88

Kintsugi for Banking: The Art of Modernizing What Already Works

Kintsugi is the Japanese art of repairing broken pottery with lacquer and powdered gold.
Rather than hiding the fractures or discarding the object entirely, the philosophy is to preserve what already has value and make the repaired connections part of its strength.

I’ve been thinking a lot about Kintsugi in the context of financial-services modernization. Because when we talk about transformation, the reaction is often to focus on replacement.

Replace the Core. Replace the LOS. Replace the CRM. Replace the legacy application.

But what if the systems themselves aren’t actually the problem?

What if the opportunity is in the fractures between them?

But are we approaching modernization in the right way? Because in many financial institutions, those systems are doing exactly what they were purchased to do. What we’ve seen is that the inefficiency comes from what’s happening around them.

And there’s some interesting evidence that the industry is beginning to see the same thing. The CCUA has identified modernization as a key priority for the credit union sector, specifically highlighting the need to address rising operational costs and evolving consumer expectations. The Bank of Canada’s 2026 Financial System Survey found that nearly all responding financial institutions are already using AI, but here’s the statistic that caught my attention:

58% identified integrating AI into their existing infrastructure and workflows as a challenge.

I think the word workflows matters. Because the next phase of modernization may not be about buying another system. It’s about improving how work moves between the systems we already have.

Take commercial lending, for example: the loan origination platform might work perfectly well. But what happens before, after and between interactions with that system? The flow typically looks something like this:

  • A relationship manager requests documentation.
  • Financial statements arrive.
  • Someone checks what’s missing.
  • Information gets entered somewhere else.
  • A credit package is assembled.
  • Risk reviews it.
  • An exception requires approval.
  • An email chain begins.
  • Another document needs updating.
  • Someone follows up.
  • Someone updates a spreadsheet.

Eventually, information makes its way back into the primary system, but none of those activities really justify replacing the lending platform. Collectively, however, they can consume enormous operational capacity.

And the same pattern exists in other processes:

  • Member onboarding and KYC
  • Commercial credit renewals
  • Mortgage exceptions
  • AML investigations
  • Fraud and disputes
  • Regulatory reporting
  • Treasury operations
  • Complaints and escalations
  • Internal approvals

We often describe these as technology problems. I think many of them are actually workflow problems. So perhaps one of the most useful modernization questions a financial institution can ask isn’t:

“Which system should we replace?”

It should be:

“Where does the work leave the system?”

  • Where are employees copying information?
  • Where are they coordinating work through email?
  • Where are spreadsheets effectively acting as workflow engines?
  • Where are experienced employees spending their time chasing documentation instead of exercising judgment?
  • Where does an exception suddenly turn a digital process into a manual one?
  • Where are two perfectly good systems creating manual work simply because they don’t communicate effectively?

These can seem like relatively small operational problems. But solving them can have an outsized impact:

→ Faster lending decisions → Greater capacity without proportional headcount growth → Better visibility → Fewer errors → Stronger audit capability → Better employee experiences → Better member experiences

I think this should also influence how we approach AI. The Bank of Canada’s research found that financial institutions generally see AI as a way to complete existing tasks faster, not as a replacement for human judgment, particularly where financial, legal and reputational consequences are significant. That feels like an important distinction. AI doesn’t necessarily make the lending decision; it extracts information from financial statements, summarizes the credit file, identifies missing documentation, surfaces the relevant policy, drafts an analyst briefing, or helps an investigator navigate a complex case.

The workflow governs the process. AI assists the people within it.

That, to me, is one of the more interesting opportunities in financial services right now. Not ripping out the systems financial institutions already trust, but rather making the work between them dramatically better.

Perhaps the next phase of financial-services modernization looks less like demolition and more like the art of Kintsugi.

  • Preserve what’s valuable
  • Identify the fractures
  • Strengthen the connections
  • Make the whole system work better because of them

As I prepare for the CCUA Financial Forum this fall, I’m particularly interested in hearing how credit unions are approaching this.

Where does the most operational friction exist in your organization today, inside your systems, or in the work happening between them?

Sources: Bank of Canada, 2026 Financial System Survey; Canadian Credit Union Association, Economic, Policy & Social Outlook.

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.

Recommended Articles

How node-wire Powers Integrations Inside formsflow.ai

A form is only the beginning of the workflow. What matters is where the submission goes next. A submission has to reach an EHR, a core banking system, an eligibility engine, or a case queue. The form is the easy part. The connections behind it are where automation projects become costly. Point-to-point integrations are built […]

Secure your agent-to-tool connections with open source solutions

node-wire + ToolHive provide the interoperability layer that enterprises need to adopt MCP at scale Enterprise AI has a quiet dependency problem. While the models get attention, the real value shows up when an agent can access the systems where the work actually happens. This could be an EHR, a core banking system, an identity […]

Don’t Let Your AI Agents Run Amok: Why You Need a BOAT

The Age of Agent Euphoria Organizations are racing to deploy autonomous AI agents capable of reasoning, planning, invoking tools, and executing business processes with minimal human oversight. The promise is compelling: a digital workforce that scales faster than any human team ever could. But as organizations move from deploying a handful of agents to hundreds […]