Why the next step may be connecting systems, decisions and work, not simply adding more technology
Credit unions don’t typically have a shortage of systems. They have systems for the core. Lending. CRM. Payments. Fraud. Documents. Digital banking. Finance. Data.
Individually, many of them do exactly what they were designed to do. But what coordinates the work that has to move between them?
A member’s experience doesn’t recognize the boundaries between applications. Neither does a commercial loan, a fraud investigation, a payment exception, a regulatory change or a complex service request. The work simply needs to move from one person, decision and system to the next.
And that creates a part of the operating model that is easy to overlook. I think of it as the operational layer: the layer that connects the specialist systems already doing important jobs and answers a different set of questions. It isn’t another core platform, and it isn’t a replacement for those systems.
What needs to happen next? Who owns it? What information do they need? What decision needs to be made? What happens if the case doesn’t follow the standard path? When does a human need to get involved? And can we prove what happened afterwards?
Those aren’t really system-of-record questions. They are systems-of-work questions, and increasingly, they may be where some of the most important modernization opportunities sit.
Systems of record know things. Operating models make things happen.
A core banking platform knows the account. The lending system knows the loan. The CRM knows the relationship. The fraud platform knows the alert. The document repository knows where the file is. And the data platform may know considerably more than any one of them.
But knowing something and coordinating what happens because of it are different jobs.
Consider a decision that requires information from three systems, approval from two people, supporting documentation, a policy check and an exception process if certain criteria are met. Each system might function perfectly, but the process can still be inefficient. That’s the distinction.
For years, modernization has understandably focused heavily on the applications themselves. But as financial institutions accumulate more capable technology, the challenge increasingly shifts toward how those technologies participate in one end-to-end operating model.
A lot is happening in Canadian financial services, making that question more urgent. The CCUA’s current priorities include payments modernization, consumer-driven banking and financial-crime prevention. Its latest federal submission also emphasizes growth, regulatory modernization and strengthening Canada’s response to fraud and scams. At the same time, the sector is facing consolidation, increased digital investment, and shifting competitive dynamics.
Each of those pressures ultimately creates more connections between systems, processes, data, decisions and people. So there’s a second question worth asking alongside “what technology do we need?”: what operating model do we need around the technology we already have?
A modern operational layer needs to do five things well
1. Connect
The first is fairly obvious. Systems need to exchange information. But integration isn’t valuable simply because two APIs can communicate. The value comes from what the connection enables: information available at the right point in a process. A decision triggered by an event elsewhere. A document automatically associated with the right case. A status change reflected across systems. An employee no longer acting as the bridge between applications.
In that sense, integration isn’t really plumbing. It is part of the business process. And this is becoming particularly relevant as Canadian financial services become more interconnected.
The CCUA, for example, recently highlighted the practical implications of consumer-driven banking for institutions managing legacy systems, mergers and third-party service arrangements. Those aren’t simply API challenges. They’re operating-model challenges.
2. Orchestrate
Connecting systems solves one problem. It doesn’t necessarily coordinate the work. That requires orchestration.
Someone applies. Something changes. A threshold is reached. A document arrives. A risk is identified. A payment fails. A policy condition is met.
What happens next?
A well-designed operating model should move work across departments and applications without requiring an employee to remember every subsequent step. That could mean triggering tasks, routing approvals, collecting documents, calling another system, waiting for a response, updating a case, escalating when an SLA is missed, or bringing a person into the process when judgment is required.
This is where the distinction between digitizing tasks and digitizing work becomes important. Automating one step might save minutes. Orchestrating the entire process can change how the organization operates.
3. Handle exceptions
This may be the most underappreciated requirement. Most processes are designed around what should happen. Financial institutions spend a remarkable amount of time dealing with what didn’t.
The borrower who doesn’t fit the standard criteria. The transaction that needs investigation. The member request that crosses three departments. The payment that doesn’t reconcile. The documentation that is incomplete. The approval that needs to go one level higher. The situation where policy requires judgment rather than a binary rule.
These aren’t edge cases in financial services. They’re part of the operating model, which means modernization can’t only improve the happy path. It needs somewhere for complex, non-linear work to live. That’s where case management becomes important.
Rather than forcing every situation through a rigid sequence, a case can bring together the people, information, tasks, documents, decisions and history surrounding an outcome. The question moves from “which system owns this?” to “who owns the outcome?” That’s a subtle but important shift.
4. Govern decisions
Financial institutions do more than move information. They make decisions.
Should this be approved? Does this meet policy? Does this need escalation? Which authority level is required? What information is missing? What happens next?
Some decisions are rules-based, some are data-driven, some require specialist expertise, and increasingly, some may be assisted by AI. A mature operational layer needs to accommodate all of them. The objective is to understand which decisions genuinely require people, and to keep them there.
Straightforward, repeatable decisions can often be automated. Complex ones can be routed to the right expertise. Policies can be incorporated into the process. Supporting information can be assembled before someone reviews it. And the decision itself can become part of the record.
That’s particularly important in a regulated environment, where efficiency matters. Explainable efficiency is much more useful.
5. Apply intelligence in context
This is where AI becomes more interesting, working as a capability inside the work itself rather than a standalone destination employees visit.
Summarize a complex case before an investigator reviews it. Extract information from documents arriving in different formats. Compare an application against policy. Identify missing information. Surface relevant historical cases. Suggest an appropriate next action. Assist with drafting correspondence. Identify anomalies requiring human attention. Or bring several analytical perspectives together before making a complex decision.
The possibilities are substantial. But context matters. AI needs to know what process it is participating in, what information is relevant, what policies apply and where its responsibility ends. Otherwise, we’ve simply created another tool outside the workflow.
The more interesting opportunity isn’t giving every employee an AI assistant. It’s making the process itself more intelligent.
This changes the modernization architecture
Once you look at operations this way, modernization starts to form a fairly logical progression.
Connect the information. Orchestrate the work. Manage the exceptions. Govern the decisions. Apply intelligence where it improves the outcome. Measure everything that happens.
None of those requires every existing platform to disappear. In fact, the opposite may often be true. Specialist systems should continue doing the jobs they are good at. The goal is to stop expecting any single one to run the entire institution.
A lending platform doesn’t need to become a CRM. A CRM doesn’t need to become a case-management system. A core doesn’t need to manage every operational exception. An AI model doesn’t need to become the operating model.
What matters is whether these capabilities can participate in one coherent flow of work.
And that matters to the member
Members don’t experience your architecture diagram. They experience its consequences. Whether they have to provide the same information twice. Whether someone can tell them where their request stands. Whether an employee has the information needed to help them. Whether something that should take hours takes days. Whether one department seems to know what another department is doing. Whether an exception feels like a normal part of doing business or sends the organization scrambling. Each of those experiences is shaped by an operating model.
The same is true internally. Employees experience whether systems help them complete the work or simply record pieces of it. Managers experience whether they can see where processes are slowing down. Risk teams experience whether controls exist inside the process or depend on people remembering them. Executives experience whether growth creates operating leverage or simply creates more administrative work.
Those are much bigger questions than any individual technology implementation.
Perhaps that’s the next phase of modernization
Not another race to accumulate technology, and not the assumption that everything old needs replacing. Instead, the focus can shift to the layer that allows the institution to behave like one connected organization, even when the underlying mix of systems is complex.
A layer where systems exchange information, work moves deliberately, exceptions are managed, decisions are governed, humans contribute where judgment matters, AI contributes where intelligence helps, and every important action leaves evidence behind.
Because ultimately, a credit union isn’t a collection of applications. It’s a collection of people making decisions and completing work on behalf of members.
The technology is there to make that easier. But the real modernization opportunity may be making sure it actually does!
Author Profile:
Moray Hickes is an Enterprise Technology Solutions leader with over two decades of experience helping organizations navigate complex technology systems and deliver high-value solutions across regulated industries.