Through AI and Crowdsourcing, Symrise Creates the Scent of Berlin


How do you capture the scent of a city? Symrise took its Philyra artificially intelligence (AI) system to Berlin to find out just that.

Over the course of six weeks (ending June 29, 2019), the company ran an interactive and collaborative project, during which interested Berliners visiting the Bikini shopping mall could help Philyra develop a new fragrance. Participants rated five different base fragrances based on their personal preferences; together with senior perfumer Marc von Ende, the AI system created the “core fragrances” based on certain phrases designed to represent the spirit of different districts of Berlin: Grunewald, Kreuzberg, Prenzlauer Berg, Schöneberg and Wannsee.

Related: [video] Symrise, IBM Research Bring AI to Fine Fragrance

The computer learned from each participant’s evaluation, and at the end of each week it created an intermediate perfume based on the tablet assessments with the help of a Symrise perfumer. Mall visitors could then sniff the intermediate perfume and share what emotions it evoked. The AI continued to learn based on these responses and answers to other open questions posed to testers.

At the end of June 2019, the final computer-generated scent was unveiled: #Berlin 3.0. Participants could receive an email with a description of the scent and then test it for themselves at the Philyra booth.

“We were curious to see where this experience of collaborative fragrance design powered by Artificial Intelligence would take us to”, says Claire Viola, vice president, digital strategy, fragrance. “And we are excited about the outcome, which enriched Berlin a unique brand-new scent.”

More in Fragrance