The emergence of developing smart net-zero cities in recent years has attracted significant attention and interest from worldwide communities and scholars as a potential solution to the critical requirement for urban sustainability. This research-in-progress paper aims to investigate the development of smart net-zero cities to propose a digital transformation roadmap for smart net-zero cities with a primary focus on data. Employing systems thinking as an underpinning theory, the study advocates for the necessity of utilising a holistic strategy for understanding the complex interdependencies and interrelationships that characterise urban systems. This paper presents a novel integration method which utilises a pre-trained large language model modified with Retrieval-Augmented Generation and supplemented with quantitative analysis to propose and validate systemic interventions for smart net-zero cities. The Large Language Model initially will be pre-trained on relevant domain-specific knowledge to generate the city system behaviour. Through an iterative process that includes city system simulation (knowledge graph) and focus group validation, the insights produced by the model will be improved. The extracted knowledge from the model will be utilised to develop a digital transformation roadmap for a smart net-zero city. Through this integrated approach, it is expected to achieve systemic intervention followed by a systemic digital transformation roadmap for smart net-zero contributing to a more holistic understanding of urban sustainability.