Why Credit Unions Need Their Own Approach to AI

On International Credit Union Day, we officially launched Cultivate AI. It was a milestone we’ve been working toward since we first shared the idea at GAC in March, and one that hundreds of credit union employees helped us reach through testing, feedback and a willingness to tell us what worked, what didn’t and what they actually needed from AI.

Now that Cultivate is out in the world, I’ve been thinking about one of the questions behind the entire project: Why does the credit union industry need its own approach to AI in the first place?

There are already plenty of AI tools available. ChatGPT, Copilot, Gemini and Claude are becoming familiar names, and many credit unions are already experimenting with one or more of them. They are incredibly capable tools. But they are also general-purpose tools, designed to serve millions of people, companies and use cases.

A credit union has a different set of needs.

When someone inside a credit union asks a question about BSA compliance, an internal policy, a regulation or the institution’s performance relative to its peers, a plausible answer isn’t enough. The employee needs to know what that answer is based on. They need to be able to check the source. In some cases, they may need to show an examiner exactly where the information came from.

That difference sounds small until you start putting AI into real workflows.

Cultivate was designed around that reality. It grounds answers in credit-union-specific information, including NCUA regulations, guidance and letters, as well as a credit union’s own policies and documents. The response doesn’t simply provide an answer; it includes the source material behind it so an employee can click through, read the relevant passage and decide whether the answer holds up.

That is a very different experience from asking a general-purpose AI tool a question and hoping it understands the context.

Take something as ordinary as reviewing a BSA policy. A general-purpose AI tool may be able to explain BSA requirements very well, but it doesn’t inherently know your credit union’s policy or whether that policy still lines up with current guidance. Cultivate can compare the institution’s own document against the relevant guidance, identify where something may no longer align and show the passages it used to reach that conclusion.

For me, that gets much closer to what useful AI should look like inside a credit union. It isn’t about producing more words faster. It is about helping an employee get to reliable information faster and giving them enough transparency to use their own judgment.

Privacy is another reason a purpose-built approach matters. Credit union employees work with information that deserves a much higher standard than “be careful what you paste into the prompt.” Cultivate was designed so member data is not used to train AI and is not sent to consumer AI tools. The credit union owns and governs its data.

I think that distinction becomes increasingly important as AI moves from experimentation into everyday work. Telling employees never to enter anything sensitive into a public tool may work as an interim policy, but it doesn’t solve the larger problem. If we expect employees to use AI to do meaningful work, then the safeguards need to be part of the environment itself.

There is another difference that I believe will become more valuable over time: Cultivate can understand the institution asking the question.

A general-purpose assistant doesn’t automatically know a credit union’s charter, field of membership, asset size, state or peer group. Cultivate is being built with that context in mind, including live 5300 Call Report data across roughly 4,400 credit unions. That creates the opportunity to move beyond generic answers toward answers that actually reflect the credit union using the platform.

That was important to us because “What should a credit union do?” is usually not a very useful question. A $100 million community credit union and a multibillion-dollar institution may be dealing with the same issue from very different positions. Context matters.

So does accountability.

One of the things I’m proudest of about Cultivate is that it did not begin with a technology company deciding what credit unions should want. It came together through collaboration among Vizo Financial Corporate Credit Union, Filene Research Institute and Vertice AI, with hundreds of beta users helping us shape what the product became. Cultivate is structured as a CUSO and is being built specifically for credit union staff and the work they do.

That structure matters because AI is going to keep changing. The models will get better. New capabilities will appear. Things that feel remarkable today will probably feel routine a few years from now. The long-term question is not simply which company has the most impressive model at a particular moment. It is who gets to decide how that technology is applied to our industry.

Credit unions have always believed that ownership and incentives matter. We built an entire financial movement around that idea. I don’t see why our approach to AI should be any different.

Christina Naylor, COO of Carolina Trust Federal Credit Union, put it particularly well during our testing: “We knew we needed something that would make our employees' lives easier and our members' lives better.”

That is ultimately the measure I keep coming back to.

Cultivate should make it easier for an employee to find the policy they need, understand a regulation, research an issue, compare information and get to a better starting point for a decision. It should help knowledge move more easily through the organization rather than sitting in one person’s head or buried somewhere in a shared drive. And it should do those things without asking a credit union to compromise the privacy and accountability its members expect.

We’ve officially launched Cultivate, but in many ways the more interesting part starts now. The same collaboration that shaped the product during beta can continue shaping where it goes next.

There will be plenty of AI built for everyone.

Our opportunity is to build the one credit unions need.

Learn more at getcultivate.ai.