AI wants to run your credit union
Back-office banking software is rarely front-page material, but that is exactly where some of the most consequential AI battles are being fought. The latest example is Fuse, which just raised 25 million dollars to rebuild the loan origination system for US credit unions. That sounds niche, yet it points to a bigger shift: AI vendors are no longer just adding smart features on top of existing systems, they are trying to replace the core itself. In this piece I unpack what Fuse is really betting on, why incumbents should be nervous, and why European cooperative banks should be paying close attention.
The news in brief
According to TechCrunch, New York based startup Fuse has raised a 25 million dollar Series A round, announced on 16 March 2026. The round is led by Footwork, with participation from Primary Venture Partners, NextView Ventures and Commerce Ventures.
Fuse was created after its founders, Andres Klaric and Marc Escapa, pivoted from an automotive lending startup in 2023. They are building an AI native loan origination system (LOS) that manages the full lifecycle of a loan for lenders, from application through underwriting to disbursement.
The company is targeting US credit unions that currently rely on aging LOS products. TechCrunch reports that Fuse already claims more than 100 customers and is going after incumbents such as listed vendor nCino and private equity owned MeridianLink. To ease switching, Fuse has set up a 5 million dollar rescue fund, offering the first 50 qualifying credit unions free platform access until their existing LOS contracts expire. Competing AI infused LOS startups mentioned include Casca and Glide.
Why this matters
Loan origination systems are the nervous system of lending. They sit between the customer interface, the credit models and the core banking ledger. If you own the LOS, you sit in the flow of data, workflow and ultimately revenue. This is why Fuse’s move is more strategic than yet another AI tool for underwriters.
For US credit unions, the upside is clear. Many are stuck on systems that take up to a year to implement, come with multi year contracts and offer limited automation. That translates into slow product launches, manual work and high operating costs. An AI native LOS that can be deployed faster and automate large parts of document intake, income verification and decisioning is directly margin accretive.
The losers, at least in the medium term, are the legacy vendors and the ecosystem around them: systems integrators, consultants and bespoke tooling built on top of old platforms. Their business model depends on complexity and multi month projects. AI based automation threatens the billable hour as much as it threatens the licence fee.
There is also a strategic angle. Credit unions have struggled to compete with digital first banks and fintech lenders that can approve loans in minutes. If Fuse can genuinely compress integration timelines and operational overheads, it gives those community institutions a realistic path to modern member experiences without hiring a Silicon Valley sized engineering team.
Of course, this is all contingent on two things: that Fuse’s AI agents are accurate enough for heavily regulated lending, and that regulators are comfortable with how decisions are explained and audited. The technology risk is real, but the distribution opportunity is equally large.
The bigger picture
Fuse is riding three converging waves.
First, the vertical SaaS plus AI trend. In almost every industry, startups are building domain specific software where AI is not a plug in, but the primary interface: think AI driven ERPs for manufacturing, AI CRMs for sales, or AI copilots embedded in legal software. Loan origination is a classic candidate because it is document heavy, rule based and expensive when run manually.
Second, we are entering the replacement cycle for 2000s era banking software. Many LOS and core banking systems were put in 10–20 years ago, often as on premises projects that have been incrementally patched. Cloud native challengers have been nibbling at the edges, but AI has given them a new story: not just cheaper hosting, but a step change in productivity.
Third, regulators are slowly moving from abstract AI principles to concrete supervisory expectations. That actually helps focused vendors. A credit union building its own AI underwriting stack would need to develop explainability, model risk management and monitoring from scratch. A specialised LOS provider can amortise that investment over hundreds of institutions.
Historically, every time a new paradigm appeared in enterprise software, a few vendors managed to turn a back office category into a strategic one. Salesforce did it with CRM, ServiceNow with IT workflows. There is a non trivial chance that the LOS category is having a similar moment: whoever wins in credit unions can later move into community banks, specialist lenders and perhaps even international markets.
At the same time, incumbents like nCino and MeridianLink are not standing still. They have capital, existing integrations and regulatory credibility. Expect them to increase their own AI features and push a narrative of safety and stability versus unproven challengers. The resulting arms race will likely accelerate innovation in a segment that has long been neglected.
The European and regional angle
On paper, Fuse is a US story. In practice, it is a preview of debates that cooperative and regional banks across Europe will face within the next five years.
Europe does not really have the US style credit union structure, but it has close analogues: German Sparkassen and cooperative banks, French mutuals, Italian and Spanish cooperative credit networks, as well as smaller savings banks and credit cooperatives in Central and Eastern Europe. Many of them still run on ageing loan engines tied to domestic core banking systems.
Under the EU AI Act and GDPR, any AI powered LOS used in Europe will have to meet strict standards on transparency, data minimisation and human oversight. That raises the barrier to entry, but it also creates a moat for vendors that can demonstrate compliant, explainable AI. European fintech infrastructure players like Mambu, Thought Machine, Backbase or Temenos are already positioning themselves around composable, API first banking stacks. Adding truly AI native loan origination modules is the logical next step.
For European institutions, the strategic question is not whether to adopt AI in lending, but where to embed it. Do they let LOS vendors become the brains of credit decisioning, or do they keep that intelligence in house and treat LOS as plumbing? The answer will vary by size: a Slovenian or Croatian cooperative with minimal IT capacity will likely lean far more heavily on vendors than a large German bank with its own data science division.
Crucially, the culture of privacy and consumer protection in Europe means that black box AI underwriting will face more pushback than in the US. Vendors inspired by Fuse will have to build tools for granular audit trails, bias detection and the ability for humans to override automated decisions. Getting that right is not a nice to have; it is the price of admission.
Looking ahead
Over the next 12 to 24 months, Fuse will be tested on three fronts.
First, implementation velocity. If the startup can show credible case studies where a mid sized credit union goes live in weeks rather than a year, the reference effect will be powerful. Conversely, a wave of delayed projects would confirm incumbent talking points that core systems cannot be rushed.
Second, regulatory comfort. Even if Fuse initially focuses on relatively simple products such as auto loans, supervisors will want to understand how its AI agents make recommendations, how edge cases are handled and how bias is monitored. A single high profile compliance failure could slow adoption across the entire segment.
Third, economics during a downturn. We are heading into a cycle where credit quality may deteriorate. If institutions using AI native LOS platforms show better loss performance because of richer data and faster reactions, the category wins. If they perform worse, the narrative could turn against automation in credit.
I expect US community banks and specialist lenders to watch credit union deployments closely, with a lag of perhaps two to three years before any serious European pilots at scale. By then, we will also see whether incumbents have successfully retrofitted AI into their platforms, or whether a new cohort of vendors has established itself as the default for smaller institutions.
For founders and product teams in Europe, Fuse is a reminder that the real AI opportunity in fintech is not another chatbot, but deeply embedded systems that quietly run regulated workflows. That is harder, slower and less flashy, but the defensibility is far greater.
The bottom line
Fuse’s 25 million dollar round is less about one startup and more about a land grab for the core of lending. If AI native vendors can prove they are faster to deploy, cheaper to run and at least as safe as incumbents, they will redraw the map of who controls financial infrastructure for smaller institutions. The open question for European cooperative and regional banks is simple: when AI comes for the loan engine, will they lead the change, or have it imposed on them by new competitors?
