Artificial intelligence is no longer an emerging technology for the financial sector. Across Dutch banks, AI is already improving customer service, lending, fraud detection and operational efficiency. As organisations move beyond pilot projects, however, they are discovering that successful AI implementation depends on far more than selecting the latest technology.
In highly regulated environments such as financial services, implementing AI means integrating it into existing systems, business processes and governance frameworks while maintaining security, transparency and human oversight. From AI-assisted mortgage applications to intelligent customer support and compliance monitoring, banks such as ING, ABN AMRO and Rabobank are demonstrating that the real challenge lies not in adopting artificial intelligence, but in embedding it responsibly into complex organisations.
Their experiences offer valuable lessons for business leaders. They also highlight a broader reality: as AI adoption accelerates, long-term success will be determined by implementation capability as much as technological innovation.
AI implementation starts with a business problem, not an AI model
Artificial intelligence is often associated with the technology itself. In reality, successful AI implementation begins much earlier, with identifying a business challenge that technology can meaningfully solve.
This is evident across the Dutch banking sector. Rather than deploying AI for the sake of innovation, banks are applying it to specific operational challenges where efficiency, consistency and customer experience can be improved. ING, for example, has introduced an AI-powered mortgage assistant that reviews application documents, highlights important information and recommends next steps before passing the case to a mortgage adviser. Rather than spending valuable time reviewing paperwork, advisers can focus on complex applications and supporting customers throughout the mortgage journey. Customers benefit from faster processing while employees remain responsible for every final lending decision. Importantly, the AI does not make lending decisions independently.
ABN AMRO has adopted a similarly practical approach. AI supports relationship managers by preparing lending documentation and summarising customer conversations, reducing administrative work and allowing advisers to focus on higher-value discussions with clients. Elsewhere within the bank, AI assists customers through digital service channels by answering common questions and guiding routine banking processes.
Although the applications differ, they share the same starting point. The objective is not to implement artificial intelligence itself, but to improve an existing process. The technology follows the business need rather than defining it.
As ING Retail Banking Head Pınar Abay explains, “Our sector remains a sector of relationships.” Rather than replacing human interaction, AI is being introduced to strengthen the way employees support customers by reducing repetitive work and allowing more time for meaningful conversations.
AI succeeds when it becomes part of the organisation
Identifying the right use case is only the first step. The real challenge begins when organisations move from experimentation to implementation. Successful AI must become part of existing systems, business processes and daily workflows rather than operating as a standalone tool.
Unlike traditional software, AI needs to work with the systems an organisation already uses. That means connecting it to reliable data, existing platforms and everyday business processes. In banking, where older systems often operate alongside newer technologies, making everything work together is often the biggest part of an AI implementation.
Dutch banks increasingly demonstrate that successful implementation is less about developing sophisticated models and more about embedding AI into the way employees work. At ING, for example, AI analyses mortgage applications, summarises complex cases and recommends next steps before passing its findings to a mortgage adviser. At ABN AMRO, AI prepares lending documentation and summarises customer conversations, allowing relationship managers to spend less time on administration and more time advising clients.
These examples illustrate a broader shift taking place across the financial sector. As ABN AMRO states, “AI is not a side experiment but a core capability that is reshaping how we bank every day.” Successful implementation therefore depends not on the AI model alone, but on how effectively it is integrated into existing systems, workflows and day-to-day operations.
Human oversight builds trust in AI implementation
Integrating AI into daily operations is only part of the implementation journey. Organisations must also ensure employees, customers and regulators can trust the decisions AI helps produce, particularly in processes with significant financial or personal consequences. Transparency is therefore essential, helping employees, customers and regulators understand how AI supports decisions while keeping human accountability clear.
Rather than replacing professional judgement, Dutch banks consistently position AI as a decision-support tool. ING’s implementation of AI-assisted mortgage processing reflects this philosophy. As the bank explains, “We decided to have humans work alongside AI agents.” While AI accelerates the analysis of mortgage applications, professional judgement remains central throughout the process.
This commitment to human oversight is made explicit in ING’s approach: “An ING employee always remains responsible for the assessment and makes the final decision on every mortgage application.” AI supports faster, more consistent decision-making, but accountability remains with experienced professionals.
ABN AMRO has adopted a similarly measured approach to scaling AI, stating that “We are moving forward with AI responsibly, not too fast, not too slow.” Governance and oversight are therefore treated as part of implementation rather than an afterthought. Readers interested in how regulation and AI governance are reshaping financial institutions can explore our related article, “The ESG–Compliance–AI Convergence: Why Your Next Risk Model Needs an Engineering Team, Not Just a Data Team,” which examines these developments in greater detail.
Ultimately, successful AI implementation is built on trust. Human oversight, clear accountability and responsible deployment enable organisations to scale AI confidently while maintaining the confidence of employees, customers and regulators.
Why AI implementation changes the expertise organisations need
As AI becomes embedded across banking operations, implementation increasingly depends on multidisciplinary teams that combine engineering expertise with business understanding.
Developing an AI model is only one part of the equation. Delivering AI within production environments requires Data Engineers to prepare and manage reliable data, Platform Engineers to build scalable infrastructure, AI and Machine Learning Engineers to deploy and monitor models, and governance specialists to ensure systems remain compliant throughout their lifecycle.
This reflects a broader shift in digital transformation. AI implementation is no longer viewed as an isolated innovation project. It has become a long-term organisational capability that requires continuous development, maintenance and oversight.
For organisations scaling AI initiatives, the challenge is often no longer access to technology. Instead, success increasingly depends on having the engineering capability to deliver AI reliably while balancing innovation with security, governance and operational resilience.
Key Implementation Takeaways
Although each bank has implemented AI differently, several common principles emerge. Together, these lessons provide a practical framework for organisations looking to scale AI successfully.
1. Start with business value, not technology
- Successful AI starts with a clearly defined business challenge. Across Dutch banking, AI creates value because it improves mortgage processing, customer service and operational efficiency, rather than being implemented simply because the technology exists.
2. Redesign workflows instead of replacing people
- Successful implementation changes how employees work rather than replacing them. By reducing repetitive administration, AI enables professionals to focus on complex decision-making, customer relationships and higher-value work. As ING Retail Banking Head Pınar Abay explains, “Our sector remains a sector of relationships.”
3. Build governance alongside implementation
- Governance should be built into AI from the start. Human oversight enables organisations to scale AI responsibly while maintaining accountability. As ING states, “An ING employee always remains responsible for the assessment and makes the final decision on every mortgage application.” Likewise, ABN AMRO notes that “We are moving forward with AI responsibly, not too fast, not too slow.”
4. Treat implementation as an engineering challenge
- Long-term value depends on strong engineering foundations, reliable data and seamless integration with existing systems. As ABN AMRO states, “AI is not a side experiment but a core capability that is reshaping how we bank every day.”
5. Invest in capability, not just technology
- Successful AI implementation depends as much on people as technology. Organisations need engineers, governance specialists and business experts to deploy, maintain and continuously improve AI systems.
While AI technologies will continue to evolve, the principles behind successful implementation remain remarkably consistent. Organisations that focus on business value, integration and people are best positioned to scale AI successfully.
AI implementation is an organisational capability
Artificial intelligence is reshaping the way organisations operate, but the experiences of Dutch banks demonstrate that successful implementation is about far more than deploying new technology. The greatest gains come when AI is integrated into existing processes, supported by strong engineering foundations and governed through clear human oversight. Ultimately, as ING observes, “For customers, what matters is not the technology itself, but getting clear, reliable answers they can trust.” That simple principle captures why successful AI implementation is measured not by the sophistication of the technology itself, but by the value it creates for customers and the confidence it inspires in the people using it.
For business leaders, this means viewing AI as an organisational capability rather than a standalone technology investment. Building that capability depends on engineers, governance specialists and business experts who can translate AI into reliable, real-world solutions.
As more organisations move from experimentation to enterprise-wide adoption, the question will no longer be whether AI can create value. Instead, it will be whether organisations have the expertise to implement it successfully, responsibly and at scale.
