Digital Human as Travel Agent: Designing and Evaluating an Avatar-Based Travel Assistant

Finding inspiration for a next trip has long meant a choice: browsing the internet - overwhelming and fragmented - or consulting a human travel agent - personal, but costly and hard to scale. LLM-based Digital Travel Assistants (DTAs) were a first answer to this problem. Now these assistants are taking the next step, moving from text-based chat toward avatars: Digital Humans with a face, a voice, and gestures that travellers can actually talk to.

This thesis is part of a research project on avatar-based DTAs, developed in collaboration with practice partner Swiss International Air Lines (SWISS); a University of Zurich GRC Short Grant has been requested to support the project, with the funding decision still pending. You will help build a forward-looking research prototype of an avatar-based travel assistant focused on inspiring travellers for their next trip, and evaluate it against a conventional chatbot baseline with around 16 participants through hands-on prototype sessions, interviews, and optional survey elements. The thesis is planned to start in Q4 2026.

While the exact scope of the prototype will be defined together at the start of the thesis, the project will include:

• Reviewing literature on Digital Travel Assistants, embodiment, and travel inspiration.
• Designing the avatar interaction concept, storyboard, and prototype logic (touchscreen, or VR glasses?).
• Owning the technical implementation end-to-end: setting up and hosting the required infrastructure and developing the avatar-based prototype itself, using suitable tools (e.g. Unreal Engine, GenAI dialogue tools, voice and avatar generation tools).
• Recruiting participants and conducting the empirical study (interviews and optional survey).
• Analyzing the qualitative data and deriving design guidelines for avatar-based DTAs - insights of direct interest to SWISS and the wider travel industry.

We are looking for a student who enjoys working independently and is comfortable taking ownership of a real research project over roughly six months, including its full technical setup. Interest in human-computer interaction, conversational and embodied AI, and qualitative research methods is expected; experience with prototyping tools (e.g. Figma, Unreal Engine, GenAI/LLM APIs) or avatar and voice generation tools is a plus. Good written and spoken English is required; German is an advantage for conducting participant interviews.

For more information please contact Jan Friedli