
Unlocking mainframe data for the AI age
As AI revolutionises business operations, organisations are discovering value in vast stores of mainframe data. But unlocking this potential requires a shift in how companies approach mission-critical systems and embrace emerging technologies
Fraudsters were busier than ever in 2024. According to credit reporting agency TransUnion, they launched 20% more digital account takeover attempts than the year before, costing consumers a median of $1,747 in losses across 18 countries.
For banks and financial institutions, the challenge of detecting and preventing this fraud is compounded by a critical obstacle: storehouses of transaction data locked away in disparate systems. Financial analysts could uncover patterns indicative of fraud in that data — if only they could access it.
They’re not alone. Across industries, professionals are grappling with the same challenge: extracting meaningful insights from vast repositories of historical data trapped in aging systems, often on mainframes.

Teasing out individual strands from what amounts to a huge spaghetti bowl of data on demand represents one of the great IT challenges facing businesses today. That’s the opinion of Michael Curry, president of Data Modernisation at Rocket Software, a global leader in IT modernisation, based in Massachusetts.
“If you think about the average bank or insurance company or government entity, so much of what happens in those environments is still happening on mainframes,” says Michael Curry. “To truly understand the current state of your business at any given time requires the ability to see what’s going on in those systems.” That means accessing data and, increasingly, accessing it with artificial intelligence (AI).
Therein lies a challenge: these mainframes, while robust and reliable, don’t always lend themselves to organising data in a way that enables organisations to harness it for AI-driven applications.
The data accessibility challenge
For modern businesses, the core mission surrounding data has evolved from accumulation to deriving meaningful strategic insight. AI can help but for accurate results AI needs data; the more, the better, according to Stewart Bond, vice president of research at IT advisory IDC, also headquartered in Massachusetts. “It is tough to be successful with AI if you are only using some of your data,” he says.
The numbers tell a stark story. Despite handling 70% of global transactions, according to IBM, mainframes remain an enigma. A recent study by Foundry reveals that only 28% of organisations leverage their mainframe data effectively, with 76% of leaders admitting they struggle to access and understand this critical information.
Only 28% of organisations leverage their mainframe data effectively, with 76% of leaders admitting they struggle to access and understand this critical information.”
Why not simply replace legacy IT systems, perhaps starting with rewriting older software? Such a drastic move rarely makes economic sense and introduces additional risks, explains Michael Curry. That’s because mainframes remain the most efficient, secure and reliable environment for many critical applications.
Instead, organisations are taking a different approach to making sense of their data — one that lets the application and its hardware stay in place.
Transforming data management
According to Michael Curry and Stewart Bond, the data access challenge can be solved with comprehensive data mapping. That means deploying systems that can catalogue the data, where it resides and how it’s all related so that other systems can make use of it.
Stewart Bond points out that cataloguing data facilitates not only analytics but security as well. “Data intelligence tells you where your data is,” he says. “There could be a comments field sitting inside a database that has a credit card number in it.” Since comment fields are not typical places to store sensitive financial information, a credit card number in one could remain widely exposed. Understanding where data is located allows companies to apply the appropriate security and access policies.
Data intelligence is already yielding real-world results. Michael Curry points to a US-located bank customer of Rocket Software that successfully mapped its entire data landscape across cloud and on-premises environments, including mainframes. The process enabled the bank to achieve several crucial objectives: ensuring regulatory compliance by tracking data lineage and calculation methods, improving customer service with real-time account information and providing more effective financial advice based on a complete view of customer holdings and account activities.
The hybrid future
Looking ahead, Stewart Bond envisions a seamless data ecosystem where artificial intelligence acts as a universal translator. AI-powered automation enabled by comprehensive data mapping could increasingly handle routine data tasks. That, in turn, could free up employees to focus on higher-level strategic analysis and decision-making, letting them contribute more value to their organisations — and customers.
For financial security teams working to protect customers from account takeover attempts, such advances (in data ecosystems) would be game-changing.”
For financial security teams working to protect customers from account takeover attempts, such advances would be game-changing. Rather than laboriously combing through disconnected data sources, they could instantly access comprehensive transaction histories across all systems, identifying suspicious patterns in real-time. Advanced AI could then proactively suggest tailored security protocols based on unique financial behavior.
“Soon, the origin of data — whether on a mainframe or in the cloud — will no longer matter. What will matter is the intelligence we can extract, the decisions we can make and the value we can create.” As Michael Curry sees it, the technology to unlock a seamless data fabric across all systems is nearly here. “We are on the cusp of achieving a unified understanding across environments, and I am confident we’ll get there very soon.”
