Natco CU audited $42M in loans in a single day | 2/19
Plus: AI adoption starts long before picking a tool, stop worrying & lean into AI, and more!
Your employees are already using AI... Your competitors are already using AI…
The only question left is whether you have a plan like this week’s featured Credit Unions.
Today, I cover:
Natco Credit Union audited $42M in loans in a single day
Grow Financial’s SVP starts AI adoption with the foundation
Financial Plus CEO says stop worrying and lean into AI
Read time: 9 minutes
Top Stories
The biggest news this week…
1) Natco Credit Union audited $42M in loans in a single day
Validating whether your underwriters are actually following guidelines used to be nearly impossible. Without a way to audit loan decisions at scale, compliance gaps could go undetected for months. And for a CDFI like Natco Credit Union, where delinquency and charge-off rates are under constant scrutiny, that blind spot carries real risk.
Natco, a $152M Credit Union based in Indiana, partnered with Zest AI and implemented LuLu, a generative AI lending intelligence tool. Loan Service Manager Stephanie Harrison led the charge after watching a product demo and immediately seeing ways it could replace work she was doing manually. She worked directly with the Zest AI team to build a custom compliance report that evaluated loans against internal underwriting guidelines and mapped results back to individual underwriters. LuLu audited $42 million in loans in roughly one day. The result: 99.7% of all decisions were within guidelines — a lovely stat to share with the supervisory committee.
The team didn’t stop there. By analyzing 24 months of charge-off data, LuLu identified the characteristics most associated with default risk. Now, when a loan application carries multiple high-risk factors, staff can step in early with tailored support to help members avoid default. On the reporting side, what used to take CEO Cindy Duke hours of manual data pulls from the NCUA website now runs in seconds. That freed-up time helped the team identify home equity lending as a growth opportunity. Natco made it a priority in 2025 and grew that portfolio by 42%. (link)
2) Grow Financial’s SVP starts AI adoption with the foundation
Most Credit Unions want to move fast on AI. The ones that stumble usually skipped the foundation. Because when your 1) data isn’t clean, 2) your processes aren’t documented, and 3) your teams aren’t aligned across channels, layering AI on top just automates the mess.
Emily Nichols, SVP of Enterprise Operations at Grow Financial Federal Credit Union, has spent 25 years watching Credit Unions evolve. She built Grow’s analytics team in 2012, before most Credit Unions were even thinking about data infrastructure. That foundation is now shaping how Grow Financial approaches AI. In 2026, the Credit Union plans to replace its contact center platform with an AI-driven solution. But Nichols isn’t treating it as a technology swap. She’s treating it as an experience overhaul that touches the contact center, digital, and branch channels simultaneously.
Her framework for getting there: start with a proof of concept on a specific use case, validate it, then scale. Don’t take the “rip off the band-aid” approach. Run the crawl-walk journey first. She’s applying the same thinking to account opening to deliver a consistent experience, whether a member calls, walks in, or goes online. (link)
3) Financial Plus CEO says stop worrying and lean into AI
Your employees are already using AI. The question is whether you’re training them or ignoring it.
Brad Bergmooser, President and CEO of Financial Plus Credit Union in Flint, Michigan, found out the hard way. Before his team had a policy or strategy in place, employees were already using AI tools to write performance reviews. His response wasn’t to shut it down. It was to reframe the whole thing. “We shouldn’t be penalizing our employees for utilizing a large language model. We need to train them.” The output, he says, is only as good as the prompt. So that’s where Financial Plus is focusing first.
On the operations side, Bergmooser sees two immediate wins: routing routine contact center calls away from agents so staff can focus on complex member issues, and using AI to handle clear-cut auto loan decisions so officers can spend time on harder cases. For a Credit Union with a unionized frontline workforce, he’s careful to frame it the right way. “It’s not a reduction in workforce mechanism. It’s a tool that, as good employers, it’s our job to upskill our workforce to utilize.”
His bigger concern is what comes next. Agentic AI, he warns, won’t just answer questions. It will act. “By the time this interview is over, an agentic AI feature will have moved a money market because a competitor is offering a higher rate.” His advice to Credit Unions: collaborate, stay at the table, and stop using smaller budgets as an excuse to wait. (link)
Tips & Use Cases
Learn to apply AI…
Small Credit Unions have no excuse to skip AI adoption: Dr. Lamont Black of Wide Open Ventures says Credit Unions under $500M in assets can access tools like Microsoft Copilot or vendor-built platforms without an in-house tech team. The real barrier is board-level fear, not budget. Black recommends starting with a formal AI risk policy to create a safe sandbox that gets leadership comfortable enough to experiment. (link)
Fed Governor Barr warns AI will reshape rates, jobs, and risk models: Barr outlined three scenarios for Credit Unions: productivity gains, an automation-driven jobless boom, or a capital spending surge that shifts risk to the financial system. Most finance firms are still experimenting while the Fed keeps rates elevated as it monitors AI’s economic impact. (link)
68% of staff using AI at work are hiding it from management: Meanwhile, sensitive data appeared in over 4% of generative AI prompts in Q2 2025. Purpose-built tools like Agent IQ limit responses to approved internal sources and can be deployed in two weeks. (link)
Only 11% of businesses have seen ROI from AI pilots so far: Rival’s Eric Fulwiler says Credit Union marketers have shifted from “what’s my AI strategy?” to “we’ve tested it and we’re still waiting on results.” He warns that as AI agents increasingly handle search and purchasing on behalf of users, Credit Unions need to optimize their content for LLMs, not just people. (link)
How Vertice AI helps Credit Unions run lookalike member targeting: The platform pulls existing core data to model member behaviors across hundreds of attributes, then scores prospects from Experian to build targeted acquisition lists. Education Credit Union used it to narrow outreach from tens of thousands of prospects to a focused few thousand, cutting the $500-per-member acquisition cost significantly. (link)
Uptiq CEO says AI should flip the 75/25 ratio of internal work to member time: Uptiq’s agents work across the core, LOS, CRM, and documents simultaneously, handling tasks like loan intake and policy compliance without human involvement. Institutions can see working agents in days, with full core integration adding a few months. (link)
30% of Credit Unions use AI but most lack formal governance: Since ChatGPT’s launch, AI has expanded from back-office scoring to drafting delinquency letters, marketing content, and self-improving chatbots. That shift creates legal, compliance, and ethical risks that boards and executive teams now need to formally address. (link)
Diversified Members Credit Union selects Scienaptic AI to modernize credit underwriting: The Michigan-based Credit Union will use the platform to evaluate members beyond surface-level credit data, aiming for faster and more inclusive lending decisions. (link)
AI-driven fraud hit a record 3.31M cases in the UK in 2024: Synthetic identity attacks drove a 12% year-over-year increase, with over 35% of UK financial institutions reporting AI-related fraud in Q1 2025. Geolocation data, name-to-address matching, and contact data validation can close the gaps fraudsters exploit. (link)
AI is now the top strategic priority for Credit Unions in 2026: CSI’s annual executive survey found AI ranked first across institutions of all sizes, with cybersecurity, digital payments, and core modernization increasingly framed as AI enablers rather than standalone goals. The main obstacle is infrastructure with legacy cores and fragmented data. (link)
Only 1 in 3 Credit Union members feel understood by their institution: White Clay CEO Mac Thompson says most CUs measure satisfaction in ways that confirm what they want to hear, rather than what members actually experience. He argues big banks have a generational lead on AI because they train models on their own internal data, while Credit Unions slapping generic AI onto poor data infrastructure will see little return. (link)
Most banks can’t define their AI strategy because they can’t define their data strategy: Cornerstone Advisors’ Ron Shevlin says Credit Unions are stalling AI implementations by talking about “data” without specifying whether they mean customer, transactional, operational, or market data. His recommendation: hire a chief data and AI officer who is a change agent first, not a technical expert. (link)
Star One Federal Credit Union uses AI to train staff and redirect member calls: New CEO Minal Gupta is integrating AI across voice and digital channels to handle member requests without live agents, and using large language models to train non-technical staff on new processes. (link)
Only 43% of Credit Unions list AI as a top spending priority, trailing banks by 25 points: American Banker’s 2026 Predictions survey found national banks lead AI investment at 68%, followed by regional banks at 54% and community banks at 44%. Credit Unions ranked last despite facing the same technology disruption risks. (link)
NatWest saved 70,000 hours in one year by deploying AI across 60,000 employees: Automated call summaries, complaint responses, and AI coding tools now write 35% of the UK bank’s code, contributing to £100 million in freed investment capacity. In 2026, NatWest is rolling out an agentic financial assistant to 25,000 customers and real-time fraud resolution through natural language conversations. (link)
51% of consumers already use AI tools for financial questions: J.D. Power research shows savings strategies, credit scores, and investing are the top financial queries going to tools like ChatGPT, yet only 27% of bank and Credit Union customers say they currently receive advice from their institution. Virtual assistants close the gap, with users scoring 35 points higher in satisfaction than non-users. (link)
TD Bank’s AI survey finds 89% of Americans are comfortable with AI: TD’s Head of AI Ted Paris says consumers are confident and curious, but cautious about privacy. TD addresses this by keeping humans in the loop before any AI output reaches customers. TD also built a contact center knowledge tool that pulls answers from hundreds of policies in real time to give every agent the same consistent response. (link)
ABNB Federal Credit Union rolls out Eltropy AI voice assistant to eliminate member hold times: The CU consolidated 7 to 8 separate vendor solutions into Eltropy’s single platform over 2025, adding chat and texting before launching AI voice in December. The assistant, nicknamed “Digi” by staff, handles routine account inquiries so employees can focus on complex member requests. (link)
Goldman Sachs deploys Anthropic-powered AI agents in core finance operations: The firm spent six months embedding Anthropic engineers inside its tech teams to build agents handling transaction reconciliation, trade accounting, and client onboarding. It reflects a broader shift among large banks toward agentic AI in structured finance functions, with Credit Unions facing the same automation pressure from larger competitors pulling further ahead. (link)
Agent IQ’s CMO says AI should handle the transactional so staff can focus on the emotional: Matt Phipps argues Credit Unions are losing their relationship banking edge as members shift to digital, and the fix is not less AI but smarter deployment. Agent IQ’s Smart Assist tool lets staff search institution-approved policies in natural language, reducing the time spent hunting for answers so more time goes to the member. (link)
Credit Unions are leading banks on GenAI but data quality will determine who sees results: Cornerstone Advisor’s 2026 report finds 59% of Credit Unions have deployed GenAI vs. 49% of banks, only 8% of Credit Unions have no GenAI plans vs. 23% of banks, and 6 in 10 Credit Unions pursuing agentic AI are targeting contact centers first. (link)
California bank cuts credit memo prep time 63% using Uptiq’s AI lending platform: The unnamed CRE and C&I lender automated document intake, financial spreading, and covenant tracking across its full credit lifecycle. The result was a 47% reduction in manual document review and 36% faster financial data extraction, freeing underwriters to focus on risk analysis instead of data entry. (link)
Funding Spotlight
Where the money is flowing for innovation…
Anthropic raises $30B Series G to scale Claude across enterprise AI: GIC and Coatue led the round, valuing Anthropic at $380B, to fund the research, infrastructure, and product development behind Claude. Run-rate revenue has reached $14B, growing over 10x annually for three consecutive years. (link)
Uptiq secures $25M to bring production-ready AI to Credit Unions and banks: The round was led by Curql, with Uptiq’s platform delivering pre-packaged AI applications across lending, underwriting, and wealth management for 140 financial institutions. Funds will expand its Qore orchestration platform so teams can deploy financial AI in days rather than months. (link)
Bracket raises $7M to scale AI treasury platform for mid-market businesses: Macquarie Group and Blackfinch Ventures led the round to fund expansion into Europe and Australia, with Bracket’s platform centralizing bank accounts, automating FX workflows, and delivering real-time treasury intelligence without legacy system costs. The company grew revenue 600% YoY in 2025. (link)
Bretton AI raises $75M Series B to expand AI compliance agents across regulated financial institutions: Sapphire Ventures led the round, with Bretton AI’s agents handling KYC, AML, and sanctions investigations in minutes rather than days. Clients have saved over $10M in compliance headcount costs and eliminated 195,000 hours of manual work. (link)
Reco raises $30M Series B to secure AI tools inside enterprise SaaS environments: Zeev Ventures led the round, with Reco’s platform continuously monitoring SaaS environments to track AI apps, agent permissions, and data flows across tools like Salesforce, ChatGPT, and Copilot. The company grew 400% YoY in 2025 on an already expanded base. (link)
ZAST.AI raises $6M Pre-A to eliminate false positives in code vulnerability detection: Hillhouse Capital led the round, with ZAST.AI’s platform automatically generating and executing proof-of-concept exploits to confirm whether vulnerabilities are real before flagging them. In 2025, the platform identified hundreds of zero-day vulnerabilities, resulting in 119 CVE assignments across projects, including Microsoft Azure SDK and Apache Struts. (link)
Samaya AI secures undisclosed funding to scale AI agents for investment workflows: NVentures and Databricks Ventures backed the round to expand Samaya’s Agent Control Plane, a purpose-built architecture that lets financial institutions design, run, and govern AI agents across earnings analysis, scenario modeling, and real-time market decisions. The platform is already in production with 10,000+ professionals at one of the world’s largest banks. (link)
Keeping up with Tech
The latest in fintech and tools…
Commonwealth Credit Union and Zest AI launch CUSO to bring AI lending to small Credit Unions: CU Lending Collective gives smaller institutions access to a custom AI credit scoring model covering auto loans, personal loans, and credit cards, with built-in fair lending compliance and hands-on operational support. Commonwealth CU exceeded 14% loan growth in 2025 after deploying Zest AI, while maintaining strong delinquency and charge-off performance. (link)
Multimodal joins Filene’s 2026 FiLab to help Credit Unions test agentic AI in lending and member services: The program runs January through July 2026, delivering sandboxed proofs of concept in account opening, consumer lending, and member assistance workflows without requiring core system integration. Participating Credit Unions will receive vendor-agnostic implementation roadmaps and ROI modeling frameworks. (link)
Anthropic releases Claude Sonnet 4.6 with 1M token context window and major coding upgrades: The model is now the default on claude.ai Free and Pro plans, with users preferring it over the previous frontier model 59% of the time in Claude Code testing. Claude in Excel now connects directly to S&P Global, PitchBook, Moody’s, and FactSet via MCP, pulling financial data without leaving the spreadsheet. (link)
Anthropic and Infosys partner to build AI agents for financial services: The collaboration integrates Claude models and Claude Code with Infosys Topaz to help regulated industries automate compliance reporting, risk detection, and legacy system modernization. Financial services use cases include faster risk assessment, automated compliance reporting, and personalized client interactions. (link)
OpenAI releases GABRIEL to turn unstructured text and images into quantitative data: The open-source Python toolkit uses GPT to apply researcher-defined measurements consistently across thousands of documents, returning a score for each one. It requires minimal technical background and is designed for economists, social scientists, and data scientists working with qualitative data at scale. (link)
OpenAI launches Lockdown Mode and Elevated Risk labels in ChatGPT to guard against prompt injection attacks: Lockdown Mode restricts how ChatGPT interacts with external systems, limiting web browsing to cached content and disabling features that could expose sensitive data. It is available now for ChatGPT Enterprise, Edu, Healthcare, and Teachers plans, with consumer availability planned for coming months. (link)
Cemex deploys AI financial agent LUCA Bot to give executives instant access to internal data: Built on Microsoft Azure OpenAI, the tool processes 120+ KPIs across regions, countries, and plants, answering natural language queries that previously required hours of searching through reports. The agent handles 400-500 queries per month with 92% accuracy on data retrieval. (link)
Provenir launches Decision Intelligence platform with agentic AI for credit risk decisioning: The platform connects data, models, and decisioning in one system, cutting strategy testing time from months to weeks while providing real-time, personalized risk decisions. Financial institutions can now access OpenAI and Anthropic models through pre-integrated APIs or private AWS Bedrock instances directly within Provenir’s workflows. (link)
OpenAI releases GPT-5.3-Codex-Spark for real-time coding with 1,000+ tokens per second: Built on Cerebras’ low-latency hardware, the model delivers near-instant responses for interactive coding tasks like targeted edits and logic refinements. It is rolling out now as a research preview for ChatGPT Pro users in the Codex app, CLI, and VS Code extension. (link)
In Other News
Related news you can learn from…
AI should not shrink the web’s freedom to learn (link)
Salesforce manager oversees AI agents like human employees (link)
OpenAI team ships a real product with zero human-written code (link)
FinovateEurope 2026 spotlights AI urgency and quantum risk for banks (link)
Vizo Financial spotlights AI-driven risk at its 2026 Risk Management Conference (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!