The development of Fully Autonomous Vehicles would fundamentally change the nature of user interactions, behaviors, needs, and activities within vehicles. Passengers who are free from driving would expect to undertake diverse in-vehicle Non-Driving-Related Tasks (NDRTs) to keep themselves occupied and free from boredom. Conversational Artificial Intelligence with its ability to have free-flowing conversations could offer a solution to improve in-vehicle user experience. To explore design innovation opportunities that Conversational AI may offer within Fully Autonomous Vehicles, this work first researches what roles and relationships can Conversational AI play towards users of fully Autonomous Vehicles. These are investigated by conducting qualitative interviews of end-users and subject matter experts, and analyzing these interviews using the Thematic Analysis method. Second, this work examines how end-users qualitatively assess the embodied user experience provided by Conversational AI roles and relationships designed for Fully Autonomous Vehicles through the means of user-technology interaction experiments. These are conducted by employing a combination of participant role enactment method and a Wizard of Oz experimental setup with a mock Conversational AI operated within a prototyped Autonomous Vehicle under laboratory conditions. Ensuing these interaction experiments, the qualitative assessment of these in-vehicle Conversational AI roles and relationships was done by conducting guided brainstorming sessions using the parameters outlined in the Subjective Assessment of Speech System Interfaces (SASSI) questionnaire along with two free-from questions. Results from both phases are presented in the form of user-desired in-vehicle Conversational AI roles, aspects that would enable the forming of Human-Autonomous Vehicle relationships mediated by Conversational AI, and user experience evaluation of the embodied idea. Based on these results, design guidelines are articulated that are aimed to inform design professionals as well as internal design strategies of Autonomous Mobility and Artificial Intelligence companies looking to innovate. Finally, we reflect on our methodological approach to indicate directions for future studies on similar lines and discuss our research’s implications for professional practice.