Urban Digital Twins (DTs) are gaining prominence for their transformative potential with the advancement in artificial intelligence (AI). However, literature on applied examples of transport DTs, especially for buses, remains limited. This study conducts a literature review and thirty-eight stakeholder interviews to map the landscape for AI-enabled, transport-focused DT development in Liverpool City Region in the UK and Busan in South Korea. It identifies key actors, their roles and responsibilities, the relationship among them as well as their motivations, priorities, and standards pursued regarding topics including big data, AI, modelling frameworks, governance, feasibility, and social equity. Based on the synthesis of cross-case findings, this paper proposes a transferable framework for guiding DT development for buses, multimodal transport and beyond, with examples of input data, DT modules, conditions needed and KPIs. Furthermore, it identifies six important considerations for DT implementation and presents practical case examples from Liverpool and Busan. First two points are on governance: fostering multi-stakeholder collaboration and funding; and pursuing DT initiatives at both national and local levels. Two other points are on data: standardising online data-sharing platform with privacy safeguards and data monetisation; and overcoming reliance on private data through data agreements and citizen participation. Final two points are on DT design: adopting modular DT solutions; and matching DT modules with appropriate level of AI-enabled dynamic simulation, visualisation and automation.