EXCLUSIVE: “From Philosophy to Production” – Marco Li Mandri, ING in ‘The Paytech Magazine’

ING’s Marco Li Mandri describes how the bank is putting its AI vision into practice

Over the past 18 months, Dutch banking giant ING has accelerated its digital transformation, pairing strong financial performance with large-scale investment in data platforms, automation – and AI. The group has reported multi-billion-euro annual profits while returning capital to shareholders – financial resilience that has created room to modernise infrastructure and scale digital innovation.

The latest AI capabilities now sit firmly at the centre of that transformation, not as a lab experiment, but as embedded capability across retail, operations and wholesale banking. Recent developments underline the shift from proof of concept to scaled deployment. More than 90 per cent of ING’s generative AI pilots have progressed into production environments, a high conversion rate in a sector where roughly two-thirds of AI proofs of concept fail to industrialise.

The bank has rolled out AI solutions across multiple markets, automated large elements of customer service and begun piloting with agentic AI in selected domains like voice agents and mortgages. At the same time, AI is being embedded in financial crime monitoring, know-your-customer (KYC) processes and internal engineering workflows, signalling enterprise-wide integration rather than isolated innovation.

It is this prioritised approach that separates ING from many of its peers.

“Last year, we measured how many of the pilots we started in generative AI made it into production, and that number is above 90 per cent,” says the bank’s Global Head of Advanced Analytics Strategy, Marco Li Mandri.

And he attributes that to prioritising projects that are ‘based on value’. In other words, ING has concentrated on domains where AI can deliver immediate customer or operational impact. Across the banking sector, AI investment has surged, but implementation maturity remains
uneven. A recent EY-Parthenon generative AI survey found that 77 per cent of banks have launched or soft-launched generative AI/genAI use cases, yet far fewer have scaled them meaningfully into production.

Governance complexity, fragmented data and organisational readiness continue to slow progress. ING, however, appears to be moving faster than the industry’s average implementation pace.

That acceleration could stem from ING’s entrepreneurial spirit, a larger technology war chest, or structural advantages in data architecture – or a combination of all three. But culture and operating philosophy undoubtedly play defining roles. Elsewhere in this issue (page 6), ING’s COO Marnix van Stiphout talks of the bank raising AI agents, each tasked with running critical operational functions under human supervision. It is a metaphor intended to capture ING’s production-led mindset: build agents, govern them and deploy them at scale.

Add to that a sustained focus on digital transformation – reskilling employees, centralising analytics platforms and embedding responsible AI frameworks.

Its internal experience is also shaping external sentiment. ING’s own 2026investment outlook identifies AI as a structural growth engine, capable of boosting productivity, attracting capital and offsetting labour shortages across the economy. That institutional conviction, rooted in hands-on deployment rather than abstract forecasting, has contributed to a notably bullish investor perspective on AI-enabled banking transformation.

Getting personal

Retail banking has been ING’s first proving ground, and hyper-personalisation sits at the forefront of it. ING has developed a global tooling layer that allows marketers to deliver highly tailored communications at scale.

“More than seven million customers globally received a personal message,” Li Mandri says, describing campaigns calibrated to behavioural data, product relevance and life-stage signals. The result? Measurable uplifts in satisfaction and engagement.

Credit decisioning provides another high-impact deployment, says Li Mandri. Machine learning models now support instant lending approvals in multiple markets, compressing wait times that stretched over days, into decisions delivered in seconds.

“We have machine learning models that now instantly provide loans,” he says. “Customers do not have to wait.”

Contact centres have been equally fertile territory. ING was among the early European banks to deploy generative AI chatbots directly into retail service environments. Today, those  systems operate across most of ING’s retail markets, handling between 65 and 75 per cent of routine customer queries.

“They help reduce friction,” Li Mandri explains, freeing human agents to focus on complex or emotionally nuanced interactions.

“And we are working to make these chatbots smarter… with the ability to execute actions, but also moving into voice,” he adds.

Voice agents represent the next interface layer – conversational systems capable not only of answering questions but also of resolving requests in real time.

“Voice bots will be able to understand what customers are asking, provide an answer, and also execute some of the actions already in that moment,” says Li Mandri.

Beyond retail, ING is embedding AI into operations and wholesale banking. Know-your-customer processes in wholesale banking – traditionally labour-intensive and document-heavy – are being augmented with AI, but also data extraction and summarisation tools are being used to improve front office productivity.

The same capabilities support sustainable finance structuring, where large datasets must be analysed to benchmark companies against their ESG (environmental, social and governance) peers.

Implementing AI, front to back

Front-office productivity is another emerging domain. Li Mandri says ING is testing AI tools that prepare client meeting briefs automatically, aggregating financial data, prior interactions and market context so that relationship managers can focus on advisory depth rather than administrative preparation. KYC, however, remains one of the most strategically critical battlegrounds. Anti-money-laundering systems are being re-engineered through a blend of machine learning and genAI. Machine learning flags suspicious transactions with greater precision; generative systems then assist analysts by extracting, summarising and contextualising case data.

“It’s about improving the efficiency… without compromising risk,” says Li Mandri.

Digital transformation is also reshaping ING’s engineering backbone. More than 5,000 software engineers now use AI as a peer-programming tool, augmenting coding productivity, accelerating testing cycles and shortening time-to-market for new digital services.

“It’s very well received,” Li Mandri notes.

Yet perhaps the most structurally transformative layer sits within agentic AI (agents set up to perform specific tasks to a set of instructions), which are now in pilot phase. Mortgage processing has been selected as the initial focus for this. Agents augmenting human underwriters by extracting data, validating policy compliance and generating documentation will shorten approval timelines, ‘potentially to within a day’, says Li Mandri. But human advisory roles
remain intact, he insists. “

Advice is still human and very important,” he stresses.

Looking ahead, ING’s roadmap centres on scaling AI across more domains and pilots withagentic AI, supported by a centralised analytics platform and workforce AI-fluency programmes. Li Mandri’s mandate is to build, deploy and govern those agents, ensuring their outputs remain aligned to customer value with augmentation over replacement, governance not opacity. In an industry still learning how to maximise the potential of AI, ING’s progress suggests the idea is not just conceptual. The AI fields are already in cultivation.

This article was published in The Paytech Magazine Issue #18, Page 29-30

The post EXCLUSIVE: “From Philosophy to Production” – Marco Li Mandri, ING in ‘The Paytech Magazine’ appeared first on FF News | Fintech Finance.