Five items worth your attention this week, each with a take. For the standing framework behind how we read vendor moves, see the 2026 CU AI vendor map.

1. The last big core provider picks its AI lab

Jack Henry will build a proprietary AI security platform on Google Cloud’s agentic defense products, the Gemini Enterprise Agent Platform, and Mandiant consulting, covering the roughly 7,400 community institutions it serves. As American Banker notes, all three major core providers have now paired with a frontier AI lab: Fiserv with OpenAI, FIS with Anthropic, and Jack Henry with Google. Our take: your core provider just chose your AI security vendor for you. Ask your Jack Henry, Fiserv, or FIS rep what ships, when, and what it costs, because these capabilities will arrive bundled whether you planned for them or not.

2. NCUA finalizes the first 11 rules of its Deregulation Project

On August 5, the NCUA Board approved eleven final rules, the first batch from its ongoing review of regulations for safety, soundness, and resilience. The rules touch chartering, field of membership, loans to other credit unions, and share insurance disclosures. None are AI rules. Our take: read the posture. This board is trimming its rulebook, which makes a standalone AI regulation even less likely near-term. Your AI obligations continue to run through existing vendor management and model risk expectations.

3. A $30M seed round argues your core cannot power AI agents

Maximum, a Miami fintech, emerged from stealth with $30 million to build a new core designed for autonomous agents, arguing that batch-processing systems cannot give agents the real-time ledger access they need. Our take: directionally right, commercially unproven. Replacing a core is a commitment you live with for ten years or more, and no credit union should make it to chase agents in 2026. But the batch-window constraint is real, and it is worth asking your core provider how their roadmap works around it.

4. Eltropy publishes an AI playbook for community institutions

Eltropy released a strategy guide claiming AI chat agents handle up to 80 percent of routine member questions and cut call volume 40 percent at institutions on its platform, per The Credit Union Connection. Our take: vendor-published numbers deserve vendor-published skepticism, but the framing is sound: treat response speed as an automation problem before a headcount problem. Ask any vendor quoting these figures for a reference at your asset size.

5. The adoption gap, quantified

Gemineye’s 2026 AI Sentiment Report, covered by CUToday, surveyed executives and staff at institutions between $250 million and $8 billion. AI adoption rated 8 of 10 in importance, yet two thirds of institutions have integrated two or fewer AI processes, and 87 percent of respondents named risks rather than opportunities. Small sample at 30 respondents, but it matches what we see. Our take: the gap between stated priority and shipped work is the whole game right now. The institutions closing it start with one narrow, internal, measurable pilot. That is exactly what this week’s 90-day pilot plan is for.

Get this roundup in your inbox every Thursday: subscribe to the AiForCU newsletter.