By now it’s fair to say most financial institutions have an AI strategy. Across the financial services industry, you’ll find artificial intelligence used for chatbots, fraud scoring, and document processing, all of which have accessible data and minimal risk of breaking anything.
But while these are useful applications, the credit union’s core systems remain frozen and outside the reach of common AI tools. This stops any AI strategy dead at the back-office door, leaving a widening gap between where financial institutions are willing to innovate and where their actual business logic lives. It’s the equivalent of painting the front of the house but not fixing the cracks in the foundation.
This gap is not a law of physics, but is the result of choices that come from misunderstanding which obstacles actually matter. Leaders overvalue the risks of AI touching the core or data governance while at the same time underestimating more pressing issues. For example, the people who understand core systems like IBM i are retiring. That’s not a risk; it’s a certainty, and by 2030 it’s not going to be a problem that can get kicked down the road.
Any AI strategy that sticks exclusively to the front office is spending the credit union’s modernization resources everywhere except where institutional knowledge is accepting its gold watch and riding off into the sunset. It’s time to fix the AI gap once and for all—and in the process save the foundational systems that keep everything running behind the scenes.
An understandable decision
You can’t blame financial organizations for being reluctant to modify the decades-old systems they use to clear transactions. Banking leaders are absolutely being rational when they don’t want to mess with the back-office infrastructure that powers their organizations.
“If it ain’t broke, don’t fix it” absolutely applies at a credit union, where breaking a working core is unthinkable. To make matters worse, the people who understand how that core works are few and aging, while the new AI talent the credit union just hired are not familiar with the platforms that keep the core running. Data governance at a credit union is no laughing matter, and exposing systems of record to outside services like AI platform-holders is a legitimate worry. Executives earn their credibility by taking issues like these seriously rather than risking their key systems in the name of innovation for innovation’s sake.
But even if the choice to keep AI out of the back office is understandable, it still creates two ticking clocks:
- The retiring developers who understand the core will take their undocumented expertise with them, creating a serious institutional knowledge gap.
- As the front office gets faster and the back office stays static, operational drag will lead to slower product launches and integration backlogs as compliance work is done by hand.
The gap between the front and back office is not neutral or static, but will compound (and worsen) over time. And while the second issue can be overcome by working to catch the back office up with the front, the loss of institutional knowledge is potentially much more serious.
The outgoing generation of experts who keep the cores running for the banking industry will either pass their expertise on, or that expertise will be lost. And while incoming professionals can rediscover these best practices on their own, the operational lag this discovery process creates could lead to considerable losses in efficiency.
Risk assessment reversed
It may seem counterintuitive, but what all this means is that financial institutions have their risk assessment backwards. Leaders are treating data governance and core stability as their absolute north star, and the loss of core expertise as an HR issue outside the scope of AI implementation. And that’s completely backwards.
Governance is a well-understood discipline the banking industry already knows how to manage, while the looming knowledge loss is, if not irreversible, an inevitable source of drag and inefficiency. When the last person to understand a system leaves, rebuilding their knowledge from scratch is a much harder task than ensuring governance.
Focusing AI implementation exclusively on the front office only makes the problem worse. Every quarter spent locking modernization efforts away from the core is a quarter closer to the cliff of absolute knowledge loss. So while back-office modernization has its risks, they’re risks credit unions must be willing to take to close the gap and preserve the precious institutional knowledge that will keep the core running smoothly into the future.
Closing the gap responsibly
The good news is that addressing these issues means holding to principles any responsible practitioner would endorse. Preventing the widespread loss of institutional knowledge and reducing drag between the front and back office requires reasonable, rational best practices that can easily leverage AI without causing undue risk. All of the steps listed below should feel familiar and safe to credit union leaders, and should be handled in-house rather than be traceable to a vendor.
- Don’t rip-and-replace where you can augment systems of record
- Keep data inside governed boundaries
- Treat documentation as a deliverable, not as an afterthought
- Direction of travel matters more than a procurement list or method
Most importantly: Capture institutional knowledge before it retires. The professionals who still keep banking systems operational are available and likely eager to share their expertise with the incoming generation. Urge these veterans to engage in mentorship while also asking them to commit their information to trackable, traceable systems that can be leveraged and built upon for years to come.
Bringing it together
A credit union can have an AI roadmap, a head of AI, and a dozen pilots in the front office. Across the financial institutions we’ve worked with, that’s the usual situation. But it’s just not the right approach if the systems that actually run the credit union haven’t been touched. The measure of a successful AI strategy isn’t how many front-office pilots are running, but whether key systems are also benefiting from innovation. And one of the best ways to leverage AI is to help preserve the human expertise that keeps systems functioning and thriving.
