Five vendor and market signals from the first two weeks of September, each verified and linkable. Four of the five involve back-office money movement: collections, payments, and balance sheet work. The pattern matters more than any single deal, and we come back to it at the end.

1. SECU, the second-largest credit union in the country, modernizes collections with AKUVO

State Employees’ Credit Union ($60 billion, 3 million members, 275 offices) selected AKUVO’s collections platform to replace manual recovery processes with playbook automation, analytics, and digital member outreach. Our take: when an institution of SECU’s size picks a category, it validates that category for everyone downstream, and collections is a textbook automation target: repetitive, rule-driven, and measurable. Worth noting for smaller shops: the quote came from SECU’s chief credit and revenue officer, not its CIO. Collections automation is a credit decision wearing a technology costume, and the reporting line reflects that.

2. Members 1st puts investment money into Delfi’s balance sheet CUSO

Members 1st Federal Credit Union ($8 billion, 630,000 members) made a strategic investment in Delfi’s CUSO, the AI-native balance sheet intelligence platform that launched its credit union service organization earlier this year with One Washington Financial and Maps Credit Union. Delfi says it has facilitated $100 million in transactions on the platform. Our take: the CUSO structure is the story here. Credit unions co-owning their AI vendors is a governance model with real teeth: aligned incentives, shared data rights, and a seat at the roadmap table. If a vendor pitch matters enough to your balance sheet, ownership questions belong in the conversation. Our contract guide covers what to ask when you are a customer rather than an owner.

3. MDT bundles AI readiness consulting into its core hosting relationship

Member Driven Technologies, the CUSO that hosts the Jack Henry Symitar core for more than 100 credit unions, partnered with OnTrac AI to fold the Ignite readiness program into its consulting arm: opportunity assessment, governance frameworks, and implementation roadmaps. Our take: your core hosting provider selling AI strategy is convenient, and convenience cuts both ways. A partner who already runs your core knows your environment better than any outside firm, and also has an obvious interest in solutions that deepen the hosting relationship. Take the readiness assessment, then pressure-test the roadmap against someone who does not bill you monthly.

4. MessagePay crosses 400 institutions and ships payment prediction

MessagePay, the Lehi, Utah payments and messaging platform, signed its 400th financial institution customer and launched AI Pay+, which forecasts when and how a borrower is likely to pay so collections teams can rank accounts beyond delinquency buckets. The company reports 86 new institutions in 2026 and integrations with Fiserv, Jack Henry, Corelation, CU Answers, and Flex. Our take: payment prediction is a sensible, narrow use of machine learning with a clean success metric, which makes it easy to pilot honestly. The screening question for any tool like this: can the vendor show the prediction lift at an institution your size, on your core? A model trained mostly on $20 billion banks may read a $200 million membership poorly.

5. A Stanford study says the readiness confidence is not supported by the data

Writing in The Financial Brand, FinRank co-founder Shyam Pradheep published results from a faculty-sponsored Stanford GSB study of 46 credit union executives. The gap: 78 percent called AI a source of competitive advantage, yet self-rated AI readiness (3.3 of 5) barely correlated with the strength of underlying data foundations (r = 0.23), and more than half of the executives rated their readiness above their own data visibility and workflow scores. Our take: this is the cautionary item of the week, and it lands on the theme we push constantly. Approved tools, a pilot, and a policy are evidence of experimentation, not capability. If your member data cannot move cleanly between systems, no vendor on this page fixes that for you, and the honest first project is the unglamorous data work.

The pattern

Collections at SECU, collections prediction at MessagePay, balance sheet intelligence at Delfi, and readiness consulting from a core hosting CUSO. The money in credit union AI is flowing to operational plumbing, not chatbots, and the one research item in the list explains why: the institutions writing these checks have learned that data foundations and working workflows decide outcomes. If you are sequencing your own first projects, this week’s companion piece on turning a pilot into a program picks up exactly there, and the vendor landscape pillar holds our full evaluation kit.

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