How Mid Oregon automates 70% of loan decisions | 4/2
Plus: Go on offense to win Gen Z, Lake Trust tackles internal AI fear, and more!
AI has raised the bar for what members expect from their Credit Union…
Whether that’s faster loan decisions, personalized financial advice, or staff who always have the right answer.
This week, I cover:
Mid Oregon Credit Union automates 70% of loan decisions
How Credit Unions can go on offense to win Gen Z with AI + data
Lake Trust Credit Union tackles internal AI fear
Read time: 9 minutes
Top Stories
The biggest news this week…
1) Mid Oregon Credit Union automates 70% of loan decisions
Kevin Cole, CEO of Mid Oregon Credit Union, says they deployed Zest AI to handle the majority of its loan decisioning. He describes the platform as a more advanced version of predictive AI, the same concept behind FICO scores that Credit Unions have used for years.
Cole describes the loan pool in three tiers: the bottom and top are easy decisions, and about half fall somewhere in the middle. Today, Zest AI is decisioning about 70% of Mid Oregon’s loan requests on a fully automated basis. The remaining 30% still go to human underwriters when marginal or complex enough to require judgment.
The result? Mid Oregon reassigned two employees from underwriting to other roles at the Credit Union rather than letting them go. The Credit Union has never laid off an employee in its history since 1957, and Cole says he has faith they’ll navigate this transition without abandoning that approach.
Beyond underwriting, Mid Oregon is using Eltropy Voice+ as one of its first conversational intelligence tools to serve members. Cole says the Credit Union has been learning with Eltropy through the process and making good progress.
At the same time, Mid Oregon has restricted staff access to ChatGPT and similar tools on its network for data security purposes. Cole says that restriction is only a limited time window, because eventually it will make the Credit Union less competitive than it wants to be. (link)
2) How Credit Unions can go on offense to win Gen Z with AI + member data
More than 6 in 10 Gen Z members are open to using AI for “what if” financial planning, according to Velera’s CU Growth Outlook 2025. Tools like ChatGPT and Gemini are already resetting what good financial service looks like for this generation. They expect their financial partners to anticipate needs, model scenarios, and deliver personalized guidance in real time. (link)
The problem, according to Lee Wetherington, Senior Director of Corporate Strategy at Jack Henry, is that most Credit Unions are missing the data to do it. Credit Unions typically hold only 20 to 25% of their average member’s financial data, limited to whatever accounts that member has at the Credit Union. Without a fuller picture, the most compelling AI use cases simply are not possible.
Wetherington says the Credit Unions best positioned to win are not waiting on regulators. They’re proactively asking members for permission to aggregate financial data from third-party relationships back to the Credit Union, so they actually know their members and can use that data to create new AI-driven products and services. The regulatory chaos around open banking, he says, is a window for aggressive Credit Unions to go on offense. (link)
3) Lake Trust Credit Union tackles internal AI fear
Krassi Kassabov, Chief Strategy Officer at Lake Trust Credit Union, dedicates a quarterly all-hands meeting to reviewing prior results and setting direction for the next quarter. Recently, he dedicated a full hour of that meeting to a single topic: AI. He posed three questions to his team:
What excites you about AI?
What scares you?
What can we do together to get over that hurdle?
Several team members were vocal about believing AI would take their jobs. Kassabov’s response was to give it a try. He draws on his own experience immigrating from Bulgaria, moving 5,000 miles away without knowing anyone or the language, to make the point. The way he got over that fear was by showing up and discovering that different does not mean bad. He tells his team the same thing about AI.
His clearest example comes from marketing, where copywriters worried AI would take over message creation. Kassabov said he needs them even more now to make sure the branding and tone of voice are right. AI can draft the message, but the human still shapes it, and work that once took hours now takes minutes.
Lake Trust is already using Posh for conversational AI to handle basic member questions, so team members can focus on more complex conversations and building emotional connections with members. Kassabov said the Credit Union is also beginning to test AI in marketing and has introduced an internal knowledge system that helps get the right content to team members at the right time.
Kassabov is clear about what AI cannot do. It cannot build trust with members the way a human can, and it cannot supply the qualitative judgment that connects data to real member situations. (link)
Tips & Use Cases
Learn to apply AI…
Genesee Co-op Federal Credit Union bets on AI to build in-house: Dan Apfel, CEO of the Rochester-based CDFI Credit Union, believes AI could let a small team of engineers build what once required a hundred. He sees it as a path for small Credit Unions to develop their own tools instead of depending entirely on outside vendors. (link)
Firefighters First Credit Union uses AI to scale emotional member bonds: Jeffrey Staw, Chief Information and Innovation Officer, says AI helps the Credit Union determine when a human interaction is needed versus when automation serves members better. As the nationwide firefighter-only Credit Union grows, that distinction is key to preserving the personal connection members expect. (link)
Vantage Credit Union uses AI to coach collections staff in real time: Matt Stegall, AVP of Credit Resolution and Recovery, says the Credit Union is analyzing member phone calls with AI and generating information cards to help employees respond faster and more effectively. He sees AI as a productivity tool that frees staff to focus on the high-empathy situations machines can’t handle. (link)
Agentic commerce is headed for your disputes queue: As AI agents begin making purchases on behalf of members, Credit Unions face a new dispute challenge: not whether a transaction was authorized, but whether the member actually intended it. AI-assisted dispute platforms can cut resolution times from 30 days to as few as 10, but most CUs haven’t built the infrastructure to handle it yet. (link)
717 Credit Union plans to process loans in under two minutes: President and CEO John Demmler says the Ohio Credit Union expects 95% of transactions to be digital within two years. The goal is to use digital tools for routine transactions while freeing branch staff to focus on deeper member financial conversations. (link)
Solaris cuts 20% of staff in pivot to AI-native banking: The German fintech, which holds a full banking license, is eliminating 80 of its roughly 400 employees as new CEO Steffen Jentsch shifts the company toward an AI-native architecture. (link)
Microsoft says cloud migration is the foundation for AI in financial services: Laurent Verdier, Microsoft’s CTO for Worldwide Financial Services, argues that legacy systems are limiting what AI can do across lending, fraud, and compliance. Credit Unions still running on older infrastructure risk falling behind as AI-ready cloud platforms become the baseline for competitive operations. (link)
Posh offers a build-vs-buy framework for Credit Union AI strategy: Build what differentiates your Credit Union and buy what accelerates it. Focus internal resources on member experience while partnering for infrastructure, compliance guardrails, and model integration. Institutions that try to build everything risk spending more time on maintenance than innovation. (link)
Casap’s dispute automation cuts errors 63% in Filene pilot: Credit Unions participating in Filene’s 2025 FiLab evaluation of Casap’s automated fraud dispute platform also reported a 122% higher ease-of-use score compared to prior systems. Staff confidence in compliance improved and member complaints declined during the pilot period. (link)
NCUA updates examination approach as Credit Unions adopt AI: Acting Director of Examination and Insurance Amanda Parkhill told a House subcommittee the agency is integrating fintech expertise into its supervision process. Credit Unions are actively using AI for lending, fraud detection, and member services, and proposed legislation would create regulatory sandboxes for testing new products with advanced compliance assurance. (link)
AI-generated paystubs now appear in roughly 1 in 5 auto loan applications: Generative AI tools have made document fraud accessible to anyone, not just sophisticated criminal networks. Credit Unions using multilayered verification that combines metadata analysis, cross-document consistency checks, and shared fraud networks can catch fakes at application time rather than discovering losses months later. (link)
Only 11% of banks have deployed AI responsibly, SAS report finds: A global SAS and IDC survey of 2,375 IT and business leaders found nearly half of banks are either underusing reliable AI or over-relying on systems that haven’t been validated. Organizations that prioritized trustworthy AI were 60% more likely to report doubling their return on AI investments. (link)
UW Credit Union sponsors free 7-part AI literacy video series: Steph Harrill Kyle, Director of Financial Inclusion and Community Engagement, says the Credit Union partnered with the Universities of Wisconsin to launch ASAP, a free series teaching members how AI works and how to use it safely. (link)
Merrill Lynch rolls out AI meeting tools to 25,000 advisors: Bank of America integrated AI into Salesforce and Zoom to help advisors prepare for client meetings, summarize discussions, and handle follow-up tasks. Meeting prep that once took 45 minutes now takes a few minutes, freeing advisors to focus on client relationships instead of administrative work. (link)
CFO-led AI projects succeed at far higher rates, Fortune study finds: Only 2% of companies have their CFO driving AI value, but when they do, 76% achieve meaningful results. That rate outperforms every other C-suite role, including Chief AI Officers and CTOs. (link)
LHV Bank pilots agentic AI for email-based customer support: Interim CEO Kris Brewster says the bank is testing Gradient Labs’ agentic AI in a controlled environment to improve response quality and turnaround times on customer inquiries. The pilot emphasizes human oversight, explainability, and auditability throughout. (link)
AI drives measurable ROI across three areas in financial services: Personalized AI engagement improves cross-sell conversion rates up to 25% and reduces churn up to 20%. AI-assisted service operations cut average handle times 20-40% and reduce cost-per-contact 30-40%. In lending and onboarding, AI document processing can cut preparation time in half. (link)
i2c captures up to 40% of fraud while keeping friction at 0.5%: Matt Pearce, VP of Fraud Risk Management at i2c, says the platform embeds AI directly into the transaction authorization process rather than adding it as a bolt-on. Machine learning builds behavioral profiles within 60 to 90 days, and consortium intelligence draws on fraud patterns across i2c’s global platform to benefit smaller Credit Unions that lack dedicated fraud teams. (link)
Most financial institutions can’t reconstruct an AI decision for regulators: Experts across multiple firms say the gap isn’t the AI itself but the surrounding infrastructure. Credit Unions deploying AI in compliance workflows should ensure every decision has a traceable audit trail, a documented policy version, and a clear record of human oversight before a regulator ever asks for it. (link)
Agentic AI cuts compliance alerts by up to 82% and speeds onboarding: A ComplyAdvantage survey found 54% of firms linked agentic AI to improved efficiency in customer screening and case investigation, while 51% reported better member onboarding experiences. Nearly 80% of organizations still take more than five minutes to clear a single sanctions alert, a gap agentic AI is built to close. (link)
Citizens Bank targets 50% of call center calls answered by agentic AI: President Brendan Coughlin says the $226B bank is piloting agentic voice AI that can handle balance checks, fraud claims, address changes, and debit card replacements end-to-end without a human. The three-year, $300M “Reimagine the Bank” project targets $450M in earnings value creation by Q4 2028. (link)
Funding Spotlight
Where the money is flowing for innovation…
Spade raises $40M Series B to turn raw transaction data into AI-ready intelligence: The round will help Spade expand beyond transaction enrichment into a full payments intelligence platform. The company processes 1.9B transactions daily at 99.9% merchant coverage, helping financial institutions reduce disputes and power agentic AI workflows. (link)
Qodo raises $70M Series B to govern AI-generated code: Led by Qumra Capital, the round brings total funding to $120M and will help Qodo scale its code review and governance platform for enterprises. Platforms like Qodo address a growing concern: AI can write code fast, but without quality controls, it introduces risk into production systems. (link)
Grand raises $5M pre-seed to replace static credit reports with real-time risk signals: The round will fund Grand’s AI-powered trust network that monitors trade partner risk and stability continuously, rather than relying on data that is months out of date. The company targets B2B trade industries where outdated credit decisions contribute to thousands of insolvencies annually. (link)
Notch raises $30M Series A to automate end-to-end workflows in regulated industries: ARR grew 12x over the past year as insurers and financial institutions adopted the platform’s AI agents for claims intake, document processing, and policy servicing. (link)
Keeping up with Tech
The latest in fintech and tools…
UiPath rolls out AI agents for sanctions screening and loan origination: Valley National Bank automated 61% of sanction-hit reviews and now processes 14,000 alerts per month using UiPath’s transaction screening agent. Suncoast Credit Union is using the platform to ingest loan documents, compare them against systems of record, and surface quick insights for review teams. (link)
Flowdock launches AI back-office automation platform for financial workflows: The platform connects to bank accounts, email, and business tools to monitor invoices, contracts, and cash flow in real time, flagging supplier overcharges, missed renewals, and cash shortfalls before they become problems. It integrates with major EU banks via Open Banking APIs and tools like Stripe, HubSpot, and Slack. (link)
FinTech Studios opens its financial intelligence engine to self-serve users: The platform continuously monitors billions of global data sources across 100+ languages and delivers domain-specific insights without manual searching or prompting. Previously available only through six-figure enterprise contracts, it now offers free and tiered access for individual users and teams. (link)
ExeQution Analytics launches hallucination-resistant AI assistant for trading: Eolas converts natural language queries into API calls rather than generating free-form responses, preventing fabricated answers by limiting the tool to approved functions on approved data. Trading, quant, and IT teams can interrogate real-time and historical market data without pulling reports through intermediaries. (link)
CUBE integrates regulatory intelligence with Microsoft Azure for automated compliance: The partnership gives financial institutions an automated way to track and interpret thousands of regulatory updates monthly across multiple jurisdictions. CUBE’s platform is now available directly on the Microsoft Marketplace for faster deployment. (link)
Stripe processed $1.9T in volume and launches AI billing and stablecoin tools: The payments giant’s new AI billing tool tracks LLM usage costs in real time, while its x402 integration lets AI agents send and receive stablecoin payments autonomously. Reports also suggest Stripe is exploring a PayPal acquisition, which would add 439 million active accounts to its network. (link)
Ani Tech launches free AI cash flow modeling app for financial advisors: The app lets advisors build and compare financial scenarios using natural language, with an AI agent called Flo monitoring client goals in the background without manual prompts. Ani Tech says agentic AI can automate 75% of non-client-facing work, potentially cutting client onboarding costs from £800 to £200. (link)
Mastercard completes live agentic payment transactions across Latin America: AI agents executed real purchases across debit and credit cards on Mastercard’s existing network, covering groceries, books, and digital goods without requiring new payment rails. The company’s Verifiable Intent framework creates a tamper-resistant record of what a cardholder authorized, giving financial institutions a clear audit trail when AI initiates a transaction. (link)
Njordium adds self-learning AI fraud detection to its vendor management platform: The module extracts and validates invoices across PDFs, email attachments, and XML feeds, flagging phantom services, non-existent products, and prices deviating from market norms. High-confidence detections route automatically into AML and financial crime prevention workflows before funds leave the organization. (link)
Flanks debuts AI co-pilot for wealth advisors managing multi-custodian portfolios: The tool lets advisors query complex family structures, cross-border holdings, and private equity capital calls in plain language, drawing on data from 600+ financial institutions across 33 countries. All outputs are descriptive only, with no investment recommendations, keeping the platform within compliance guardrails. (link)
Microsoft 365 Copilot adds long-running agentic workflows with Cowork: Copilot Cowork lets users delegate multi-step work by describing an outcome, with the system creating a plan and carrying tasks forward across files and tools. The update also improves Researcher with a Critique feature that uses separate Anthropic and OpenAI models to draft and review outputs, scoring 13.8% higher on the DRACO deep research benchmark. (link)
Google releases Gemini 3.1 Flash Live for real-time voice AI applications: The updated model delivers lower latency and improved tonal understanding, adjusting responses based on user frustration or confusion in real time. Credit Union contact center developers can access it via the Gemini Live API, with enterprise deployment available through Gemini Enterprise for Customer Experience. (link)
In Other News
Related news you can learn from…
Agentic AI payments and leadership moves dominate March (link)
Perplexity shares lessons from building voice-first AI agents (link)
AI cuts threat detection time 36% when kept in a support role (link)
Australian financial institutions embed AI into AML compliance workflows to cut false positives (link)
Community Corner
Memes and visuals…
Thanks for reading!
Until next week,
— Credit Union AI Guy
Enjoying this newsletter? Forward it to a colleague who might appreciate practical AI insights.
Have a question or want to share your own AI experiment? Reply to this email or connect with us on LinkedIn. We’re building a community of curious Credit Union professionals. Come join the conversation!