Stuck in first gear: The real reason why banks can’t scale AI  

Most banks are stuck in first gear when it comes to AI. They have the models and the talent, but they are slow to recognise their tightly coupled core technology and the data traps – which means most of their AI ambitions are collapsing. The uncomfortable truth? The biggest crisis facing financial institutions is not a technology gap; it is a crisis of inertia.

Australian banks are spending tens of millions of dollars on AI but are using it like a “digital Band-Aid”, according to AI experts at Thoughtworks.

Consider the humble onboarding process. Banks spend an average of $15 million annually on customer onboarding alone. Yet, only 4% of new accounts are approved on the same day, according to recent reports, and 18% of new customers simply give up.

The remaining 78% are stuck in an agonising multi-day wait, just to give the banks their business.

Banks built on monolithic legacy systems are unable to keep up with competition that effectively leverages the powerful differentiator – AI, to transform and reinvent entire business models.

Uncomfortable truths

“The banking sector is spinning its wheels when it comes to implementing AI,” says Omar Bashir, Enterprise Modernisation Expert at Thoughtworks, a global technology consultancy that integrates design, engineering and AI.

“Management teams shuffle through proof-of-concepts searching for workable solutions and yet fail to implement anything that might yield any meaningful change.”

Bashir, who works with global banks to scale their AI projects, discusses the growing realisation among CXOs of what it takes to steer an entire bank in a new direction.

“Banks now understand that their own data is trapped in technology that is no longer serving them,” he says. “They are trying to bolt AI onto monolithic legacy systems with limited access to quality data, and it is simply not working.”

Bashir wants banks not to look at AI as just another efficiency tool.

“It is a fundamental opportunity to redefine value, operations and customer engagement. This is the next wave of financial transformation, and by remaining stuck in first gear, banks are missing out on opportunities in plain sight.”

Scaling beyond pilot projects

Despite growing investments in AI, many banks struggle to translate innovation into measurable gains, Bashir says.

“This is a trap into which good ideas disappear and wither away.”

Smart AI experts are being employed by the banks, but existing infrastructure is dragging down the innovation potential, Bashir says.

“This is a tangible failure to scale. We have seen as low as only a third of the processes being fully automated, while the remaining still require manual intervention.”

Ultimately, this internal failure becomes a customer-facing disaster, he says.

“In many cases, customers are looking in through the front window, and they like what the bank has to offer, but they just cannot find a way through the front door and ultimately walk away.”

Escaping the innovation theatre

Banks need to focus on implementing a viable foundation that helps them leverage AI for business success, according to Gia Truong, Head of Data and AI at Thoughtworks.

Consider the Indian bank that Thoughtworks partnered with to fix its onboarding. A “systematic, foundational approach” helped the bank streamline the entire process with a human-assisted AI journey, Truong says. The result: over 100 agents had their query handling accelerated, and 1,500 employees got instant access to performance data.

“To scale AI like this, banks need a new playbook,” says Truong. “We call it the FOREST framework – six dimensions of AI readiness that help organisations overcome common blockers and scale pilots into production with confidence.”

Truong argues that banks fail when they ignore the ‘T’ in FOREST – Trustworthy AI. “This is not a nice-to-have,” he says. “A KPMG report found 82% of respondents are worried about data breaches. The trust deficit is real.”

“An Australian Securities and Investments Commission report in late 2024 found many financial licensees do not even disclose their use of AI,” Truong says. Trust, he says, is built by focusing on the ‘E’ – the Experience of Human-plus-AI. “This is designing the system to be human-centric right from the start. The task is to augment the human experience, not just replace it.”

But the deepest challenge is governance, he says. “This goes beyond data sovereignty – where data is stored. Whose world view is the AI model encapsulating? Is it considering cultural nuance, or just codifying bias?”

“Governance is knowing the difference between a correct response and an ethically correct one. Without those guardrails, banks are not building an asset; they are building a liability.”

Success with a competitive edge

The AI revolution is turning out to be less of a technology challenge and more of a test of leadership, Truong says, adding that the only path is a commitment to building systemic capabilities necessary for continuous innovation.

“Achieving a sustainable return on investment requires banks to embed innovation into every layer of the business, which requires modernising technology, unlocking the data for value through insights, scaling experimentation and humanising the relationship between people and machines,” he says.

“Everyone else? They’ll just keep polishing their proof-of-concept decks, wondering why their customers left and their transformation never arrived.” Learn more at thoughtworks.com/en-au