What is an AI governance framework for a credit union, and how is it different from an AI policy?

An AI governance framework covers ongoing risk and business process for managing models, agents, and evals as they run. An AI policy is a static document. Kirk Drake, founder and CEO of CU 2.0, argues most credit unions have the second and mistake it for the first.

What should an AI implementation roadmap for a credit union include in the first 90 days?

A skills catalog built from existing job descriptions, a documented workflow system replacing outdated SOP manuals, and an open call for staff volunteers rather than a mandated rollout. Kirk Drake compresses what took his own company 18 months into 30 to 60 days for credit union clients.

Is AI data privacy a real concern for credit unions, or mostly a perception problem?

Mostly perception, according to Kirk Drake. Zero trust environments, role-based access, PII scrubbers, and hosting patterns like Microsoft Foundry already solve the technical risk. Platforms like Multimodal's AgentFlow build that architecture in rather than treating it as a separate project.

Should a credit union hire a VP of AI or run an AI readiness assessment first?

Neither, in Kirk Drake's view. Skip the dedicated hire, since AI adoption works better as an organization-wide strategy than a department. Instead, send an open invitation for volunteers and treat the 5 to 7 percent who opt in first as the real readiness signal.

How do you close an AI skills gap without a company-wide mandate?

Let visible results do the recruiting. An opted-in group of early volunteers typically shows measurably more output within six to twelve weeks, which pulls the next wave in voluntarily rather than through a top-down training requirement.

Why does Kirk Drake say 30 years of skills now count for zero?

Because the specific tool proficiencies, PowerPoint, spreadsheets, pivot tables, that people built careers on have been commoditized by AI. Drake's point isn't that experience is worthless, but that treating the loss as sunk cost is what keeps institutions from moving forward.

Finance AI

September 2, 2026

The FAQ Bot That Burned Through a Credit Union's AI Spend

Zach Cox runs business innovation at Credit Union 1, translating AI initiatives into measurable ROI; Aaron Cain, a former creative director, runs digital innovation and owns how those initiatives actually look and feel to members and staff.

Bareerah Shoukat

Writer

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:

Credit Union 1's first AI use case, a low risk FAQ bot, got shut down after a vendor's 15 minute interaction window billed every chat as two interactions
The bot burned through the credit union's spend without moving a single business metric it was built to improve
Credit Union 1 splits AI ownership in two. Zach Cox (Director of Business Innovation) owns the ROI case, Aaron Cain (Director of Digital Innovation) owns the build and member experience
AviaryAI's outbound voice agents now call up to 7,000 members ahead of a merger conversion, easing the inbound call spikes that used to follow every deal
AI driven headcount change at Credit Union 1 shows up through normal turnover. Tasks get absorbed a few percentage points at a time, and fewer people get hired back into the same total workload
A generative chatbot once quoted a hypothetical disclosure example back to a member as though it were a real dispute resolution. Credit Union 1 killed the bot that day and started building its own AI specific knowledge base.

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

Guest: Zach Cox, Director of Business Innovation, Credit Union 1 Guest: Cain, Director of Digital Innovation, Credit Union 1 Host: Ankur Patel, CEO & Founder, Multimodal

What went wrong with Credit Union 1's first AI chatbot?

Credit Union 1's first deployed AI use case was a high volume FAQ bot meant to be low risk. The vendor counted every chat in 15 minute increments as one interaction. Sessions kept running in the background even after a member closed the window, so a single "hi" and walk away registered as two interactions, minimum, every time. The bot burned through the credit union's spend without touching a single business outcome the team had set out to move.

"Don't let the tail wag the dog," Zach said of the lesson.

"Vendors are incentivized to get us using the product, not necessarily the right outcome, but at the end of the day, we chose them to represent us."

Credit Union 1 pulled the plug and rebuilt their intake and vetting process before adding another use case.

How should credit unions evaluate AI vendors?

The lesson Zach draws from the pricing miss comes down to ownership. Vendors can inform pricing and product decisions, but the credit union carries the reputational risk when something goes sideways with a member, so that judgment has to stay with them. Cain pointed to a second layer of the same problem: the vendor's billing logic never accounted for how members actually behave, closing a window without formally ending a session, and nobody caught the gap until the invoice arrived.

Their fix going forward is running small, controlled tests before any full rollout, sometimes just an hour long, to see how a use case performs before committing real budget to it.

Why does Credit Union 1 split AI ownership across two roles?

Many credit unions run AI decisions through committee or hand them entirely to IT. Credit Union 1 built two roles for it instead: Zach Cox evaluates whether an initiative should happen and measures quality of interactions, reduced escalations, and ROI after go live, while Cain, a former advertising creative director, shapes what the experience looks and feels like for members and staff.

"I think that's where I tend to come into play, that cross departmental collaboration,"

Zach said of the handoff between his data driven prioritization and Cain's build side execution.

How is AI used for credit union mergers and member communication?

Credit Union 1 is working through roughly ten mergers this year, and reaching every affected member individually was never something a human team that size could pull off. Using AviaryAI's outbound voice agents, they've connected with as many as 7,000 members in a short span ahead of a conversion, warning them about incoming debit cards and other changes before the switch happens.

"AI is really the only path forward if we're trying to do that much mass communication at scale"

Cain said, describing a merger team of roughly ten people that could never have made those calls itself.

Does AI cause layoffs at credit unions?

Zach describes it as absorption rather than replacement. AI takes on a percentage of what any one role does, mostly the repetitive parts, and that reduction stacks across a full team over time until the total headcount needed for the same workload starts to shrink.

"It allows for replacement through attrition where no one's role is fully replicated and eliminated"

Zach explained, "but as tasks get absorbed one layer at a time, there's fewer people for the same total amount of work." Contact centers already see 30 to 40 percent annual turnover, so that reallocation happens through normal churn.

What does data readiness actually require before deploying AI?

Credit Union 1 initially pointed its AI at the public website, assuming compliance vetted content was automatically safe for a chatbot to draw from. A member chatted in about a disputed charge, the bot keyed off the word "dispute," found an unrelated example scenario buried in a disclosure page, and read it back to the member as fact. Credit Union 1 shut the bot down that day and started building an AI specific knowledge base instead of repurposing website copy.

How This Works in Practice

Credit Union 1's approach to AI in 2026 has been to deploy, learn from what breaks, and formalize the infrastructure around it afterward.

"We're going, right? Let's deploy, let's learn, let's fail, let's iterate, and let's just keep moving forward"

Zach said. "It's given us opportunity to really take a step back and say, hey, we've done it, but now we do need that infrastructure in place."

That's the gap AgentFlow closes: governed, auditable AI workflows that let credit unions move past one-off pilots without recreating the vendor pricing and data readiness traps Credit Union 1 learned the hard way. Explore AgentFlow for credit unions

Want more credit union AI perspectives? Check out how CUltivate is rebuilding vendor trust for smaller credit unions and why the 1% CUSO investment cap is becoming a bottleneck for credit union AI.

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

Frequently Asked Questions

1. Why did Credit Union 1 shut down its first AI chatbot?
A vendor billing model counted every chat session as two interactions minimum, burning through spend without improving the business outcomes the bot was built for.

2. How should credit unions evaluate AI vendors?
Own the outcome, not just the tool. Zach Cox's rule: vendors are incentivized to drive usage, so the credit union has to own that judgment call itself.

3. Does AI cause layoffs at credit unions?
At Credit Union 1, AI driven headcount change happens through attrition. Fewer people get hired back into a shrinking total workload rather than existing staff being cut.

4. How do credit unions use AI for mergers?
Credit Union 1 uses AviaryAI's outbound voice agents to call thousands of affected members ahead of a merger conversion, easing the inbound call spikes that typically follow.

5. What is AI data readiness for a credit union?
It means content built specifically for AI intake rather than website copy repurposed for a chatbot. Credit Union 1 found reformatted content produced far better AI responses than raw articles.