The integration of smart buildings and smart cities has become a key element of modern urban planning, necessitating a thorough assessment of various service factors to ensure resource sharing and improve the overall impact of integration on smart city performance. This study investigates the use of advanced AI models, specifically ChatGPT by Open AI and Bard developed by Google as artificial intelligence experts to evaluate 26 factors related to smart building services. This approach aims to provide a comprehensive guidance methodology, which sheds light on the peculiarities of using AI large language models (LLMs) to assess smart building integration in the broader context of the efficiency of smart cities. The study investigates the session-to-session consistency of responses generated by both LLMs, assesses their stability, identifies performance fluctuations, and reveals efficient strategies of prompt engineering in the smart building and smart city domains. By comparing the strengths and weaknesses of each model, the authors encompass critical aspects such as model biases and knowledge sources upon which the models rely. The research results contribute to understanding the capabilities and limitations of utilising ChatGPT and Google Bard as AI experts as well as identifying areas for future potential applications.