ING receives roughly 85,000 customer interactions every week through phone calls and online chat in the Netherlands, and its traditional scripted chatbot could resolve only 40–45 percent of them — leaving nearly 16,500 customers per week waiting for live agents. To close that gap, the bank worked with McKinsey & Company and QuantumBlack on a seven-week joint project that produced a generative AI chatbot, deployed into ING's mobile support chat in September 2023.
"By prioritizing real customer needs, security, and ease of use, ING was able to develop a bespoke customer support tool that gives users the best possible experience," said Marnix van Stiphout, ING Chief Operating Officer.
The initiative grew out of pressure on banks to deliver faster, more personalized support without compromising compliance, security, or customer trust. ING treated the deployment as much a governance exercise as a technology one, involving risk and compliance stakeholders throughout development to keep the chatbot within strict safety boundaries.
The joint team began with a deep assessment of ING's existing chatbot infrastructure, then built a sophisticated multi-step response pipeline that combines knowledge retrieval from multiple internal data sources with AI-driven ranking of potential responses by relevance and usefulness. The system includes disambiguation logic that presents several possible answers when a customer's intent is unclear, and custom guardrails built with tools from QuantumBlack Labs prevent it from offering sensitive financial guidance such as mortgage advice and investment recommendations.
The chatbot was initially piloted with 10 percent of real ING customers in the Netherlands; since launch, thousands of customers have interacted with it, making it one of the first real-life customer-facing generative AI pilots in Europe. Within the first seven weeks of use it helped 20 percent more customers resolve their issues than the previous solution, reducing pressure on customer service call centers and shifting more customers from phone support to digital chat.
Customers experienced faster query resolution, more detailed and personalized responses, and improved overall satisfaction. ING has since used the project as a foundation for enterprise-wide AI adoption, investing in capability building across more than 50 support functions including Risk, Contact Center Operations, Analytics, and Technology teams.
ING and the McKinsey team aim to double the chatbot's performance in the near future and build a scalable model that can be extended to all ING countries. "This project has helped establish a solid technical foundation that puts ING at the forefront of gen AI applications within the banking industry," said Bahadir Yilmaz, ING Chief Analytics Officer.