Credit unions are stuck running loan operations on software that predates the iPhone, and Fuse just raised $25M to drag them into the agent economy.
The Signal
Credit unions process $750 billion in loans annually on loan origination systems (LOS) built in the 1990s. These aren't just old. They're actively hostile to modern automation. Every loan application flows through manual underwriting steps that could run autonomously. Every document gets reviewed by humans who could be supervising AI agents instead.
Fuse is building an AI-native LOS from scratch. Not bolting machine learning onto legacy code. Building for a world where agents handle document verification, risk assessment, and compliance checking in seconds instead of days. The $25M Series A from undisclosed investors funds the core platform build. But the $5M "rescue fund" is the sharper play. That's cash to help credit unions escape vendor lock-in from legacy providers who charge extraction-level fees to institutions that can't afford to switch.
The timing matters. Credit unions are member-owned cooperatives, not profit-maximizing banks. They serve 140 million Americans, often in communities big banks ignore. But they're getting crushed on efficiency. The average credit union loan takes 3-5 days to process. Digital-first lenders do it in hours. That gap is an existential threat when members expect Stripe-speed everything.
Fuse is betting that credit unions will trade their legacy systems for agent-powered infrastructure if someone covers the switching costs. They're probably right.
The Implication
Watch which credit unions sign on first. They'll become testing grounds for how well AI agents can handle regulated financial operations at scale. If Fuse works, every community bank and regional lender faces the same rebuild-or-die question. For workers in loan operations, this is the preview: your job isn't disappearing, but it's becoming agent supervision instead of manual processing.
Source: TechCrunch AI