Research
We began by designing a physical scent album of 24 base aroma families. Through a custom-built mobile companion app, we collected user preferences and emotional associations with each scent. This data was used to:
Build individual olfactory profiles
Train a base AI model to generate custom scent cycles
Avoid repeated exposure through adaptive sequencing
In the early stages of the product, we conducted research, held focus group interviews, and trained the AI model to create personalized scents based on a sample album of 24 scent families.
Hundreds of hours were invested by leading fragrance experts to explore and digitize scents, translating them into an AI-driven fragrance model. In parallel, industrial designers and engineers developed multiple iterations of the smart diffuser to ensure exceptional scent distribution and the stability and reliability of the hardware.
In the early stages of the product, we conducted research, held focus group interviews, and trained the AI model to create personalized scents based on a sample album of 24 scent families.
