From Stanford AI labs to banking leadership, Casca co-founder Lukas shares how dual expertise is reshaping small business lending with responsible AI.
Lukas, could you tell us a bit about your professional background and journey – how your experience in AI research at Stanford and in banking led you to co-found Casca?
My path to founding Casca was really shaped by two parallel tracks that eventually converged. The first started early in my life with a question I asked myself as a kid: if irresponsible businessmen could have such a big negative impact on the world—like we saw during the financial crisis—couldn’t responsible entrepreneurs have an equally big positive impact?
That question led me to social entrepreneurship in college, where I convinced 75 people to join me in building projects through Enactus. But I learned something crucial during a trip to Senegal: the most sustainable positive impact often came from regular companies that were well-led, not just those labeled “social enterprises.” A local factory producing cattle feed from fish waste created jobs that lifted farmers out of poverty—that’s when I realized purpose-driven leadership could happen in any business.
The second track was banking and AI. After working as Chief of Staff at Avaloq, one of the world’s leading core banking system providers managing over $4 trillion in assets, I went to Stanford for my MBA and dove deep into machine learning and data science. That’s where I met my co-founder Isaiah Williams, who had been building conversational AI systems that were automating millions of real estate interactions. We became instant friends, winning AI hackathons together and using machine learning to predict loan defaults.
What became clear to both of us was this gap in the market: there were brilliant people in AI who didn’t understand banking, and experienced bankers who didn’t understand AI. We had the dual knowledge. And honestly, we decided to start Casca together at Isaiah’s wedding in Mississippi—because when you find a co-founder you trust and a problem worth solving, you don’t wait.
The problem we chose was small business lending. It’s the most underserved and broken system in banking, and we knew we could fix it.
What inspired you to tackle the challenges of small business lending, and how has your vision for Casca evolved since the company’s founding?
The inspiration came from seeing both sides of a fundamental disconnect. On one hand, you have over 30 million businesses in the U.S.—from small family shops to mid-market enterprises—struggling to access affordable capital. They’re stuck choosing between traditional banks with 60-90 day approval processes or online lenders charging them triple-digit APRs, sometimes up to 19%.
On the other hand, I saw bankers who genuinely wanted to serve these businesses but were held back by their systems. Working with our first customer, Bankwell Bank, we watched loan officers spend countless hours on emails, phone calls, and document requests just to guide one borrower through the process. The human bottleneck was killing everyone’s ability to operate efficiently.do
I saw people in AI that didn’t know banking, and people in banking that didn’t know AI. With our dual knowledge, we wanted to focus on one part of banking to actually fix—and small business lending was the obvious choice because it’s the most broken.
Since founding Casca, our vision has actually expanded beyond where we started. We began focused specifically on SBA 7(a) loans and small business lending, but we quickly realized the platform we were building could transform commercial real estate, personal loans, asset-backed lending—really the entire lending ecosystem. The core insight remains the same: 90% of manual work in lending can be automated. We’re just getting started.
Small businesses drive nearly 40% of U.S. GDP and two-thirds of new jobs, yet many wait weeks for funding. Why hasn’t lending infrastructure kept pace with this growth?
The infrastructure simply wasn’t built for the volume, complexity, and speed required today. Most banking technology in use is 15 to 30 years old—built before APIs, automation, and modern user experience were even industry expectations. These legacy loan origination systems still rely on dozens of manual steps and outdated workflows.
Here’s the fundamental problem: you can’t bolt AI onto a 15-year-old architecture and expect transformation. Many legacy providers tried this with Robotic Process Automation in the past, and those experiments largely failed. Some even price per user seat, which creates a perverse disincentive to automate workflows that would reduce manual touches.
We built Casca as an AI-native platform specifically to eliminate these bottlenecks, not wrap automation around them. The difference is night and day. When you design a system from the ground up for automation—natively orchestrating document collection, applicant communications, and underwriting workflows in one place—you can actually deliver the speed and transparency that small businesses deserve.
Think about it: our customers are processing loans 30 times faster than industry averages. That’s not incremental improvement on legacy systems—that’s rebuilding the foundation entirely.
Community and regional banks have capital but are losing ground in small-business lending. What structural challenges are slowing them down, and how can technology help?
Community banks aren’t losing because they lack capital or appetite—they’re losing because their systems force a 60- to 90-day process that borrowers simply can’t afford to wait for. Meanwhile, alternative lenders approve loans in hours but often charge triple-digit APRs that hurt businesses in the long run.
Technology is the bridge that lets community banks compete on speed without sacrificing the responsible underwriting and affordable rates they’re known for.
Here’s what we’re seeing with Casca: Bankwell Bank, our first customer, is now processing hundreds of small business and commercial loan applications each month through our platform. More than 60% of those applications are initiated outside of working hours—on weekends, late at night—because borrowers can start when it’s convenient for them, not when the bank is open.
The platform automates document collection, status tracking, and borrower communications, which means bankers receive a fully assembled loan package ready for review. That’s 90% less manual effort per applicant. And because the experience is so much smoother, Bankwell is seeing nearly 3x higher conversion rates from initial lead to completed application.
When local financial institutions can offer competitive rates AND fast processing, they become the lender of choice. Capital stays in the community where it belongs. That’s what technology enables—and it’s exactly what community banks need to win back market share.
You built Casca’s AI-native loan origination platform. How are you using responsible, explainable AI to improve the speed, efficiency, and fairness of small business lending?
In banking, compliance has to be the #1 priority for AI—it only takes one noncompliant or incorrect message for a bank to turn off an AI system that’s communicating with customers. AI can’t be a black box in this industry; its actions have to be explainable.
Our goal is to be “the most trusted partner in Banking AI,” which means the integrity of our system always comes first.
We implement this through two core principles:
First, human-in-the-loop monitoring around the clock. Every message sent out to an applicant is checked to be compliant, accurate, and friendly. We’re not just automating and hoping for the best—there’s oversight on every customer-facing action.
Second, explainable AI with a complete audit trail for every action. Every decision the system makes, and the data used to generate it, is recorded and archived for auditability. If a regulator asks why a particular communication was sent or how a document requirement was determined, we can show them exactly what happened and why.
This approach to responsible AI is what allows us to automate up to 90% of the manual work while actually improving fairness. By standardizing workflows and removing human inconsistency, we ensure that every applicant gets the same clear requirements, the same transparent process, and the same opportunity to complete their application—whether they’re applying at 2 PM on a Tuesday or 11 PM on a Saturday.
How does AI help banks balance automation with regulatory compliance, ensuring fair lending without sacrificing speed or scalability?
This is where being AI-native rather than bolting automation onto legacy systems makes all the difference.
In a legacy system, compliance is often an afterthought—you build the workflow, then try to add compliance checks on top. With Casca, compliance is embedded in the platform architecture from day one.
Our approach has three layers:
First, standardization creates fairness. By automating the application process and document requirements, every borrower gets the same clear, consistent experience. There’s no risk of one loan officer asking for different documents than another, or providing different information. This standardization actually improves fair lending outcomes.
Second, the audit trail provides accountability. Every action the system takes is logged with the data that informed it. If a regulator questions any decision, banks can demonstrate exactly what happened, when, and why. This auditability gives banks confidence to move faster because they know they can defend their processes.
Third, human oversight stays in the loop. We automate the tedious work—document collection, applicant reminders, status updates—but critical decisions still involve human judgment. Bankers review the decision-ready packages our system assembles. They’re not removed from the process; they’re elevated to focus on actual underwriting rather than administrative tasks.
The result is scalability without sacrificing compliance. With our platform, every application follows the same compliant workflow, whether it’s the first one or the thousandth.
What trends or shifts do you see in small business lending over the next 3-5 years, and how will AI shape this future?
The next 3-5 years will be defined by the death of the legacy loan origination system and the rise of automation-native platforms.
Here’s what I mean: the economic pressure on banks is only increasing. Small businesses need faster access to capital to compete, and they have more options than ever—including alternative lenders who can approve in hours. Community and regional banks have to match that speed while maintaining responsible underwriting and competitive rates. The only way to do that is through automation.
We’re going to see a few key shifts:
First, borrower expectations will fundamentally change. Once businesses experience applying for a loan at 10 PM on a Sunday and getting a response in days instead of months, there’s no going back. The banks that can’t offer that experience will lose market share.
Second, the definition of “competitive advantage” will shift from balance sheet size to operational efficiency. The winners won’t necessarily be the biggest banks—they’ll be the ones who can process the most applications with the least friction. Technology becomes the equalizer.
Third, lending will expand beyond traditional categories. As platforms like ours automate 90% of the manual work in small business lending, we’ll expand into commercial real estate, personal loans, asset-backed lending—anywhere there’s a fragmented, outdated process that technology can fix.
AI will be the engine driving all of this. But the critical distinction is between AI that’s bolted onto legacy systems versus AI that’s native to the platform. The future belongs to platforms built from the ground up for automation.
With all the buzz around generative AI, which applications do you see as truly practical for banks today, and which are still just hype?
This is a great question because there’s definitely a lot of hype in the market right now.
What’s practical today:
Document processing and data extraction is incredibly practical. Banks deal with mountains of tax returns, financial statements, personal guarantees—all unstructured data that used to require manual review. AI can now extract, validate, and organize this information with accuracy that matches or exceeds human performance.
Intelligent workflow automation is another huge one. Rather than just following pre-programmed rules, AI can actually understand where an application is in the process, what’s missing, and what the next best action should be—whether that’s sending a borrower a reminder, requesting additional documentation, or flagging something for human review.
Borrower communication and guidance is transforming the experience. AI can answer common questions, provide status updates, and guide applicants through complex requirements in natural language—making the process feel more like talking to a helpful advisor than filling out government forms.
What’s still hype:
Fully autonomous underwriting without human oversight is, in my view, not ready for prime time—and may never be appropriate in banking. The regulatory environment requires explainability and accountability. More importantly, relationship banking is built on judgment and trust, which AI should augment, not replace.
Generic chatbots that claim to “solve banking” are mostly hype. We’ve seen too many banks deploy customer-facing AI that gives incorrect information or fails basic compliance checks. In banking, you can’t iterate your way to compliance—you have to get it right from day one.
The key distinction is between AI that’s deeply integrated into banking workflows with proper oversight versus AI that’s meant to look impressive in a demo. The practical applications are the ones where AI removes friction from existing processes while maintaining—or improving—control and compliance.
A quote or advice from the author:
“You can’t bolt AI onto 15-year-old systems and expect transformation. Rebuild from the ground up—and make compliance part of the foundation, not an afterthought. In banking, speed without trust isn’t innovation, it’s risk.”
Stay Ahead of the Financial Curve with Our Latest Fintech News Updates!
