakiraman, where AI uses data to string several agents into a multi-task purchase workflow.

“That is the more medium-term use, and all of it is powered by data,” he said.

The long-term case is a bit more out there, and still a little while away, Janakiraman admits: quantum computing and AI.

“Quantum is still a while away,” he said. “But it can profoundly change… how many data can be analysed at the same time, and how granular it can get. And that will profoundly change the way AI is going to be used.

“We are seeing that movement towards data being increasingly important in how operations are run, treasury operations are done, procurement operations are run, and finance operations are run. All of that is going be powered by data.”

Security

Robinson made it clear the security and anonymity of the data was crucial for its use in the production of any insight. Both bank policy and regulatory expectations are clear in how data can be safely used.

“We welcome [the] clarity of message around what we are and are not allowed to share,” she said. “I think it gives us really good rails to run on, and it gives us a very firm position within our bank of what we think is appropriate.”

Robinson said modern data security processes meant by the time they are ready to use, the data has “almost changed in its profile from being that original data point to something else”.

“There's technology that allows you to take a dataset and then de-identify it, which means that you can no longer see that it's mine or Hari’s… and it becomes aggregated,” she said.

“And when we do both of those things we operate within a policy and then we have secure data transfer rules, we actually create quite a strong framework.”

Listen to the podcast above to find out more.