Membership growth hides a divide

Credit union membership is growing. But most credit unions aren't sharing in that growth. New numbers show the gap. Also this week: AI moves deeper into everyday banking, and a payments project gets a clearer path to launch. This is CUAI Weekly.

Let's start with the membership picture. The NCUA released its second quarter state level report on September 23rd. Across federally insured credit unions, membership rose overall. Yet about 56% of institutions had fewer members than a year earlier.

Those two things can happen together. Growth across the system doesn't mean growth at the typical institution. The median membership change was a decline of 0.6%. Among credit unions losing members, more than half had less than $50 million in assets.

Geography mattered, too. Vermont and Alaska led median membership growth. Arkansas and Nebraska had the steepest median declines. These are year over year comparisons through June, not a reading of September activity.

For a leadership team, the useful question is where your own institution sits within that picture. Compare your membership trend with similar institutions before treating the national growth headline as evidence that your market is healthy. The report describes the divide; it doesn't establish what's causing it.

Now, to AI inside the working day. Narmi announced an AI Marketplace on September 24th, bringing its banking tools into one hub. The company says the available features include account opening review assistance, message drafting, marketing tools, and natural language search.

One example is DecisionAssist. It summarizes identity verification information and makes recommendations for staff reviewing new accounts. Narmi reports that average review time fell from eight minutes to less than two. That's a vendor reported result, not an independently verified benchmark or a promise for every institution.

The distinction between available features and the roadmap matters here. Narmi lists document agents and agentic secure messaging as upcoming capabilities. They shouldn't be treated as already delivered.

If you're evaluating tools like these, ask to see the workflow with your own approval rules: what information the model sees, what the employee checks, and how an error gets corrected. A faster review is valuable only if the decision remains reliable.

Finally, a development in the plumbing behind payments. The Clearing House selected Quant to support its On Chain Money Initiative. The planned network would let participating institutions clear and settle tokenized deposits: digital representations of deposits, moved on a different kind of ledger.

The announcement describes connections to existing payment systems, including RTP and CHIPS. It targets availability in the first half of 2027. That's a planned launch, not a service every credit union can switch on today.

Participation details are still to come. For credit union payments teams, this is a development to track alongside existing real time payment options. Ask about access, integration, and the member or business problem it would actually solve before committing resources.

Those are three developments from the week ending September 27th. You'll find the original reporting and sources with this briefing. For more industry coverage, visit Credit Union AI News.