By Joey Rault, Chief Revenue Officer, Codat
For most of the past decade, banks have treated digital transformation as an aspirational destination: a multi-year program that ends with a new core, a better app, and a cleaner operating model.
That way of thinking is now behind us.
In 2026, the banks that win will turn AI into operating leverage and make that leverage visible at board level.
This is the shift I’m watching closely: operating leverage is moving from an internal efficiency goal to a board-level priority, and it will become a direct measure of whether a bank’s technology investments are paying off.
At the same time, the era of tools built on fragmented data is ending. Banks are moving away from isolated tools and toward horizontal platforms that support onboarding, credit, treasury, servicing, and operations from a unified data foundation.
These two trends are tightly linked. Together, they define what serious AI adoption in banking actually looks like. Based on Codat’s existing relationships with many of the world’s largest commercial banks, here is what we’re seeing as In and Out for 2026.
IN: Operating leverage as a board-level KPI
Banks have always cared about efficiency ratios and cost control.
Operating leverage is the next step. It’s not simply about cutting spend, it’s about scaling output and customer value without scaling headcount and complexity at the same rate.
In practice, this means smaller, more advanced teams supported by AI and cleaner tech stacks delivering a lower cost base. It also means boards start asking harder questions about what the bank gets in return for every dollar invested in technology.
Many banks are currently stuck in a frustrating middle ground:
- Relationship managers chase documents instead of deals.
- Credit teams lose days to manual workflows and repeated data entry.
- Operations teams build workarounds for systems that should talk to each other.
- Card and treasury teams lack actionable spend insights.
The result is that some of the highest-value people in a bank are doing some of the lowest-value work. Procedural friction consumes the time of even the sector’s most skilled relationship managers.
AI can remove a large portion of that friction, but only if it’s deployed in a way that changes how the bank operates end-to-end, not in pockets.
When AI is integrated properly, it becomes a multiplier. Admin work drops, cycle times shrink, and teams make decisions faster with better information. Customers feel the difference in speed, clarity, and confidence.
That is what operating leverage looks like in the real world. It shows up by winning more opportunities, driving more card volume, and preventing churn. And once boards can see that impact, operating leverage becomes a competitive moat.
OUT: Tool sprawl built on fragmented data
For the last couple of years, “AI strategy” in many banks has looked like a growing collection of tools.
One for meeting notes. One for document processing. One for customer service. One for policy search. One for credit memos. Each tool has a business case. Each tool can show value in isolation.
But when these tools are deployed on top of fragmented, inconsistent data, they create drag instead of leverage.
That’s why investment in isolated AI tools will decline in 2026, as banks move toward horizontal platforms built on a unified data foundation.
Banks are realizing that AI does not work like traditional software procurement. Its performance is dictated less by features and more by the quality, consistency, and governance of the data feeding it.
When even the best AI sits on top of data that’s conflicting, unclear, or incomplete, it sends the system into a tizzy. Inconsistent customer records, duplicated financial data, conflicting risk flags, and unclear data lineage could all result in bad AI outcomes.
In banking, trust is everything. Customers don’t just want an answer. They want to know why. They want to trust the decision, and they want a human who can explain it clearly.
That is why explainability and data lineage are the price of entry. If a system offers a recommendation, banks need to know the source of the data, how it was prepared, and what processed it.
And if you cannot explain a result in clear language, it will not hold up in front of a customer.
Fragmented tools make this harder. Horizontal platforms make it possible.
If you’re a bank executive or board member looking at 2026 planning, the question isn’t “should we use AI?” The question is: “are we building leverage, or accumulating tools?”
AI cannot compensate for poor data. It only magnifies it. In our work connecting banks to accounting, commerce, and payments data, we see this pattern repeatedly: the institutions that move fastest are those that solved their data fragmentation problem first. Clean, permissioned, real-time data is the foundation of sustainable automation and insight.
The future belongs to banks that treat AI like infrastructure, not innovation theatre: horizontal, integrated, measurable, and built on data their boards can trust.
