The rapid evolution of Artificial Intelligence has established it as an essential component across professional disciplines and fundamentally altered traditional workflows. In the field of architecture, AI tools are increasingly utilized for visualization and ideation, yet their application in technical building physics and efficient energy design remains underexplored, particularly regarding reliability and accuracy. This thesis investigates the integration of conversational AI agents or Large Language Models (LLMs), such as ChatGPT and Google Gemini, into the early design workflows of young architects and students, specifically focusing on energy-based decision-making. The research methodology utilizes a primary school case study in Southern Senegal, characterized by strict financial constraints and an absolute reliance on passive climatic strategies to evaluate an integrated, three-phase design pipeline.