What happens when code is generated faster than a business can agree on the process behind it? At MoIn Group, we use AI to build ServusOS, an operating system for insurance brokers that itself uses AI to interpret documents and automate everyday work. Bringing it into existing businesses exposed a difficult reality: different people follow different processes, documents contradict each other, and a fix for one case can quietly become the wrong rule for the next.
In this candid case study, Fabian Rothe shares concrete examples of those conflicts and the approaches the team is developing to address them. He explores how to identify who owns a business decision, establish which sources take precedence, and use AI to turn recurring exceptions into explicit, testable rules – with human judgment at the points where the business must choose.
Product, engineering and design practitioners will leave with practical ways to distinguish model errors from unresolved business decisions, prevent local fixes from becoming conflicting logic, and move from greenfield prototypes to AI-powered workflows in established organizations.
Chief Product Officer at MoIn Group. Ex-DeepL.
Fabian Rothe is Chief Product Officer at MoIn Group, where he leads the development of ServusOS, an AI-powered operating system for insurance brokers. Previously, he worked at DeepL, where he automated processes across the self-serve business, and at Trusted Shops, where he helped build the legal tech business. Today, he works hands-on with AI coding agents and domain experts to turn complex business processes into software with explicit rules, clear decision ownership and reliable workflows.
