Finopotamus Unscripted One on One: Jim Perry Talks AI
By Finn O’Potamus
Finopotamus Unscripted is our forum for video panel discussions. The format calls for Finopotamus Publisher John San Filippo to serve as moderator and three subject matter experts to share their knowledge and experience. That all changed when we requested an interview with Market Insights, Inc. Senior Strategist Jim Perry – a request originally intended to support a written article.
“I had Jim on our TechSolutions4CUs podcast a few months ago, so I know what a wealth of knowledge and insight he brings to the table,” said San Filippo. “Based on that experience, I knew two things. First, I knew the interview would run close to an hour and we’d be forced to wrap it up before we really wanted to. Second, I knew that a 900-word article – or even an 1,800-word article – wouldn't do the interview justice. I’d have to leave something important out.”
Thus was born this first-ever Finopotamus Unscripted One on One. Here’s just a brief recap of the interview.
Navigating Frontier AI and Back-Office Reality
The discussion kicked off around the stark contrast between high-level frontier AI headlines and the everyday realities of credit union back-office operations. Perry noted that while stories of rogue agents and model risks often trigger science-fiction panic among executives, pulling back is the wrong reaction.
“If they use that as an excuse to slow down their own movement towards adopting any level of AI functionality within their organizations, they’re going to fall further and further behind,” said Perry.
While frontier developers contend with core architectures, guardrails, and rapid recursive improvement, community financial institutions face practical concerns like hallucinations, accuracy, and operational reliability. Perry stressed that while consumer-facing generative models produce variable outputs, enterprise deployments must be bounded.
“We can set the rules in place to govern a deterministic answer,” Perry explained. “And then that’s where you want the human looking at the answer, or it’s handing off an answer to a human being.”
The Urgent Need for True AI Governance
As agentic workflows begin executing tasks—such as combining multiple agents to assemble comprehensive marketing campaigns in minutes—governance cannot remain an afterthought or a delegated checklist.
“These AI capabilities and governance have to advance together,” said Perry. “Right now, I think we’re really falling behind in the industry on the governance side of it.”
Perry warned against boards abdicating oversight to operational staff or treating policy as a passive exercise. With AI embedded in core systems and third-party tools across the enterprise, governance requires ongoing board-level scrutiny, continuous contract reviews, and clear operational dependency plans.
“Make sure that you’re really digging into all of your vendor agreements, your core agreements, making sure that you understand precisely what safeguards are in place,” said Perry. “If one piece of tech goes down for a while, it’s going to shut down what part of our operations? And if it does shut down part of our operations, what’s our backup?”
The Disruption of the Member Interface
Beyond operational efficiency, the conversation turned to the threat AI poses to member relationships. As consumers grant tools like ChatGPT, Claude, and specialized agent platforms direct authority over their financial lives, third-party agents increasingly sit directly between credit unions and their members.
Perry pointed to common consumer friction points such as certificate of deposit (CD) rollovers, where institutions historically relied on member inertia. Autonomous agents capable of monitoring rates and reallocating funds eliminate that friction entirely.
“Everybody right now is concerned we could wake up one day to a deposit runoff that we didn’t even foresee,” Perry noted.
Coupled with the broader convergence of open banking and digital assets, this shift makes discoverability vital. If a credit union’s data architecture and unique offerings are not machine-readable, AI agents simply will not recommend them.
“If AI can’t find you, if you don’t come up in an AI-driven search, you’re invisible,” Perry declared.
Rethinking the Primary Financial Institution
The dialogue also tackled the debate over the apparent demise of the primary financial institution (PFI). While San Filippo argued that members crave simplicity and would prefer a unified “digital PFI” over managing dozens of disparate apps, Perry offered a sobering view based on consumer data.
“The traditional understanding of primacy is dead,” said Perry. “Generally speaking, we don’t think of any single institution as being our only option.”
Perry emphasized that financial institutions can no longer expect to be everything to every member at every stage of life. Instead, they must excel at specific niches where their value proposition clearly outperforms competitors.
“All financial institutions can hope for moving forward, and organize themselves around, is: How can I be the credit union that they choose for this need going forward?” Perry said. “Because that’s how consumers are organizing their financial lives right now.”
Advice for the Credit Union C-Suite
When asked for a single actionable priority that leadership teams should tackle immediately, Perry pointed straight to vendor management and technological agility.
“Take another very close look at all of your vendor relationships, and really determine how they’re using AI,” advised Perry. “Make sure that you have the kind of optionality that will allow your credit union to evolve alongside the technology, as opposed to getting left behind by committing to something for a longer period of time than the technology will actually last.”
