How a leading bank reinvents IT incident management using AI

Summary

Sector: Banking

The challenge: A leading European bank was overwhelmed by a growing number of IT incidents across its complex technology estate, with IT teams spending hours manually figuring out the root cause of incidents. The Leadership team recognized that they needed a faster, simpler way to diagnose and manage these issues, and to set the foundations for AI-driven automation in future.

The solution: The bank used Agentic AI to cut IT incident diagnosis time significantly, improve reliability, and create a strong foundation for scaling AI-driven IT operations across the organization.

Key results:

80% reduction in root cause analysis time (from up to 60 minutes to under 10)
600+ hours of manual effort saved each month
More consistent, data-driven incident assessments
Strong foundations laid for future end-to-end incident automation

The challenge

Manual troubleshooting causing delays in fixing IT problems

As one of Europe’s top five banks, the client operates a vast, complex IT estate including servers, databases and cloud platforms. As this set up became more complex, IT problems increased sharply – particularly those related to virtual machines – with around 1,500 incidents reported every month.

Leadership recognized that traditional, manual methods could no longer keep pace with this growing IT complexity. They needed a faster, standardized, and more reliable approach that met strict governance, privacy, and security expectations.

Finding the cause of IT problems involved teams having to pull information from many different tools. This often took an hour per issue, creating inconsistent results that put heavy pressure on teams, slowed down fixes and affected service reliability.

The bank’s ambition went beyond finding faster ways to solve IT problems. The goal was to put in place AI that could help manage IT issues from start to finish, and, over time, automatically fix some of them – while meeting strict governance, privacy, and control requirements.

The solution

Using AI to manage IT incidents at scale

To address these issues, the bank partnered with Wavestone to design and deploy a modular, reusable, and scalable Agentic AI solution specifically designed to handle IT problems.

Starting with ‘virtual machines’, the system automatically pulls together real-time system health and performance data, and uses multiple AI Agents to quickly identify what went wrong and suggest next steps.

From the outset, the IT architecture solution was designed to be flexible and future-proof, so it could be reused across different systems and wouldn’t lock the bank into one technology provider.

The solution fits neatly into the bank’s existing ecosystem, with humans staying firmly in control of decisions.

Thanks to close collaboration and strong existing AI foundations, the solution was up and running in production in just three months (for the first use case: Virtual Machines).

The results

Faster fixes, more reliable systems, and a foundation for future automation

The impact of the agentic AI solution was immediate and measurable, cutting the time to find the cause of IT problems from up to an hour to ten minutes or less, an 80% reduction in diagnostic time.

This helped teams fix issues faster, improve system reliability, while freeing up operations and engineering teams to focus on higher value activities instead of firefighting.

The success of the project, and strong executive sponsorship, led the bank to expand the solution to more systems, creating a well-governed platform to scale AI across IT and support wider transformation.

By putting the right controls in place and building flexible AI tools that give clear visibility into systems, the bank has laid solid foundations to expand AI-driven incident management (AIOps*) and speed up its broader IT transformation.

* AIOps is the use of AI and machine learning to automate, analyze and optimize IT operations, enabling faster detection, diagnosis and resolution of issues across complex systems.