From humans to machines: Researching entrepreneurial AI agents built on large language models

Entrepreneurial AI agents (e.g., Large Language Models (LLMs) prompted to assume an entrepreneurial persona) represent a new research frontier in entrepreneurship. We explore whether such agents exhibit the structured profile of the human entrepreneurial mindset, a construct long established as a driver of entrepreneurial activity in humans. Drawing on the biological concept of host-shift evolution, we investigate whether the characteristic components of this mindset (i.e., basic personality traits, characteristic adaptations, and self-concept) emerge in a coherent constellation within AI agents. Using established psychological instruments, we examine whether entrepreneurial AI agents display systematic patterns that mirror the internal structure of the entrepreneurial mindset. Our findings indicate that such coherent profiles do emerge, consistent with a human-like entrepreneurial mindset structure. At the same time, our results highlight important limitations, including stereotype amplification and the likelihood that responses reflect statistical simulation or mimicry rather than genuine cognition. We argue that these findings signal a potential shift in entrepreneurship research: from focusing solely on the psychology of human entrepreneurs to also examining the simulated mindset structures of artificial entrepreneurial agents. This opens new research agendas on mindset–behavior consistency, human–AI collaboration, and the frameworks needed to study the emerging psychology of entrepreneurial AI. • AI agents built on LLMs can simulate core features of the entrepreneurial mindset. • Entrepreneurial mindset traits appear to spill over from humans to AI systems. • Role-prompted LLMs reproduce structured, human-like entrepreneurial profiles. • Entrepreneurial AI agents act as creative but non-neutral collaborators. • Findings highlight opportunities and risks in human–AI entrepreneurial practice.