How a Wyoming credit union CEO built an AI back office
StagePoint Federal Credit Union serves about 9,000 members from a branch in Laramie, Wyoming, and two more in Casper. It also operates one of the more ambitious artificial intelligence programs among small credit unions.
The initiative has been developed largely in-house through the vision of President and CEO Tyler Valentine and the technical leadership of Geoff Shimp, vice president of technology, who have worked together to build and deploy AI solutions across the organization.
Over the past two years, the rural, community-chartered cooperative has moved underwriting, its knowledge base, vendor reviews, fraud tracking, and even staff performance reviews onto AI, with Valentine himself assembling many of the dashboards and workflows that hold it together.
Valentine has led the credit union for 18 of his 25 years there, and treats staying near the leading edge as a small-shop necessity. He also wears most of the operational hats: he runs vendor management, serves as the HR lead, and, in his own words, creates his own dashboards for business operations.
“I think small credit unions, oftentimes, we have to buy out-of-the-box solutions that are created by others,” he said. So rather than wait for a vendor, he experiments and looks for ways new technology can fit a roughly $130-million balance sheet.
Valentine bought a personal ChatGPT subscription as soon as the tool went mainstream three to four years ago and showed his team how to use it for letters, marketing copy, and research. That familiarity set the stage for bigger moves.
Starting small, then going all in
The first big step was forced by circumstance. When the employee who handled StagePoint indirect lending had a medical emergency and was not expected to return, her workload landed on several colleagues at once. StagePoint had already researched Zest AI, an underwriting-automation credit union service organization, and the staffing gap made the case to act. The credit union signed on about two years ago and has since added Zest’s LuLu lending-intelligence platform and its Strategy module.
Around the same time, StagePoint adopted CurrentWave AI, a closed knowledge base for credit unions. The team loaded in policies, procedures, training guides, and vendor manuals, including documentation from its core processor and its card processor, Velera. Staff now ask plain-language questions, such as how to place a stop payment on a check, and get step-by-step answers that cite the source document. Because the system is closed, it learns only from what StagePoint feeds it.
That changes the frontline experience. Valentine points to a recent run of questions about home equity line of credit limits, likely from a new mortgage loan officer getting up to speed. Where she once clicked layers deep into a policy folder while a member waited, she now asks the question and gets a concise answer on the spot. Staff can also flag any question the system cannot answer, showing leaders where the documentation has gaps.
Building its own systems
For everything that did not come in a box, StagePoint turned to a business license of Anthropic’s Claude. Valentine and his IT lead built it themselves. “If you can articulate it, Claude can build it,” he said.
The result is a full internal intranet that now houses HR workflows, one-on-one meetings, and performance discussions, split between an employee portal and the manager dashboards Valentine designed. Those dashboards do more than display information. Supervisors get pinged when a task goes unfinished, overdue items surface on their own, and other leaders can see at a glance which managers have fallen behind. The same platform runs vendor management, where it reads uploaded documents such as financial statements and System and Organization Controls 2 reports and returns a summary, plus IT ticketing and fraud-case tracking.
The payoff shows up first in Valentine’s own week. His vendor-management dashboard now serves up what he calls “little cards” that flag which reviews are coming due and which have slipped overdue, work he once chased across a spreadsheet, his task bar, and calendar reminders, and often crammed in just before an NCUA examiner arrived. He estimates the dashboard alone saves him 20 to 30 hours a year. The analytics matter as much as the time: IT ticketing data now reveals, by device and user, whether a pattern signals a real training problem rather than a hunch, while the fraud module tracks cases, trends, and dollars prevented.
Compliance is the obvious question, and Valentine addresses it head-on. He says StagePoint runs Claude on an internal server walled off from the internet, reachable only from inside the building. No member data sits in the system, he notes, and the employee information that does cannot be reached from the outside world or used to train anything beyond the credit union’s own content.
“It’s not hallucinating because it’s only learning from the information that we’re giving it,” he said.
What members notice, and what comes next
Members do not see most of this directly, but Valentine argues they feel it through more confident, better-informed staff. The credit union is also evaluating options to bring AI to member-facing chat and phone support, though it has not settled on a direction.
Accounting is the next frontier. Instead of reconciling card-processor and core totals together and hunting for the difference, Valentine wants AI to compare the two line by line and flag what posted twice or never posted. Looking outward, he is also forming a credit union service organization to package the tools he built for small credit unions that lack the time or staff to build their own.
His message to peers is blunt.
“AI is here, it’s not coming, it is here,” Valentine said, and says the institutions that lean in will thrive. Starting early, he adds, was about staying in control of the technology: the credit union wanted to learn the tools “before it was smarter than us.” For a cooperative that has served Laramie since 1935, the goal is continuity.
“We’ve been around for 91 years,” he said. “We want to be here for the next 91.”
