The challenge
How might a conversational agent collect dietary context and respond through a more expressive, human-like interface?
My role
In a team of two, I co-developed the Furhat and ChatGPT integration, structured conversation flows and worked on the robot's gaze, facial expressions and other social cues.
The approach
- Designed dialogue to collect health, lifestyle and food preference information.
- Implemented logic for a prototype that generates personalised meal plan suggestions.
- Iterated the timing and expression of non-verbal robot behaviour.
What emerged
The result was an interaction prototype exploring how conversational logic and social presence work together. It was not presented as a clinically validated dietetic service.
