A Conversation with John Allison, CMO of Ceto, and Dr. Lamont Black.
The highest-impact first step is an internal, employee-facing AI assistant for contact center staff, not a member-facing chatbot. This lowers risk while immediately improving the member experience.
Dr. Lamont Black, fintech professor at DePaul University and Filene Research Institute Fellow, outlines this exact sequence below.
We sat down with Dr. Lamont Black to get a direct answer on where that starting point actually is.
Dr. Lamont Black put it plainly when he sat down with Ceto: "Unless you have that broader picture and that sense of direction, you're just going to be doing this in a very ad hoc, bolt-on way."
AI isn't a technology decision. It's a leadership decision. It spans every business line: lending, operations, member service, and marketing. The mistake most institutions make is treating it as an IT department initiative instead of a leadership-table conversation.
Plain language summary: Talk about AI as a leadership team before buying anything. This single step prevents the "bolt-on" failure pattern most institutions hit six months into a rushed AI purchase.
Most leadership teams worry AI will damage the member relationship. That fear is valid; it's usually aimed at the wrong target.
The version of AI most people picture is a phone tree: a robo-voice, a frustrating menu system. That's not what's being discussed in 2026.
Members already experience AI-driven personalization through Netflix, Spotify, and Amazon. When a credit union doesn't close that gap, members feel the difference. And that gap compounds every year it's left unaddressed.
Plain language summary: AI doesn't replace the human touch. It's what makes high-touch service possible at scale as a credit union grows.
This is the step most institutions skip.
| Starting Point | Risk Level | Member Exposure | Typical Outcome |
|---|---|---|---|
| Member-facing chatbot | High | Direct | Hallucination risk, compliance exposure |
| Employee-facing AI assistant | Low | Indirect | Faster service, no public-facing risk |
Dr. Black recommends starting with an internal knowledge assistant for contact center staff, not a chatbot on the website.
"Humans doing what humans do best, relating to the member, making it feel warm and empathetic. The computer is doing what computers do best: finding information, retrieving it, and delivering it," he said. "We're both operating in our zone of genius."
New contact center employees often don't know every policy by heart. When a member calls with a question, they have to dig through internal files to find the answer, and the member waits while they search.
An AI assistant fixes this quietly. The employee asks it a question in plain language and gets the answer instantly. The member never interacts with the AI at all, they just get a faster, more confident answer from the person they're already talking to.
Two guardrails matter most, according to Dr. Black:
Data privacy. Member data and proprietary institutional data should never enter a public AI training model. "That's kind of the biggest no-no," he said. A clear internal AI policy on secure data environments solves this without shutting down experimentation.
Bias and explainability. Machine learning isn't rule-based the way traditional coding is. The model learns on its own and can produce implicit bias even without race or other protected characteristics built in as direct inputs. Dr. Black pointed to lending as the clearest risk: a model can replicate something close to redlining without ever using race as a factor, just by weighting other data points that correlate with it. "What we're seeing from a regulatory perspective is explainability," he said, making sure institutions can show how a model reached its decision.
Where should a credit union start with AI?
Start with leadership alignment, not technology. Put AI on the leadership team agenda before any vendor conversation, then begin with an internal, employee-facing use case before deploying anything member-facing.
What is an employee-facing AI assistant?
An internal tool that lets staff query company policies, procedures, and product information in plain English, returning instant, accurate answers instead of requiring manual document searches.
Is AI for credit union chatbots too risky to start with?
Member-facing chatbots carry the highest hallucination and compliance risk. Most credit unions should prove out AI internally first, according to Dr. Lamont Black.
How does AI improve member experience without losing the human touch?
AI gives staff instant access to information so the human interaction stays focused on the member instead of searching for answers. The member still talks to a person; that person is simply better equipped.
What is the board's role in credit union AI adoption?
Setting ethical guardrails, data privacy policy and bias monitoring, rather than managing technical implementation. This is a governance function, not an IT function.
Credit unions leading on AI in 2026 aren't the ones with the largest technology budget. They're the ones who had the leadership conversation first.
Watch the full conversation with Dr. Lamont Black recorded directly by Ceto, covering the member experience fear, the leadership alignment question, and the practical first step every credit union can take.
About Dr. Lamont Black
Dr. Lamont Black is a fintech professor at DePaul University's Driehaus College of Business and a Fellow at the Filene Research Institute, where he leads the center on the Credit Union of the Future. He works directly with credit union leadership teams across the country on AI adoption strategy and is a regularly published voice on AI in financial services through Filene's research, articles, podcasts, and webinars.
Stay in the conversation:
Follow Dr. Lamont Black on Linkedin for ongoing insights on AI and credit unions.