Will AI Replace Human Underwriters? Banxware’s Approach to the Future of Lending

How AI is Changing Underwriting at Banxware

Our Risk Analytics Team has fundamentally reshaped our approach to credit decisions. Gone are the days of manual, subjective assessments. Today, we use machine learning and AI to analyze large datasets and pinpoint risks and opportunities with far greater precision. This transformation enables faster, fairer, and more scalable decisions.

A key advantage: While traditional lenders often misjudge SMEs using surface-level indicators, our AI-driven models dig deeper. By considering not only financial statements but also market trends and unique business signals, we ensure that promising, dynamic companies, so crucial for economic development, don’t slip through the cracks.

“We’re not automating for automation’s sake,” says Jose Vollmann. “Our goal is to make the underwriting process smarter, faster, and more accessible without losing the expert knowledge that makes Banxware unique.”

AI in Action at Banxware

• Automated Data Extraction:
  Every day, customers send us diverse financial documents: Excel exports of BWAs, balance sheets, and bank statements. What once demanded hours of manual review is now handled by our machine learning models. They automatically extract and organize all relevant figures, helping us process applications with impressive speed and accuracy.
• LLMs for Deeper Insights:
  We’re piloting Large Language Models (LLMs) to take our analysis even further. These AI models scan our processesd data to spot anomalies or unusual activities which could be overseen by our teams. While LLMs are already enhancing our risk detection, their ability to fully grasp business context is still growing.
• Continuous Model Improvement:
  Our scoring models never stand still. With every new case and every human decision, we feed real-world feedback into our system, constantly training and updating our algorithms.

Will AI Replace Humans?

AI has brought huge advances to credit underwriting, but at Banxware we see it as an enabler, not a replacement, for human expertise.

“AI is incredibly powerful for the repetitive and analytical parts of underwriting. But when it comes to understanding context, handling exceptions, or making a judgment call for a unique business, that’s where human expertise still shines,” explains Jose Vollmann.

Consider this real-world example:

"Urban Bee Solutions GmbH," an urban beekeeping and rooftop honey production company, applies for Sofortfinanzierung to expand its hive network for the spring pollination season. The business uploads its financials. Our AI efficiently extracts all relevant numbers and immediately flags a large, one-time equipment purchase and an unusually high inflow from an international food distributor, both outside the business’s typical monthly patterns.

Given our fast payout commitment, the process advances quickly. Before approval, a Banxware underwriter reviews these flagged items. The underwriter recognizes that the large equipment purchase is for new urban hive installations required by a new municipal partnership, and that the international payment is a bulk pre-order for limited-edition “city honey”, a rare but regular occurrence each spring.

Where a purely automated system might classify these irregularities as high risk, the underwriter’s understanding of the industry and seasonal business model ensures that the loan is confidently approved and paid out the next day.

But what happens next is just as important: Each time our underwriters clarify such cases, whether it’s identifying a seasonal investment, spotting a legitimate pre-payment, or noting a new pattern, they document their decision and reasoning. This feedback flows straight back into our AI models, teaching them to better distinguish between typical and atypical cases in the future.

“Every time we make a decision, whether by a model or a person, we feed that information back into our systems,” Jose emphasizes. “So our technology and our team are learning together."

This ongoing cycle ensures our AI grows smarter and more precise, while Banxware’s unique underwriting expertise is captured and multiplied with every new application.

Looking Ahead: The Future of Underwriting

With tools and frameworks like LLMs and Agents evolving at high speed, could some underwriting be fully automated soon? For standard cases and well-known industries, the answer may be yes. In the near future, AI could independently handle many straightforward applications, especially where we have extensive historical data and have trained our models on thousands of previous human decisions.

However, in the diverse and complex landscape of SME lending, there will continue to be situations where human expertise is essential, whether it’s understanding a unique business model, evaluating incomplete data, or weighing the impact of unusual market events.

Right now, we firmly believe in the value of human–AI collaboration. But who knows what the next five years will bring? Perhaps, as our technology and data grow, we’ll see our own AI underwriter autonomously making credit decisions in well-understood segments.