Entrepreneurs traditionally use 'learning-by-doing' and 'learning-by-thinking' as alternative approaches to iteratively build business models for their new ventures. However, both approaches face criticism in how they address novelty and uncertainty, which are crucial to successful entrepreneurship. While Generative AI (GenAI) is increasingly used in entrepreneurial tasks, the practices through which it becomes a learning resource for entrepreneurs remain unexplored. Based on a qualitative study, we present a process model that illustrates how entrepreneurs incorporate GenAI into business model design through five resourcing practices. These practices transform GenAI into a valuable resource for facilitating learning during the design process. This approach, which we term 'Learning-by-Conversing,' introduces a generative startup methodology to complement the lean startup model. We distinguish two modes of learning by conversing - reflexive learning and confirmatory learning - based on how novice and experienced entrepreneurs engage with it. By proposing a learning approach that integrates GenAI with entrepreneurial efforts, we bridge the 'thinking' versus 'doing' debate in business model generation and deepen our understanding of GenAI's role in entrepreneurship