Welcome to AI for FIs, from Dixon Strategic Labs. Each week, this newsletter curates critical developments in agentic AI and explains why they matter for your credit union.
More than half of credit unions have deployed AI tools. Only 16% have a roadmap. A study of 809 institutions found that the gap between adoption and strategy costs 428 basis points of ROE, with smaller institutions hit nearly four times harder than large ones. Meanwhile, Lake Michigan Credit Union is processing 60% more loan files with the same team. Closing that gap is the work ahead.
For the most ambitious of you, we also broke down two major 2026 AI studies from Cornerstone Advisors and NVIDIA side by side. Full analysis at the bottom.
Ready set let’s go:
American Banker · Feb 3, 2026
Accenture warns that AI prompt engines will let members optimize idle cash returns across institutions. Even a small pricing shift could put 22% of U.S. banking pretax income at risk. PayPal and BBVA are already partnering with Open AI platforms.
Why it matters: If members start asking agents to find better deposit rates, credit unions relying on low-cost deposits face real margin pressure.
(Note: If you hit a paywall on the American Banker link, the author Michael Abbott posted a solid summary of the core thesis here on LinkedIn.)
The Financial Brand · Feb 13, 2026
Cornerstone Advisors' 2026 banking report finds 59% of credit unions have deployed GenAI, and 17% are exploring agentic AI. But analyst Ron Shevlin warns that weak internal data quality will undermine AI efforts. The report also highlights Model Context Protocol (MCP) as a critical enabler for agentic AI.
Why it matters: The 17% exploring agentic AI face the same bottleneck as everyone else: there is no AI strategy without a data strategy.
Board of Governors of the Federal Reserve System • Feb 16, 2026
Fed Governor Michael Barr called AI a likely "general-purpose technology" and disclosed that the Fed itself is using LLMs internally. One project cut database migration time by 50% and caught 30% more issues during testing. Barr warned that early-career workers in AI-exposed roles are already seeing employment declines, and urged investment in retraining.
Why it matters: Credit union leaders can take the Fed’s numbers straight to a board conversation about what's realistic to expect from AI adoption.
📡 On Our Radar
Engage fi research finds only 16% of FIs have an AI roadmap despite 67% already implementing AI tools.
A study of 809 FIs found GenAI adoption caused a 428-basis-point ROE decline, with smaller institutions hit nearly four times harder.
Lake Michigan CU's agentic AI disclosure tool lets the same staff handle 60% more mortgage files without adding headcount.
BECU acquired EarnUp's AI team and is building an AI financial adviser called Becca for its 1.5 million members.
NIST launched an AI Agent Standards Initiative covering interoperability, security, and identity for autonomous agents.
Ethan Mollick published a guide to choosing AI in the agentic era, covering models, apps, and harnesses across OpenAI, Anthropic, and Google.
⚙️ Tools & Vendors
Sphinx raised $7.1M in seed funding for browser-native AI agents that automate AML and KYC compliance workflows.
Creatio launched six banking AI agents for referrals, retention, onboarding, and loan servicing, each deployable in 10 weeks.
Claude Sonnet 4.6 topped the Finance Agent benchmark at 63.3%, beating every competitor including the more “sophisticated” (and expensive) Opus 4.6.
🔬 Deep Dive
Cornerstone and NVIDIA Agree on Where AI Agents Are Headed
TL;DR:
Two major 2026 surveys from Cornerstone Advisors and NVIDIA tell a similar story
AI adoption in financial services nearly doubled in one year
42% of the broader industry is already moving on agentic AI. Credit unions are at 17%.
Top use cases: knowledge management, fraud detection, back-office ops
Biggest blocker: data quality
The Cornerstone report is covered above. NVIDIA also dropped their "State of AI in Financial Services" survey this month (839 respondents across the broader financial services world). Read side by side, the two reports reinforce each other.
The Numbers Line Up
Cornerstone's numbers on credit union AI adoption are above. NVIDIA's tell the same story from the broader industry: 65% of financial services firms are actively using AI, up from 45% in 2024.
On agentic AI specifically, NVIDIA found 42% of financial services organizations are already using or assessing AI agents. Compare that to Cornerstone's 17% for credit unions. That gap is worth watching.
Cornerstone: Credit unions' AI status across four technology categories. 17% have deployed agentic AI. Another 25% plan to invest this year.
Where Agents Are Landing First
Three use cases keep showing up across both reports:
Knowledge management. NVIDIA's respondents cited it as the #1 agentic AI workflow (56%).
Fraud detection. 75% of Cornerstone's respondents see agentic AI opportunity in pattern recognition and early fraud detection. NVIDIA found 24% of consumer finance respondents already cite fraud and AML as a top-ROI use case.
Back-office operations. NVIDIA found 52% of agentic AI adopters using agents for internal process optimization. Cornerstone found credit unions targeting operations (51%), lending (57%), and IT (47%) for deployment in 2026.
NVIDIA: These are top five workflows where financial services firms are deploying AI agents. Knowledge management leads at 56%.
The pattern across all of these: repetitive, high-volume, rule-bound work.
The Data Problem, Quantified
NVIDIA says data-related issues are the #1 challenge in AI adoption (40% of respondents, up from 33% last year). Privacy, sovereignty, data scattered across systems.
NVIDIA: Data-related issues remain a top barrier to AI adoption, led by performance reliability and lack of internal capacity.
Cornerstone put a number on it. They rated how effectively financial institutions use data across five business functions: strategic planning, sales/marketing, credit analysis, operational delivery, and data access/analysis. Each function scored out of 100. The average total across all five was 241 out of 500. Sales and marketing was the worst, with two-thirds of institutions scoring below 50.
The institutions seeing returns (NVIDIA says 89% report both increased revenue and decreased costs) have their data house in order.
Budgets Are Moving
Cornerstone: 83% of credit unions plan to increase tech spending in 2026. 22% plan increases above 10%.
NVIDIA: nearly all respondents said AI investment will increase or stay flat. 44% expect increases above 10%.
The spending priorities shifted too. NVIDIA found the #1 AI priority is now "optimizing AI workflows and production cycles." Many financial institutions have moved past pilots and are investing in making what they've already built work better.
What to Do About It
Mind the gap. 42% of the broader industry is already using or assessing AI agents. Credit unions are at 17%, with 25% more planning to invest this year.
Start with your messiest workflows. Knowledge management, fraud review, document processing. That's where agents deliver today.
Fix your data before you buy AI tools. If your core, digital channels, and CRM don't talk to each other, no AI agent can fix that.
Both reports are free. Get em while they’re hot:
Whether you're wrestling with the gap between AI adoption and strategy, weighing the ROE risks that hit smaller institutions hardest, or figuring out how to turn early pilots into the kind of results Lake Michigan CU is seeing, I'd love to hear what's shaping your thinking.
Drop me a note and say hello at [email protected].
How this newsletter is made: Brent curates links, Claude (Opus 4.6) generates summaries and the intro, Brent edits and publishes on Beehiiv. ⚡ Alakazam ⚡.
Know someone who should be reading this? Send them the subscribe link:
