The rapid advancement of generative artificial intelligence (AI) has introduced powerful tools for reimagining how knowledge is produced, formalized, and validated. Large language models (LLMs), in particular, enable the articulation of complex ideas through natural language and the generation of outputs that resemble models, algorithms, or theoretical frameworks (Bender et al., 2021; Floridi & Chiriatti, 2020). While this capability has sparked interest in prompt engineering, existing approaches remain largely pragmatic, focusing on optimization heuristics rather than exploring the epistemological and methodological implications of prompts as instruments for theory-building (White et al., 2023). To address this gap, we introduce Prompt-Driven Conceptual Modeling (PDCM) as a systematic methodology that uses natural language prompts, mediated through generative AI, to produce structured conceptual artifacts. We further define Generative Theoretical Prototypes (GTPs) as provisional outputs of PDCM that serve as scaffolds for subsequent refinement and validation. Drawing from traditions of conceptual modeling (Booch, 1994; Object Management Group [OMG], n.d.), computational creativity (Colton & Wiggins, 2012; Boden, 2004), and design science research (Hevner et al., 2004), this paper argues that PDCM provides a reproducible framework for integrating human intuition, AI mediation, and scholarly rigor. Using case studies—spanning platform ecosystems, distributed computing architectures, and mathematical modeling—we illustrate how PDCM functions in practice. We conclude by positioning PDCM and GTP as contributions to both professional and academic education communities, identifying future directions for validation, benchmarking, and cross-disciplinary adoption.