In a recent profile in Forbes Belgium, we share how we’re tackling one of the biggest bottlenecks in European mortgage lending: the slow, largely manual credit decision. While the front end of the mortgage journey is now highly digital – simulators, online onboarding, document upload portals – the real “time-to-yes” moment often still relies on people opening PDFs, visually comparing data, retyping information and sending incomplete files back and forth. From our base in Antwerp and across six European markets, we’ve built Oper to address exactly that gap in the process.
The article highlights how our software focuses on the “decision core” of mortgage origination: underwriting, document processing and policy checks. Our AI agent, Herman, acts like a junior credit analyst – recognising documents, checking completeness, structuring data and surfacing signals for a senior underwriter, without ever taking the final decision. We work strictly within each bank’s own data sources and credit policies, and our algorithms are designed to reason, not guess: when Herman encounters something he doesn’t understand, he stops and hands it over to a human. Drawing on insights from our “Mortgage Trends 2026: A New Era of Intelligent Mortgage Origination” report, our CEO and co-founder Geert Van Kerckhoven explains how rising regulatory requirements, more complex files (often 20–50 documents per case) and a shortage of credit analysts have created a structural “time-to-yes crisis” just as banks want to grow their mortgage books without proportionally increasing headcount.
Looking ahead, we share an ambition that’s simple to describe but challenging to achieve: moving the industry towards near real-time, safe and compliant mortgage decisions for the vast majority of cases, fundamentally improving the customer experience while making underwriting more scalable. The profile also touches on broader questions around pricing, segmentation and profitability in mortgage lending, underlining the need for better data-driven decisioning.
.jpg)
