ING: Turning AI Pilots Into Production

In this conversation, Marco Li Mandri of ING offers a concrete, execution-focused view of what successful AI adoption actually looks like inside a global bank. While many organisations remain stuck in pilot mode, ING has taken a markedly different path. Over the past year, more than 90% of ING’s generative AI pilots made it into production — a figure that immediately sets the bank apart.

Li Mandri attributes this success to a disciplined focus on high-value priorities, regardless of complexity. Rather than spreading AI experiments thinly across the organisation, ING concentrated its efforts where impact would be measurable and meaningful. One of the clearest examples is in retail banking, where the bank developed a hyper-personalisation tool enabling marketers worldwide to create highly tailored campaigns. The result: more than 4.6 million customers received messages personalised to their individual needs, preferences, and behaviours — at scale.

Credit decisioning represents another major breakthrough. ING deployed machine-learning models that allow loans to be approved instantly, significantly improving customer experience while maintaining robust risk controls. Faster access to credit, Li Mandri explains, is one of the most tangible ways AI can improve people’s financial lives.

Customer service has also been transformed. ING was among the first banks to launch customer-facing generative AI chatbots in production. These are now live across most retail countries, automating 65–75% of basic customer queries and reducing friction across everyday interactions. Importantly, this is not the end state. ING is now making these assistants smarter — enabling them to execute actions, explore voice interfaces, and eventually operate as more capable digital agents.

That ambition extends into agentic AI, which ING is approaching with caution and clarity. The bank has chosen mortgages as a starting point — one of the most complex, regulated, and data-heavy products. Here, agentic AI is used to augment human decision-making: extracting data, checking policy compliance, and preparing documentation, while humans remain firmly in control.

In wholesale banking, generative AI is already improving efficiency in KYC processes through automated data extraction, and supporting front-office teams in structuring sustainable finance deals that require extensive benchmarking and analysis. Similar tools are now being tested to help relationship managers prepare meeting briefs, freeing time for higher-value client interaction.

Operations and technology are equally central. ING is using a blend of ML and GenAI to improve AML effectiveness, focusing attention on true risk while maintaining — or increasing — control standards. Meanwhile, more than 5,000 software engineers now use AI as a peer-programming assistant, reporting higher productivity and faster time to market.

Li Mandri’s message is pragmatic: AI delivers value when it moves beyond experimentation and into daily work. ING’s focus is not on hype, but on execution — turning AI into a reliable engine of productivity, scale, and customer impact.

The post ING: Turning AI Pilots Into Production appeared first on FF News | Fintech Finance.