I spent fifteen years working inside a community bank before I ever advised one on digital strategy. I learned banking first and applied AI second. Most of what I hear about AI these days doesn't come from a place of excitement. It comes from boards and executives who are cautious, and rightly so. Banking runs on trust, and trust doesn't leave much room for guessing.
So instead of another AI hype piece, I want to walk through the questions I actually hear from community bankers, and how I answer them.
It’s strategy first, not technology
The first thing I tell every bank is to not start with the technology. Start with your strategic goals. What's actually slowing your team down? Is it productivity, staffing, efficiency, documentation? AI only earns its place once you know what problem you're pointing it at. Chase the technology first, and you end up with a tool nobody uses and a line item nobody can justify. Align it with something the bank already cares about, and you've given adoption a reason to take hold.
That said, I understand the pull to just dive in and start experimenting. Once people get a taste of what AI can do, it's genuinely hard to stay disciplined. Someone opens it up to draft a compliance memo, and an hour later they're asking it to write the toast for their daughter's graduation party. That's not a knock on anyone. It's just what happens once you realize AI can actually help with more than you initially expected.
But that curiosity needs a container. Before your team starts testing tools, sit down and get honest about what's standing between you and your income goals, your growth targets, your efficiency ratios. Then look at whether AI is actually one of the things that can move those numbers, or whether you're just following the crowd because everyone else is talking about it.
Once that's clear, the smallest possible step is the right one. I tell bankers to open a voice conversation with an AI model, whether that's Copilot, ChatGPT, or Claude, and just have a conversation with it… literally. Each of these models offers a “voice mode” that gives you the ability to just talk to it, and it talks back.
Once you get comfortable with the idea of having a conversation with an AI, ask it what a community banker should be thinking about before adopting AI. It'll feel a little awkward the first time. I remember how strange it felt when I did it myself. It gets easier over time, and I now fire up a conversation while driving to the airport for my next event or walking my dogs. The ideas that come out of those conversations - blog topics, presentation angles, ways to frame a problem - have become a vital part of how I work these days.
Is this going to replace our people?
No. But that fear is real, and leadership needs to address it head-on rather than hope it fades on its own. I ask bankers to think of AI as a coworker, not a replacement, and there's an important distinction buried in that word. We give tasks to a loan system or a CRM. We work collaboratively with AI. We converse with it, we challenge it, and it can push back or offer something we hadn't considered. That's a different relationship than anything most bank employees have had with technology before, and it needs to be treated that way. When was the last time you had a conversation with Excel? Never.
I've seen employees who manually pulled data from four separate spreadsheets every week for years, simply because that's what they were trained to do and nobody had ever asked them to question it. Handing that task off frees the person to spend more time with customers, more time in the community, and often, more time at home at the end of the day instead of staying late to finish paperwork.
Leadership has to make that case explicitly, though, or the fear will win by default. People need to hear, directly, that this isn't being brought in to thin out the staff. It's being brought in to take the repetitive, data-heavy, generative busywork off their plate so they can be more present with the parts of the job that actually require a person.
There will still be resistance, and that's normal. I think about it the same way I think about fax machines. Somebody out there still prefers writing on paper and feeding it through a machine, and I understand the comfort of that. But email is how we communicate now, whether you love it or not. AI adoption is heading in a similar direction. The banks and the bankers who adapt early are going to have an easier time than the ones who wait until they have no choice.
None of this means charging in with rose colored glasses on. Security, privacy, and fraud are real concerns, and bankers think about them constantly, which is a good instinct, not an obstacle. The same discipline that governs a lending decision should govern how a bank rolls out AI. Understand what could go wrong before you scale anything up. Be careful about what information goes into a model until you're confident it can be trusted. Risk doesn't disappear because a tool is useful, and pretending otherwise is how banks get burned.
What does this actually look like once it's working?
More often than not, it looks like individual employees getting comfortable enough to ask “why do we still do it this way?” and having a system in place that lets them explore the answer without worrying they'll get in trouble for questioning the status quo. That permission has to come from leadership, and it has to be genuine rather than a line in a memo.
This is also why I keep saying it's a top-down and bottom-up effort at the same time. Leadership needs to treat AI as a strategic priority worth real time and resources, not a shiny object to chase because a competitor mentioned it in a press release. But the people closest to the daily work are the ones who will actually spot where it helps, because they're the ones living with the inefficiency every day. Adoption works best when both directions move together instead of one waiting on the other.
Over time, some of those employees become the person other staff go to first, before anyone even thinks to call in outside help. I've watched that happen inside banks I work with directly, where individuals who started out skeptical are now the internal go-to on AI questions for their whole department. That shift, more than any dashboard or press release, is the real sign the adoption worked.
What this actually requires is a clear sense of what you're solving for, a willingness to start small enough that the first step feels almost too easy, and a plan for handling risk the same way you already handle it everywhere else in the institution. No in-house tech team needed, no single vendor bet required, just a starting point every community bank can work with, regardless of size or budget.
