Design Principles for AI-Enhanced Process Automation: An eDSR Approach to Intelligent Data Validation in Financial Decisions

Over the past decade, automation of business processes has been evolving onto more advanced and intelligent solutions that go beyond basic capabilities of rule-based robotization. This paper presents a Design Science Research (DSR) study conducted to develop and evaluate design principles for integrating fuzzy logic algorithms with Robotic Process Automation (RPA) to address complex data validation challenges in financial services. Following the echelon-based Design Science Research (eDSR) methodology, we decompose the design process into three interconnected echelons: problem diagnosis, solution design, as well as demonstration and evaluation. Our research addresses the gap in prescriptive knowledge for designing intelligent automation systems that can handle imperfect data matching scenarios while maintaining accuracy and compliance standards. The iterative research process conducted at a multinational IT corporation, enabled us to develop design principles guiding the integration of Jaro-Winkler, Levenshtein, and N-gram algorithms within RPA, resulting in a solution that achieved a 67% reduction in false rejection rates while maintaining 97% accuracy in data validation processes. Our main contributions include: (1) prescriptive design knowledge for enhancing RPA with fuzzy logic capabilities, (2) validated design principles for balancing automation flexibility with risk management in financial contexts, and (3) empirical evidence implementing Task-Technology Fit (TTF) theory at intelligent automation contexts. It provides actionable guidance for practitioners implementing AI-enhanced automation in complex, regulated environments.