FounderPulse-AI: Design and Pilot Evaluation of Bounded Intelligence for Startup Founders Under Entrepreneurial Strain

Abstract

Intelligent decision support systems increasingly assist users under uncertainty, but limited research explains how they can support startup founders under entrepreneurial strain without crossing into coaching, clinical, or advisory authority. This study designed and pilot-evaluated FounderPulse-AI, a retrieval-grounded, prompt-governed, nonclinical decision support artifact for Greek startup founders. Using design science research, the study combined a qualitative needs analysis with 20 founders, co-founders, and ecosystem actors and a four-week comparison involving 100 founders and co-founders assigned to FounderPulse-AI or a structured reflection journal. FounderPulse-AI produced stronger gains in reflective clarity, perceived decision quality, perceived usefulness, trust in AI, usability, boundary clarity, escalation appropriateness, and intention to continue use, while privacy concerns decreased more in the journal condition. The main contribution is a bounded AI founder support architecture and a Responsible AI Founder Support Model that connects intelligent interaction, decision recovery, privacy sensitivity, and human escalation.