By Steve Morgan, Senior Director of Industry Markets at Pega
Banks are entering 2026 with unprecedented momentum around AI and automation, yet many are discovering that technology is no longer the hard part. From generative and agentic AI to increasingly sophisticated workflow automation, the tools are not the limiting factor. What remains uncertain is whether banks are prepared for the human, operational and governance change these technologies demand.
After several years of experimentation, AI is moving into the core of banking operations. The challenge now is not innovation, but execution, specifically how institutions manage the scale of change AI introduces across jobs, workflows and decision-making.
Why Change Management is Becoming Crucial for Banks’ AI Plans
There is a huge, almost feverish focus on new technologies like agentic AI. But as banks seek to realize true value from the AI and automation wave, many are discovering the hardest part isn’t deploying the technology, but managing change.
This is not because of AI and automation replacing people, but how the impact of these technologies will change jobs and workflows. Before banks can experience any meaningful benefits, they still need to map, analyze and redesign how work gets done in order to understand how those processes will be transformed.
For many institutions, this work is only just beginning, even as different AI and automation projects are already in underway. The risk in 2026 is that change management programs will not be planned and executed properly, delaying the benefits of otherwise promising AI investments.
That is why a growing number of banks are starting to look at how change management itself can be enabled by AI tools that help map out processes and track how work evolves as automation is introduced.
Why 2026 Will Be a Bridge (not a Breakthrough) Year for Banking AI
The industry is benefiting from how rapidly agentic AI tools and platforms are expanding, especially those that have been designed and configured to help banks drive significant transformation. It has never been easier to build an agent, connect it to specific processes, automate steps, and track automation.
Yet the reality is that banking is a regulated industry built on low tolerance for error. Digital transformation initiatives rightly take time when they touch customer money, credit decisions, fraud management or regulatory reporting.
That is why most banks should expect the bulk of AI-driven benefits to materialise in 2027 rather than 2026. Managing expectations, both internally and externally, will be just as important as managing change itself.
Why Only Governed Agentic AI Will Scale in Banking
In 2026, we can expect to see agentic AI to make meaningful inroads into specific areas of the bank. Current project pipelines suggest early focus areas will include customer engagement, servicing and operational processes.
Consumers are already becoming accustomed to AI in their daily lives, and provided banks are transparent about how agentic AI is used (and it delivers the right outcomes) resistance should be limited.
What determines success with these projects is whether banks choose agentic AI technologies designed specifically for regulated environments. Expect to see banks increasingly filter out solutions that rely solely on prompt-based configuration and grant agents unrestricted freedom to reason and act without sufficient governance.
Creativity certainly has its place, but it falters in regulated tasks such as resolving a credit card fraud case or making a lending decision. Black-box AI that cannot meet banking standards for control, auditability and oversight will face rejection.
Where Banks Must Draw the Line on AI Automation
As AI and automation projects progress, banks will need to be much more serious about when and how AI should stop and where humans should take over. This is about asking how automated we want to be wherever we apply these technologies.
Consider unsecured lending on credit cards. A bank might currently manually review 15% of applications, but that process could be fully automated, and in many cases, an AI agent could operate the same, if not better than a human.
But regulators and internal audit teams are unlikely to accept AI outputs alone. They will need human validation and oversight to some degree, meaning banks will need to rethink not just how many people they need, but how roles evolve as automation expands.
This takes us back to how much AI-associated digital transformation will need to be accompanied by super thorough change management exercises.
Why Human Judgement Still Matters in Banking AI
Banking is ultimately grounded in human expertise, trust and empathy, particularly in higher-value or high-stress situations. From my own experience running lending operations and collections, the limits of automation become clearest in moments of pressure. Small business owners facing financial stress often value speaking with an advisor who has helped similar businesses through similar challenges. That sense of understanding and certainty still matters.
AI may eventually handle more of these interactions, and acceptance will grow over time. But in 2026, not every customer (or situation) will be comfortable relying entirely on an AI agent.
AI will do so much more for banks during 2026, but people will need to be there interpreting, improving, and validating what AI does.
The banks that succeed will be those that treat change management, governance and human judgement not as obstacles to AI adoption, but as the foundations that make it possible.
