Banking has always been about relationships. But for the last two decades, the industry has been on a slow march away from them. Online banking moved the experience from the branch to the living room. Mobile banking moved it to the pocket. Each step made transactions more convenient, but each step also pushed the human connection a little further away.
Now, a new shift is underway. AI in banking isn't just adding another channel. It's bringing the relationship back, and community banks are the ones leading the charge.
We recently hosted a webinar with C.J. Conrad, Senior Vice President of Operations and Innovation at Middlesex Federal in Somerville, Massachusetts. C.J. has been in the industry since 2009 and has lived through every phase of digital banking, from building one of the first mobile banking workarounds on a Fiserv platform to now implementing agentic AI that can execute real financial transactions. His perspective is rare: part technologist, part banker, part designer, all practitioner.
What followed was one of the most honest conversations we've had about where banking is actually headed.
Watch the full webinar: From Digital to Agentic Banking - with C.J. Conrad of Middlesex Federal
From the branch lobby to your car's dashboard
Every generation of banking technology has moved the experience one step closer to the customer. C.J. traced this arc from the beginning:
"It went from bank, driving to the bank, to your home but still attached to your desktop, to outside but still attached to your phone, and then almost becoming detached from your phone to just have an agent that's doing stuff for you."
What makes this moment different from the previous shifts is that the customer no longer needs to interact with a screen at all. C.J. painted a picture that any parent would recognize: you're in the car, driving your teenagers somewhere, and that's when your mind drifts to finances. What bills need to be paid. What money needs to move.
"You'd be able to just hit a button on your CarPlay and tell the bank, hey, I need to move $5,000 into the 529 plan to pay a college bill, and it would be able to do it for you."
That's not a chatbot answering frequently asked questions. That's an AI agent executing a real financial transaction on behalf of the customer, within the bank's own infrastructure. That's agentic banking.
Beyond cost cutting: why AI agents are a growth play
Most conversations about artificial intelligence in banking default to the same playbook: cut costs, deflect support tickets, replace call center headcount. C.J. has a fundamentally different view of what AI agents can do for a bank, one that starts with the emotional reality of how people relate to their money.
"These agents can not only satisfy the mechanics of banking, but they can satisfy the emotive drivers behind the questions that people are asking, and almost anticipate and help without being another human being."
He referenced Sylvia Porter, a financial writer from the 1950s, whose observation still holds: money is rarely just paper and coins. It's security, it's fear, it's the gap between your best month and your worst. Small business owners know this viscerally. So do parents saving for college, retirees planning their next decade, and twenty-somethings who know they should be investing but don't know where to start.
The challenge is that most of these people won't ask their banker the hard questions. They're afraid of being judged, or they don't want to waste someone's time with what feels like a dumb question. AI changes that dynamic completely.
"Not somebody who you perceive is going to judge you, but the agent," C.J. said.
When customers feel safe enough to ask the real questions, banks get to serve them in ways that were never possible at scale before. The customer gets better answers. The bank gets deeper relationships. Everyone wins.
The question that changes everything
One of C.J.'s most compelling points came through a design analogy drawn from his pre-banking career. If a client walks in and asks for a green smiley face on a highway billboard, a designer can deliver exactly that. But if the designer asks "what are you trying to accomplish?", they might come up with ten better solutions that actually achieve the goal.
Banking has the same problem. Traditional digital banking is fundamentally reactive. The customer clicks a button, the system executes. If the customer asks the wrong question, they get a technically correct but practically useless answer.
"With AI, it says, what do you want to accomplish today? And the more people start to trust it, your ability to help that person achieve their goals is much broader."
This is the real unlock. Someone who comes in saying "I want to open another account for my kid" might actually need guidance on 529 plans, education savings strategies, and long-term financial planning. A well-trained AI agent can have that conversation, probe for intent, offer options, and model scenarios, just like the best bankers always have. The difference is that it can do this for every customer, around the clock, without the bank needing to turn every teller into a cash management specialist.
C.J. pointed out that personal finance management tools have existed for years but never saw strong adoption. AI might finally change that, because the interface is a conversation, not a dashboard. You don't need to download data and build a spreadsheet. You just say "can you up the eating out budget to three times a month and show me what that does?" and the agent handles it.
Why banks need to own the AI conversation
Here's where the strategic stakes get high.
Customers are already using ChatGPT, Claude, and other AI tools to ask financial questions. Every one of those conversations represents a relationship touchpoint happening outside the bank's walls, data the bank never sees, context it can never act on.
"It's so important to keep those conversations in the trusted source, in the bank, with the people that are trying to keep your money safe and your privacy and your confidentiality sacred," C.J. emphasized.
When a bank deploys its own AI agent, those conversations come back inside the institution. The bank learns what customers are worried about, what they're planning, and what might cause them to leave. That's intelligence that drives better products, more relevant outreach, and stronger retention.
The alternative is what Tyllen Bicakcic, Payman's CEO, described from Payman's own experience: when they moved their business funds from one bank to another, all the old bank saw was a transaction. No context. No reason. No chance to respond. An AI agent that customers actually talk to gives the bank something it hasn't had since the days of the branch lobby, a window into intent.
Authenticity isn't optional anymore
C.J. made a point that might make some bankers uncomfortable: if a customer asks the AI agent about the best CD rate in the area, the answer can't just list your own rates.
"This has to be an authentic experience. If somebody asks the AI what the best CD rate is in the area, it can't just be our CDs, because that's disingenuous."
This requires a shift in institutional mindset. C.J. acknowledged that he's worked at banks where certain marketing messages were off limits because they might make customers aware of better options elsewhere. But in an AI-driven experience, that approach falls apart. Customers will find out anyway. The question is whether they find out from your agent, who can then offer a compelling reason to stay, or from someone else's.
"I'm banking on the fact that the bulk of them, if your AI agent says, ours is 10 basis points lower but I can open that for you right now and move the money into it, that they're gonna say that is a convenience worth 10 basis points."
Trust is built through transparency. And in an AI-powered banking experience, authenticity isn't just a nice-to-have. It's the foundation that everything else rests on.
Compliance without compromise
No conversation about AI in banking is complete without addressing the regulatory reality. Middlesex Federal is taking a deliberate, methodical approach that other institutions can learn from.
The key principles are straightforward. Every transaction requires explicit customer approval, the human stays in the loop. There's a complete audit trail of every interaction. The bank's compliance team, with decades of experience across Reg E, UDAP, and the full regulatory landscape, is actively involved in writing the prompts and building the agents alongside the technology team.
"The bank is creating the agents alongside of Payman. I think that is so important that the bankers understand what they're creating and how," C.J. said.
The system runs on existing payment rails through the bank's existing digital banking provider and core. AI doesn't change the plumbing. It changes the interface and the intelligence layer on top. And with full chain-of-thought visibility into how the agent reasons and makes decisions, banks and regulators can see exactly why a particular action was taken, what tools were called, and who approved it.
C.J. also highlighted the institution's proactive approach to regulatory engagement: keeping their primary regulator, the OCC, fully apprised of what they're doing, anticipating what an audit would look like before one happens, and educating their existing auditors so everyone grows together.
The community bank advantage
For years, fintechs have eaten into community bank market share by reducing friction. Faster onboarding, slicker apps, fewer clicks. Community banks have watched this happen while sitting on assets that fintechs can only dream of: deep customer trust, regulatory credibility, and real deposit relationships.
AI agents level the playing field.
"This really gives an opportunity for community banks to start to reduce that friction at the same level, and in some cases, potentially even better than some of the fintech, non-bank competitors that we face," C.J. said.
An AI agent that understands intent, executes transactions, answers complex financial questions, and does all of this within a trusted, regulated, FDIC-insured environment is something no fintech can easily replicate. Community banks have always had the trust. Now they have the technology to match the experience.
What the light bulb moment actually feels like
When asked about the moment it all clicked, C.J. described what a colleague who's been working closely on the implementation told him:
"It's like a light went on, or like an explosion in your head, and you're like, whoa, this is massive."
That moment, when you stop asking "can it do this?" and start asking "why can't it do this?", is when the real transformation begins. C.J. said it happened to him when he saw the first Payman demo. It happened to his colleague Karen Grindrod as she worked hands-on with the implementation team. And once it clicks, the ideas don't stop. New use cases emerge across the bank. The question shifts from "should we do this?" to "where do we start next?"
Middlesex Federal is in UAT now. Friends and family testing is next. The roadmap keeps expanding, not because someone in a boardroom decided it should, but because the people actually building with the technology keep finding new problems it can solve.
The future of banking isn't about replacing humans. It's about giving every customer access to the kind of thoughtful, personalized financial guidance that used to require a dedicated relationship manager, and giving every bank the tools to deliver it.
AI agents make that possible. Community banks are the ones building it.
Watch the full conversation: From Digital to Agentic Banking - with C.J. Conrad of Middlesex Federal
Interested in exploring agentic banking for your institution? Talk to our team about what's possible.
