This is a summary of an episode of Main Street AI, an educational podcast on AI led by our founder. Join 3,700+ business leaders and AI enthusiasts and be the first to know when new episodes go live. Subscribe to our newsletter here.

TL;DR:

• Jason Stverak wants the NCUA to create a sandbox with firm rules on what credit unions must achieve and room to experiment on how
• Regulation can't keep pace with AI, so he wants the regulator to learn alongside the industry
• Small credit unions can adopt AI by building it together through CUSOs, since none can afford to build alone
• Credit unions are ideal test labs for AI vendors, and a rural credit union that adopts a tool proves it anywhere
• Lending decisions need human oversight because fair lending laws can't be handed to a program
• Credit unions need a solid B, not an A plus, to serve members

Before we dive into the key takeaways from this episode, be sure to catch the full episode here:

Why Credit Union Regulation Can't Keep Up With AI

Jason's regulatory argument comes down to timing.

"Regulation cannot keep up with the evolution of technology. It just can't."
Jason Stverak

By the time a law is written, he says, the industry is in a whip and buggy while technology is on a moonshot. His answer is broad guidelines for consumer protection, with flexibility inside them.

What an NCUA Sandbox Would Look Like

Jason describes hard and fast rules on what credit unions must achieve, with more freedom on how they get there. It isn't the wild, wild west. A credit union could experiment as long as it serves members and doesn't violate federal law or regulation.

He also wants the regulator to keep learning as the industry does, and to act as a matchmaker. Instead of a flat "you can't do this," the NCUA could say that credit union A tried something and is similar in size, so the two should talk. Ankur adds that a neutral regulator is a good outcome, and an enabling one would be better.

The timing matters. John Crews became NCUA chairman on August 24 and said he intends to focus on enabling innovation and right-sizing the regulatory framework. Jason called the incoming leader an incredible choice.

Small Credit Unions Can't Build AI Alone

Jason says DCUC is 63 years old and represents more than 200 credit unions with over $525 billion in assets and 40 million members. Consolidation frames his concern. The industry had more than 10,000 credit unions 10 to 15 years ago. NCUA counts 4,214 federally insured credit unions today, and Jason puts about 3,000 of them at $300 million in assets or below.

"We're working on other avenues so that they don't have to make the choice of consolidate or die."
Jason Stverak

A large bank can write one check and build in-house, but only for itself. Credit unions pool. Twenty, thirty, or fifty of them form a CUSO to develop a product none could afford alone. DCUC's CU Unplugged event shows the model. Credit unions and their tech staff met in San Francisco with no agenda, wrote code for days on real problems, and left with new apps and AI programs to run at home.

Credit Unions as AI Test Labs

Jason pitches the tech sector too. If AI works at a credit union in Casper, Wyoming, or his hometown of Medora, North Dakota, he says, it can work in Washington or Charlotte. Vendors get real-world feedback, and early users might get a reduced rate.

Boards, Budgets, and the Attrition Question

Credit unions range from "what is AI?" to boards asking for it in the 2027 budget. Jason says volunteer boards are underrated here, because even the smallest credit union has someone raising AI at the meeting. Trust builds in small steps, starting with something like automated meeting notes. Once a board is comfortable, larger investment gets an easier case. He mentions that when a longtime employee retires, a credit union might move that role's resources to agents, without eliminating human connection.

Lending Needs Human Oversight

Jason draws a firm line on lending.

"The technology is only as good as the person who programs it."
Jason Stverak

Federal fair lending law can't be handed to a program automatically. Bad data could lead a model to steer away from certain neighborhoods, which is the redlining problem CRA was created to fix.

A B Is Enough

Ankur worries the biggest firms will pull further ahead. Jason agrees the haves move faster, then offers a different standard for credit unions.

"We may not be getting an A plus, but we don't need an A plus to serve our members."
Jason Stverak

Credit unions have always been the little guys, he says, and they do more with less. Maybe a credit union can't afford the $200 subscription but can afford the $100 one. Like big-screen TVs, he adds, today's premium capability gets affordable within a few years.

Why the Mission Matters

Jason's motivation is credit unions that see past a three-digit score. His example is a veteran leaving the Air Force who needs money to buy a truck and start a business.

How This Works in Practice

Jason names back-room work and initial loan screening as the practical starting points. Multimodal's AgentFlow puts agents on that kind of file work, reading and routing loan documents beside the existing loan origination system, so credit union teams get relief without replacing their core.

‍

Want more on financial services and AI? Check other episodes here.

Frequently Asked Questions

Why should the NCUA give credit unions an AI sandbox?
Regulation can't keep pace with technology, according to Jason. A sandbox would set firm outcomes and leave room on method.

What would an NCUA AI sandbox look like?
Hard rules on what must be achieved, with flexibility on how. The regulator would also learn alongside the industry and connect similar credit unions.

Can small credit unions afford AI?
Yes, by building it together. Jason says 20 to 50 credit unions can form a CUSO to develop what none could fund alone.

Can AI make lending decisions at a credit union?
Not without human oversight. Fair lending laws can't be handed to a program, and AgentFlow keeps people in review of loan files.

Why are credit unions consolidating?
Competitive pressure, changing demographics, and the ability to compete all play a role. The count has fallen from over 10,000 to about 4,200.

How should a small credit union fund AI?
Start with small wins that build board trust, then scale. Jason says resources can shift when staff retire.

‍

‍