Westpac at work on a new agentic ecosystem

Westpac is poised to accelerate its adoption of agentic AI across the bank having started a fresh program of work on a new data ecosystem purpose built for the technology.

The ecosystem, which Westpac is calling its “agentic data ecosystem”, will build on the bank's previous work developing an intelligence layer that chief executive Anthony Miller revealed to investors late last year.

Speaking at a Snowflake customer event in Sydney, chief digital, data and AI officer Andrew McMullan described the ecosystem as “one single living system” for Westpac’s business definitions, data lineage, governance, quality and ownership, creating “a shared intelligence foundation”.

“Based on our framework of trust, we want to build trust into the way we create and use intelligence, because most organisations don't struggle with a lack of data - [they] struggle with confidence in it," he said.

"[They] struggle whether or not [there's] the right controls and lineage; [and on] the time spent trying to debate the numbers instead of acting on the data and making intelligent decisions.

"That is what the agentic data ecosystem is designed to solve at Westpac,” McMullan said.

The bank has increased the number of data sources feeding into its intelligence layer to 285, up from the 251 it reported this past November.

McMullan said that the layer has accelerated its time to build and deploy AI models five-fold, while making the process of converting insights based on customer data into engagement six times faster.

The bank has also sped up its ability to convert customer data into offer-based marketing campaigns.

“Rather than building it from scratch, the right offer can now call to action and surface from a pre-approved catalogue, matched to the engagement, consistent on brand, already assured, and then we go through all of the inclusions and exclusions and select the audience. And what used to take months now takes days,” McMullan said.

AI for bankers

McMullan also revealed some of the bank’s early efforts to structure AI into the work routines of its employees, based on integrating Westpac’s intelligence layer with Snowflake virtual assistants: CoWork and its equivalent for data engineers, CoCo.

It has built three main platforms that are broadly designed to reflect bankers’ needs across a typical working day, internally naming them “Start Your Day Informed”, “Create a Better Day” and “Sleep Tight at Night”.

Start Your Day Informed operates as a market intelligence tool for bankers that provides them with a daily update on based “thousands” of signals collected from across multiple bank systems.

The bank provided less detail about Create a Better Day’s specific functions but McMullan said it was used across multiple teams to analyse economic activity with the ability to narrow by geographic areas.

“What emerged as we were building this was truly fascinating. Commercial bankers are using it to prepare for conversations and identify opportunities that they otherwise just wouldn't have seen. Institutional bankers are telling is that it's helped to strengthen underwriting discussions and challenge assumptions,” McMullen said.

McMullan also gave less attention to explaining the final piece, Sleep Tight at Night, but offered clues that the platform is integrated with the bank’s risk management and credit control systems.

“[We] want all our customers to sleep well at night, knowing that everything that we were helping them with during the day is going to happen overnight, because risk credit, and control are all monitored continuously,” he said.