Gemini and ChatGPT were both useful in the AI-assisted workflow, but they played different roles. Gemini was more effective during the research stage, especially in gathering contextual information about Malaysia’s tropical climate, traditional Malay architecture, passive design strategies, the Bunga Raya, and the Rukun Negara principles. Its responses were more factual and research-oriented, providing background knowledge that helped build the foundation of the project.

ChatGPT was more effective during the concept development stage. It helped transform the research information into architectural ideas, such as the central Unity Court, five surrounding principle-based spaces, inverted hibiscus roof form, porous timber screens, and shaded walkways. While Gemini provided the raw research and context, ChatGPT helped organize the information into a clear design concept, spatial distribution, and presentation narrative.

The main difference is that Gemini supported understanding, while ChatGPT supported interpretation and design translation. Gemini helped answer “what is the context?”, while ChatGPT helped answer “how can this become architecture?” Both tools were useful, but human judgment was still needed to select relevant information, reject unclear ideas, refine the concept, and make the final design decisions.