ITWeb TV: FNB moves past AI hype to measurable business value

As financial institutions move beyond artificial intelligence (AI) experimentation and begin scaling deployments, First National Bank (FNB) says it is assessing the value of its AI investments through four key pillars.

The financial institution says it is focused on whether the technology is producing measurable business outcomes by measuring and assessing revenue generation, operational efficiency, fraud prevention and the impact of AI on employees, she notes.

Over the last few years, FNB has been introducing AI across customer-facing banking and employee workflows, including its FNB banking app and NAVi, which supports advisers with real-time customer insights and routine tasks.

In an interview, Dané Liebenberg, AI executive for FNB Retail and Business Banking, tells ITWeb TV how the organisation assesses whether its AI deployments deliver value across the business.

“We look at revenue from a sales conversation perspective. The next one is efficiencies, like how many efficiencies have we gained across different processes, including cost to serve? And then we assess our defence model from a fraud losses perspective: how is that model operating at the back-end?

See also

“The fourth area is the effect AI has on employees, particularly whether it can remove repetitive administrative work and allow staff to spend more time on activities requiring human judgement and advice.”

One of the clearest measures of AI’s operational impact at FNB is the amount of time that can be removed from existing processes, she continues.

Liebenberg points to back-office activities where automated models have reduced the amount of manual work required, allowing processes that previously took several days to be completed more quickly.

“Let’s take a back-office example, where something took five days to do. Right now, you can use certain models to do it in two days. So, there’s a lot of examples where we try to free up employees’ time.”

SA’s major banks told ITWeb early this year that they are entering a new phase in their AI journey in 2026, shifting from pilot projects and isolated experiments, to large-scale deployment across core operations.

Standard Bank says it is shifting its AI strategy from experimenting with individual tools and use cases, to embedding AI as a core organisational capability.

Dané Liebenberg, AI executive for FNB Retail and Business Banking. (Photograph by Lesley Moyo)

---

Nedbank, meanwhile, is testing a new hybrid banking model that combines self-service technology, digital on-boarding and human assistance in retail environments, as banks rethink the role of traditional branches.

FNB is already using AI beyond customer-facing applications, with Liebenberg highlighting NAVi, an AI agent designed to support customers and the bank’s advisers.

For employees, the system is intended to make relevant information available through their existing interface, reducing the amount of time spent searching for information and handling routine administrative work.

“The AI platform acts as a co-pilot for financial advisors, taking over repetitive administrative tasks, while assisting with personalised financial planning to scale their services and focus on holistic wealth management.

“The broader objective is to give employees more time to focus on customer advice rather than administrative activities.”

This forms part of FNB’s wider effort to use AI across customer journeys and internal processes, while retaining human involvement where judgement, advice or customer support remains important, she states.

FNB sees the evolution of banking technology moving beyond mobile applications towards what Liebenberg describes as an “intelligence interface”.

The shift is intended to make banking interactions more proactive, with technology helping customers navigate services without requiring them to manually search for every solution.

“The technology is really relevant if it solves the right problem. I think that is really what I’m passionate about. The bank’s objective is to make customer journeys more intuitive and contextual, using data to understand what a customer may need at a particular point in their interaction with the bank.”

She believes this could ultimately result in technology taking on more of the administrative burden associated with banking, while customers continue to interact with the bank through increasingly intelligent digital channels.

Liebenberg says AI can also be used for checks, balances and verification that previously required employees to carry out more routine tasks.

The same approach is being applied to processes such as know-your-customer requirements, where FNB continues to perform the required checks but has introduced technology and process optimisation to make the experience more efficient.

Liebenberg says document verification processes are now being performed in a fraction of the time.

As traditional banks compete with digital-only challengers, FNB also sees its established data infrastructure as an important asset.

While newer banking players can move quickly without the burden of older infrastructure, Liebenberg says established institutions have accumulated extensive customer data and sophisticated fraud and compliance models over many years.

“The branchless newcomers don’t carry that legacy, right. But where the traditional banks have advantage is that deep data and insight we have, which is quite critical. Fraud models and compliance models have been built over years with the use of some of these insights in mind.”

For FNB, that data is also central to developing more contextualised customer experiences and AI applications that can operate within existing banking processes.

As AI becomes more involved in customer interactions and financial services processes, FNB says governance and human oversight remain central to its deployment strategy.

Liebenberg says AI use cases go through governance processes, risk assessment and human oversight, particularly where technology could influence customer decisions.

“Every AI use case is important. It goes through the right governance, the human oversight on it and the right risk models. Adequate risk sign-off is essential prior to deployment. Because ultimately, we offer a service that we really want to customers to trust, therefore we need to earn that trust.”

She adds that the bank’s longer-term objective is not simply to deploy the largest number of AI models, but to build an interface that makes banking more useful, while maintaining customer trust.

Looking ahead, Liebenberg provides a glimpse into what the future holds for FNB: “We expect banking interactions to become increasingly conversational and proactive as generative AI develops further, while security, governance and responsible AI deployment remain important considerations.”