Autonomous vehicles (AVs) could improve public transport (PT) and reduce private car use, improving transport-related issues, such as emissions and accidents. However, widespread public adoption and acceptance of AVs is crucial to achieve this potential. This requires replacing traditional staff support in conventional PT vehicles. One approach is to employ virtual in-vehicle assistants (VIVA) tailored to the varied needs of PT passengers. This research involved a focus group workshop with experts from the field and a systematic literature review to identify the requirements for VIVA. Through iterative cycles, including five think-aloud sessions with experts, a set of design principles was established. As one of the first studies in that regard, we provide comprehensive design knowledge.