My Advice to Banks on AI: Michael Curry of Rocket Software

I spoke with Michael Curry, President of Data Modernisation at Rocket Software, about the critical AI and data decisions facing bank executives today. Michael leads efforts helping global enterprises unlock the full value of their data across mainframe to hybrid cloud environments.

Over to you Michael - my questions are in bold:

Can you give us an introduction to you and an overview of your organisation?

I'm Michael Curry, president of the data modernisation business unit at Rocket Software. In my role, I focus on helping global enterprises unlock the full value of their data by modernising how it's accessed, governed, and made available for usage in AI-models, cross core systems and the cloud. Rocket Software is a global leader in modernisation and the trusted partner to many of the world's most missioncritical organisations. We help businesses maximise the value of their data, applications, and infrastructure so they can deliver the essential services that power our modern world. From the mainframe to hybrid cloud environments, our mission is to make modernisation reliable, secure, and achievable for every organisation we serve.

If you were advising a bank CEO today, what would you say is the single biggest mistake they're making with data and AI?

The biggest mistake banks are currently making is treating their AI program as if it is something separate from the rest of their IT plans and budgets. Agentic AI will be embedded in every process and application within the business over the next few years. The only way that will work is to embed AI skills and tools into every function. This is especially relevant to data. Making clean, trusted data available to agentic AI across the enterprise is critical to the success of AI programs. This is because data is the fuel of AI. Of course, this requires a framework to protect and govern the data and its use across these new agentic interactions.

What's one AI or data capability banks should prioritise in the next 12–18 months, and why?

Banks should prioritise building a unified, AIready data foundation that connects mainframe, cloud, and distributed systems in real time. This single move accelerates everything by providing a consistent data substrate for agentic AI. This substrate needs to provide the governance around the data as it is used within the agentic interactions. It also requires a semantic translation function to map the complex underlying technical data up to business terminology that business users understand. Once in place, this substrate can be used for new AI use cases ranging from fraud detection to regulatory reporting to deploying copilots, because models run on complete, trusted data.

Where do you see banks overestimating AI, and where are they underestimating it?

This isn't specific to banking, but many companies are overestimating what AI can currently do. For example, some companies see the groundbreaking capability AI has to understand, write, and translate code, and they immediately assume they can use this to easily refactor core systems within their environment. Unfortunately, the complexities of these applications far outstrip the ability of code translation alone to solve the problem.

What does "good" actually look like when AI and data are working well inside a bank?

When AI and data are working well together, banks operate with more resilience, speed, and control. Incidents are resolved quickly using naturallanguage interactions; analytics and LLMs run on governed, realtime data; migrations happen with zero downtime; and security is built in at every layer with Zero Trust access. Operations get simpler, costs drop, uptime improves, and compliance risk goes down.

What's the hardest AI or data decision bank executives are avoiding right now, and why?

70-90 percent of enterprise data is estimated to be unstructured {Gartner}. In order for AI-models to be fully effective, banks need to be able to access unstructured data in addition to structured. However, there are major challenges with accessing this type of data (e.g. gaining insights from customer onboarding documents that have sensitive data). While generative AI can bring enormous benefits in the ability to parse, understand, and analyse this unstructured data, bridging this data to AI without inadvertently exposing it to new risks is the ongoing challenge.

Thank you Michael! You can connect with Michael on his LinkedIn Profile and find out more about the company at rocketsoftware.com.