Rule-based and AI-assisted workflows for DBP: Insights from applied research

Abstract

Digital building permit procedures still rely heavily on fragmented, document-based workflows, requiring manual checks for completeness, consistency and regulatory compliance. This paper presents an applied research project investigating how rule-based and AI-assisted workflows can improve the quality and efficiency of building permit applications, with a particular focus on the pre-submission phase. The project, conducted by the Institute of Virtual Design and Construction at FHNW with squaretech GmbH and the City of Winterthur, develops a “Step 0” framework combining an automated compliance checker and an AI-supported chatbot. The system combines deterministic rule-based methods with selected AI components and geospatial information, including GIS, where these provide clear added value. The paper presents the conceptual framework, system architecture, pilot validation and initial implementation insights. It further discusses key challenges related to data quality, legal requirements, explainability, transferability and integration into existing administrative workflows.