This study investigates how generative artificial intelligence (AI) models—specifically ChatGPT Plus (with and without the Projects feature) and DeepSeek—can support the design of LEGO® SERIOUS PLAY® session plans. Through a systematic reflexive content analysis methodology, we evaluate the originality, structure, and methodological fidelity of AI-generated LSP session plans across two distinct thematic areas: Creative Leadership and Building Team Trust. Our multi-layered evaluation framework includes content analysis, session structure mapping, and a temporal triangulation protocol that compensates for the limitations of single-researcher evaluation. Findings reveal that ChatGPT Plus with Projects exhibits superior consistency and alignment with LSP principles (scoring 13.7/15 on our composite index), while DeepSeek demonstrates strengths in information synthesis but limited session structuring (8.9/15). Comparative analysis of AI-generated facilitator prompts reveals significant qualitative differences, with ChatGPT Plus with Projects producing questions with greater metaphorical depth and nuance, creating space for participants to explore complex dimensions of the themes. The study contributes to the fields of creativity, participatory facilitation, and AI by providing an evidence-based framework for assessing AI support in creative methodologies and offers practical guidelines for facilitators seeking to leverage AI as a collaborative partner in session design. Our prompt engineering protocol, detailed in this paper, provides a replicable template for facilitators and researchers exploring AI-human collaboration in experiential learning design.