As the banking sector faces pressure to accelerate growth and technology adoption amidst cost challenges and a rapidly evolving digital landscape, FStech News Editor Alexandra Leonards speaks with Daniele Tonella, chief technology officer at ING, to explore the bank’s digital priorities, the AI trends it is engaging with, and the key strategic pain points it is currently navigating.

With the rapid advancement of technology, evolving risks and regulations, and mounting competition from challengers, the traditional banking industry is undergoing unprecedented change.

Incumbents are under significant pressure to keep pace; those that fail to adapt risk missing growth targets, falling short of profit expectations, and disappointing customers who now demand seamless, tech-enabled experiences. At the same time, financial institutions must integrate emerging technologies into legacy systems while managing the long-term costs of advanced tools such as generative AI (genAI).

Recent data from the Treasury Committee illustrates the risks of failing to balance old and new systems effectively. Over the past two years, nine of the UK’s top banks and building societies have reported at least 803 hours of unplanned technology and system outages.

Technology as more than just a tool

Financial institutions adopting the latest technology must recognise that it is more than just a plug-in solution. Real benefits often require a cultural shift and a holistic digital strategy.

Daniele Tonella, chief technology officer at ING, believes the bank's approach is distinctive within the sector.

“It’s unique in a couple of ways – the first being the level of engineering I’ve observed,” he tells FStech. “Of course, I’m not referring to all 20,000 engineers, but I’ve encountered several who, in my experience, exceed benchmarks I’ve seen elsewhere.”

Tonella, who previously held leadership roles at UniCredit, AXA, and Swiss Life, says that at ING, even the commercial side of the bank demonstrates a strong awareness of technology.

“Colleagues on the commercial side don’t think in terms of tools – which reflects a lower level of maturity – that is, ‘I have a problem, buy me a tool, problem solved,’” he explains. “They think in terms of client value and experience.”

This mindset has enabled what Tonella describes as a radically different approach to technology.

“Moving away from seeing technology as a tool or something you can buy, to something that you are,” he continues. “That’s the key difference.”

A digital culture

Since joining ING less than a year ago, Tonella has continued to uncover new and innovative technologies being developed and deployed within the bank.

“Some of the most interesting solutions are hidden,” he explains. “I recently discovered a tool developed by one of our teams in Romania that assesses the risk of software deployment using a graph database – a very novel approach.”

While he finds such developments impressive, they are largely considered business as usual within the bank.

“They say, ‘Of course we have it – why are you asking?’ And I respond, ‘You don’t realise what other firms do or don’t have,” he continues.

Another strength of ING’s technology strategy, Tonella says, lies in its approach to Agile.

“Our implementation of Agile requires constant attention, as it involves a degree of decentralisation,” he adds. “We delegate a lot to tribes and squads, which is powerful culturally, because it means those closest to the work make the decisions – not someone in the City.”

He notes that this creates the challenge of orchestrating such freedom.

In previous roles, his main leadership challenge was often to inject energy into the organisation.

“Here, the issue is the opposite – it’s an organisation with excess energy," adds the CTO. "The challenge is focusing that energy.

"But it’s preferable – like surfing a big wave rather than a flat surface. It’s less stable, but more fun.”

ING’s platform engineering approach is also evolving how Agile functions and how software is developed and deployed.

“This helps us balance agility and autonomy with control,” says Tonella. “It allows us to maintain or accelerate the release of client value while avoiding the trap of spiralling complexity – the central challenge for any large tech organisation.”

A challenging landscape: stability, standards and economic pressure

Despite these strengths, Tonella acknowledges there is room for improvement.

For example, ING is enhancing the stability of its operations after several public production incidents.

“These incidents must not happen – stable production is a baseline expectation,” he says, referring to the critical systems supporting ING’s customers.

Like many traditional banks, ING has faced outages and ATM issues in recent years.

“If production is unstable or perceived to be so, no other message from the bank resonates,” he adds.

The bank is also progressing in its journey to consolidate technologies under global standards.

“We need to complete this, as in some cases we’re in the worst position – both global and local services are still active, so we incur double the cost,” he continues.

The most pressing strategic challenge, however, remains cost pressure, particularly given the current economic landscape.

“Technology is about managing the time axis,” says Tonella. “I can build something today that delivers software, but it brings long-term operational costs.

“Like buying a car – you also commit to insurance, fuel and maintenance. Tech is the same.”

He notes that those requesting technology do not always grasp the long-term implications.

“This is a strategic issue,” he explains. “With business growth demands and limited budgets, we must ensure the operational burden doesn’t suffocate our the available means to be spent on innovation, new things, and everything else. That balance is a very delicate one.”

This ultimately requires both foresight and internal persuasion.

“Some people in the organisation, even in my team, may say ‘this is my budget, I want to spend it on change,’ without realising it may lock them in," Tonella tells FStech. 2Three years later, only half their budget remains.

"Foresight is about having a strategic vision – architecturally and directionally – and asking: where are we heading?”

Tonella stresses the importance of delegating accountability within a clear strategic framework.

“If architecture, strategy and vision are unclear, how can I expect 400 tribes to act coherently?” he asks. “You must provide direction before you can enforce alignment.”

Agentic AI

Tonella describes ING’s approach to cutting-edge AI, including agentic AI, as “conservatively aggressive.”

“We’re exploring intensively,” he says. “We’ve identified five areas for genAI use cases, with around 40 under consideration.

"Some have been industrialised, some are scaling, and others are in early exploration.”

He explains that agentic AI follows the same logic – identifying and achieving measurable benefits.

“What’s hidden behind all of this is you cannot have effective AI without a well-managed data layer,” he says. “The old saying was ‘garbage in, garbage out.’

“With AI, it’s ‘garbage in, disaster out.’ So, there is some invisible plumbing necessary to have proper industrialised AI.”

He also outlines the security risks posed by external genAI tools.

“Some attacks, like poisoning the well, involve publishing vulnerable code snippets online, which then get absorbed into AI models such as ChatGPT or Copilot,” explains the CTO. “Developers may unknowingly use compromised code.”

Another challenge is that unlike traditional deterministic software, AI models evolve post-deployment.

“You don’t just deploy and forget – someone has to operate and monitor the model to ensure it behaves as intended,” he says. “Managing data layers and model lifecycles are emerging disciplines – and essential.”

Asked whether ING is prioritising foundational disciplines before exploring agentic AI, Tonella says both are progressing in parallel.

“The commercial side moves fast – they want growth and value – so we must experiment while building the groundwork,” he adds.

How AI is improving operations

While Tonella champions AI, he believes its real value lies not in mimicking humans, but in managing complexity.

The bank is using genAI to reduce calls to its helpdesk from customers unable to locate features in its mobile app.

“It’s frustrating for clients – if I ask them to call for something that should be visible, it’s a poor experience and generates call centre costs,” he says. “So we’re integrating an AI-powered search function to guide users directly in the app.”

On the engineering side, ING is also piloting AI in its security operations centre.

“Experts monitor attack signals, but 90 per cent of alerts are false positives. It’s tedious work,” he says. “We’re trialling AI to help engineers filter out false alarms based on historical patterns.”

Previously, engineers reviewed up to 10,000 alerts daily, with only 500 relevant.

“Pre-filtering improves the chance of spotting real threats,” he explains. “It’s AI not as a human replacement, but as human augmentation.”

Short-term vs long-term

Tonella acknowledges that long-term AI benefits are often deprioritised by businesses chasing quick results.

He mentions Klarna which last August said AI would play a role in further staff cutbacks following a reduction of employees from 5000 to 3800 driven by natural attrition.

At the time, the company's chief executive Sebastian Siemiatkowski suggested AI had been a driving force for its 27 per cent revenue growth in the first half of 2024. He also told the Financial Times that the firm could employee as little as 2,000 people as a result of using AI across customer service and marketing.

Speaking about the business' AI customer service assistant, which was launched in February last year, Siemiatkowski said: "Our AI assistant now performs the work of 700 employees, reducing the average resolution time from 11 minutes to just two, while maintaining the same customer satisfaction scores as human agents."

But in May, the chief exec admitted that its AI-driven job cuts had “gone too far”, with the FinTech now testing a new cohort of staff “in an Uber-type of setup” that can log in and work remotely and take up customers’ requests as representatives of the Buy Now, Pay Later firm.

He also revealed that the business is planning a recruitment drive so that people using its customer services have the option of speaking to a real person rather than a chatbot.

“It was a very painful thing, a disaster; they essentially fell prey to marketing," continues Tonella.

In theory, AI could transform time-intensive processes such as mortgages, but Tonella suggests that such processes should be streamlined at a more fundamental level for long-term sustainability rather than just using AI to carry on existing workloads.

“If an agent processes 400 pages of documents in a week instead of six months, that’s progress,” he notes. “But perhaps it should only be 10 pages in the first place.

“AI won’t reduce that complexity automatically. It takes effort, pattern recognition and strategic application to achieve sustainable simplification. The technology is ready – but the focus is still too short-term.”

For traditional banks, a holistic strategy combining cultural change, education, and careful experimentation – as demonstrated by ING – may prove essential in meeting today’s challenges: increasingly sophisticated AI-driven cyber threats, costly outages, mounting technological debt, and economic uncertainty.

---