Photo Jacqueline Wild © Kyndryl
By Jacqueline Wild, Managing Director of Kyndryl for Switzerland and Austria
In the financial sector, artificial intelligence is seen as one of the key levers for boosting efficiency, automating processes and improving decision-making. Banks, insurance companies and financial services providers are already testing AI in many areas: customer service, fraud detection, document processing, risk analysis and software development. The need to act is pressing. Margins are narrowing, customer expectations are rising, regulatory requirements are multiplying and cyber risks are becoming more complex.
Yet operational AI deployment still often falls short of expectations. Many initiatives operate within specific teams or in controlled test environments, without managing to integrate into the company’s critical business processes. The real AI challenge therefore lies not only in the technology. It lies in the ability to make AI reliable, integrated and scalable, so that it creates genuine added value for the business.
The latest Kyndryl Readiness Report[1] clearly illustrates this paradox: AI spending rose by an average of 33% compared with the previous year. As many as 54% already report positive returns on investment from AI. Yet 62% have still not managed to move beyond the pilot stage.
In the financial sector, discussions often focus on models, algorithms and new applications. By contrast, the underlying IT infrastructure, although often decisive, remains less visible. Many institutions still rely on core systems that have been built up over the years. Data is stored in different silos, interfaces are complex, processes are legacy-based and regulatory requirements are high.
This is a major obstacle for AI. Models need consistent, up-to-date and easily accessible data. They must be able to integrate into existing processes. Outputs must be traceable. And sensitive information must be protected against uncontrolled processing. When data is fragmented or core systems cannot be integrated flexibly, AI remains confined to isolated experiments.
Here too, there is a significant gap between ambition and reality. According to the Readiness Report, 90% of organisations believe they have the tools and processes needed to test and deploy new ideas quickly. Yet more than half believe their own technology environment is holding back innovation. Last year, 90% of business leaders already considered their IT infrastructure to be “best in class”, but only 39% judged it fit for future challenges.
For Swiss and European financial institutions, the challenge is particularly complex. They must innovate without compromising trust, stability or regulatory compliance. An AI pilot can be launched relatively quickly. Integrating it into core business processes is an entirely different matter.
This raises questions that go far beyond the simple choice of a technology solution: where do the data come from? Who can verify the results? How can errors be detected? Which systems are connected? How can data protection, supervisory obligations and internal controls be ensured at the same time? How can directives and controls be embedded in automated, machine-readable processes so that they are applied consistently? And who bears responsibility when AI prepares processes or influences decisions?
IT is therefore becoming a strategic issue for executive management and boards of directors. Data architecture, resilience, integration and security determine whether AI can truly drive cost reduction, better processes and new business models. Technology is therefore no longer just an operational infrastructure; it is becoming a genuine lever for steering the business.
Cybersecurity plays a central role. Financial institutions have always been among the most attractive targets for cyberattacks. AI is changing this risk landscape on both sides.
It can help detect anomalies more quickly, automate security processes and identify threats earlier. At the same time, cybercriminals can use AI to make phishing campaigns, social engineering or malware more sophisticated.
The Kyndryl Readiness Report identifies cybersecurity as the most common area of AI application. That is hardly surprising: in a sector built on trust and availability, cyber resilience is a prerequisite for any digital transformation.
For financial institutions, this means that AI readiness and cyber-threat readiness can no longer be separated. Any strategy aimed at deploying AI at scale must also include data-access management, identity and authorisation models, the security architecture and the ability to respond in an emergency. Otherwise, new technologies will not only improve efficiency, but also expand the attack surface.
Beyond infrastructure and security, employees also play a decisive role in success. AI will only deliver results if employees understand it, accept it, trust it and are able to use it effectively. According to the Readiness Report, 87% of executives believe AI will profoundly transform the jobs in their company over the next 12 months. Yet only 29% believe their workforce is sufficiently prepared to use AI successfully in its day-to-day work.
This issue is particularly important in the financial sector. Many processes rely heavily on human expertise and are sensitive to risk. Employees must not only master the tools, but also be able to analyse results critically, assess data quality and identify risks. This increasingly requires bridge roles – such as Human Systems Architects – who define how employees, decision-making processes and AI systems interact. Introducing AI is therefore not just an IT project. It is a transformation project that requires continuous training, clear guidelines and trust.
The financial sector has fully grasped the importance of AI. Investment is rising and the first results are visible. But the next phase will be more demanding: the goal is no longer to multiply pilot projects, but to turn AI investments into economically measurable added value.
That requires modern data architectures, integrable core systems, solid governance, robust cyber resilience and employees capable of using AI responsibly. Institutions that put these foundations in place will be able to integrate AI into their operational value chain. They will be able to automate processes, identify risks earlier and serve clients faster.
The real question is therefore not which bank is experimenting with the largest number of AI applications. It is which institutions are able to evolve their technological and organisational foundations so that artificial intelligence becomes reliable, transparent and value-creating. In a trust-based sector, what will make the difference is not the most ambitious AI roadmap, but the ability to implement it on a solid footing and scale it successfully.
[1] https://www.kyndryl.com/ie/en/insights/readiness-report-2025
Find all our Business Culture articles
