This paper works on an approach for designing chemical warehouse layouts using Large Language Models (LLMs) like GPT-4. The main objective of the study is to generate layouts that minimize safety risks and optimize accessibility. Our methodology talks about framing the design process as a Constraint Satisfaction Problem and specifying constraints within GPT-4 through contextual guidance and prompt engineering. The constraints chosen consider chemical compatibility, proximity of frequently accessed chemicals, and space utilization. We looked into some key aspects in getting expected results from Generative AI tools such as specificity of prompts, providing examples, diverse prompt variations and monitoring for any response biases. Using the outputs we obtained from iterative prompting; we generated a dataset of 400 sample layouts. We finally used a constraint satisfaction evaluator specifically designed to output a custom evaluation metric score for the layouts based on the constraints to analyze the effectiveness of this method. Overall, the generated layouts conformed well to the constraints, demonstrating a fairly high satisfaction rate. The findings thus showed what potential LLMs have when it comes to automating this design process, and presents a novel representation of this problem unique to chemical warehouses. It, however, also identified limitations of the model in compliance with increasing number of rules and highlighted the necessity of a larger database and supervised fine-tuning.