Before we dive in, a personal note: my fascination with this industry comes from an unexpected source—my younger brother Jeremy Fragrance, who has become the world's leading fragrance influencer. While his energetic videos and signature white suits might raise eyebrows, his impact on how people discover and appreciate scents is undeniable. His journey has given me a unique window into this traditionally closed industry that's now being revolutionized by AI.
My interest in AI's role in fragrance creation was initially sparked back in 2019, when I had the opportunity to speak with Achim Daub, then President of Symrise. During our conversation, he shared insights about how they were beginning to implement AI in their fragrance development processes. That early glimpse into the intersection of artificial intelligence and perfumery has had me following the industry's digital transformation ever since.
Perfumery's Persistent Challenges
For centuries, creating a new fragrance has been an artisanal process shrouded in mystery and tradition. The typical development cycle could stretch anywhere from 6 to 18 months, with perfumers manually mixing ingredients, waiting for formulations to stabilize, and making hundreds of incremental adjustments before finalizing a scent.
This painstaking approach created several critical business challenges:
Time-to-market pressures in an increasingly trend-driven industry
High costs from extensive trial-and-error development
Limited exploration of possible ingredient combinations
Dependency on a small pool of master perfumers with years of training
Difficulty translating consumer preferences into actual formulations
As one industry executive put it: "Creating perfume was like composing music without being able to hear the notes until the entire symphony was finished."
Multiple Approaches to AI Adoption
Unlike industries where a single AI approach dominates, fragrance houses have developed diverse and complementary systems. Four major players have emerged with distinctive technologies:
Givaudan: The Interactive Visualizer
Givaudan's Carto system reimagined fragrance creation through a digital interface that visualizes scents in an "Odor Value Map." Perfumers use a touchscreen to explore formulations while a connected robot instantly creates physical samples for testing. As Calice Becker, VP of Creation, explained: "Carto invites perfumers to dare and experiment with combinations that might not have been obvious choices."
Symrise & IBM: The Apprentice System
In partnership with IBM Research, Symrise developed Philyra, an AI "apprentice" that analyzes thousands of formulas, raw materials, and historical performance data to suggest novel combinations. This system famously created two commercial fragrances for O Boticário in Brazil—marking the first AI-designed perfumes to reach market.
DSM-Firmenich: The Digital Service Platform
Taking a different approach, Firmenich (now DSM-Firmenich) launched Scentmate, a digital platform that democratizes fragrance creation for small brands and entrepreneurs. As the company described it: "Scentmate compresses what traditionally took weeks into days," opening fragrance development to businesses that couldn't previously access it.
IFF: The Consumer Insight Engine
International Flavors & Fragrances (IFF) introduced ScentChat, an AI-powered mobile app that gathers and analyzes real-time consumer feedback. Unlike systems focused on formulation, ScentChat bridges the gap between consumer language and technical perfumery, using AI to translate subjective descriptions into actionable insights.
How They Made It Work
These companies followed different but complementary pathways to AI implementation:
Building the Data Foundation
All major players invested heavily in digitizing their fragrance knowledge. This meant:
Creating structured databases of thousands of formulas
Developing consistent taxonomies for scent descriptions
Standardizing performance metrics across fragrance families
Capturing consumer response data in machine-readable formats
Symrise's approach with IBM exemplifies this challenge—they compiled hundreds of thousands of formulas and their performance data to train Philyra's neural networks.
Varied Technical Approaches
Each company leveraged different AI technologies:
Givaudan built a visual interface with predictive algorithms
Symrise/IBM used deep learning to analyze formula patterns
Firmenich created a streamlined digital workflow platform
IFF focused on natural language processing to interpret consumer feedback
Strategic Integration with Humans
A common thread across all implementations was positioning AI as an assistant, not a replacement. As Achim Daub of Symrise noted, their AI system serves as an "apprentice" to make perfumers more productive. Similarly, Givaudan deployed Carto globally but kept perfumers central to the creative process.
Cross-Functional Teams
Success required collaboration across disciplines. Firmenich formed a multidisciplinary "startup-style" team with experts in AI, product design, and fragrance to co-create Scentmate. This blending of technical and domain expertise proved crucial for building tools that perfumers would actually use.
Common Hurdles Across the Industry
Despite different approaches, companies faced similar obstacles:
Technical Limitations and Data Challenges
Unlike text or images, scent lacks abundant digital datasets and a clear "language." Creating large, labeled datasets of fragrance formulas with sensory profiles proved costly and labor-intensive. Even state-of-the-art algorithms struggled with the non-linear nature of scent, where small formula changes can produce dramatically different results.
Industry Resistance and Cultural Barriers
The fragrance world's storied heritage of artistry bred resistance to AI. As perfumer Emmanuelle Moeglin cautioned, over-reliance on AI might "dilute the individuality of perfumery and risk losing customer trust in originality." Companies had to carefully manage this cultural transition.
Data Privacy and Intellectual Property Concerns
Formulas represent core intellectual property for fragrance houses. Using AI systems—especially those involving third parties like IBM—required strict governance to protect these trade secrets. One industry expert noted intellectual property is at risk with AI because "the technology allows more people access to data that was once siloed."
Talent Gaps
Implementing AI required specialists who understood both computer science and perfumery—a rare combination. Companies addressed this through partnerships (like Symrise leveraging IBM's researchers) or by building internal capability through training programs for cross-disciplinary expertise.
Industry-Wide Transformation
AI has dramatically transformed fragrance creation in several key areas:
Speed and Efficiency
Companies now develop fragrances in days instead of months. Teams run more tests with fewer resources and use expensive ingredients more efficiently.
Creative Expansion
Perfumers discover ingredient combinations they wouldn't have tried before. New fragrances perform better in consumer testing. Companies like O Boticário have tapped new markets with AI-designed scents for millennials.
Business Innovation
Scentmate now brings fragrance creation to small brands. EveryHuman's AI kiosks create custom perfumes in stores. YSL uses EEG technology to match brain responses to ideal scents.
Sustainability
EcoScent Compass analyzes environmental impacts of formulas. Failed trials produce less waste. AI optimizes formulas to use fewer ingredients while maintaining quality.
"AI transforms how we tell new scent stories while preserving the artistry of our profession," says Maurizio Volpi of Givaudan Fragrance & Beauty.
Cross-Industry Lessons to Apply
For leaders looking to implement AI in creative or technical fields, the fragrance industry's journey offers valuable lessons:
The fragrance industry teaches us valuable lessons about implementing AI successfully. Make humans and AI true partners by keeping domain experts in charge and training AI to enhance their work, not replace them. Treat AI as an apprentice that helps experts work better.
Build your data foundation first by organizing existing knowledge before adding AI. Create clear labels for your domain expertise and connect different data sources so AI gets the full picture.
Mix technical and domain teams by forming groups with both AI experts and industry veterans. Create bridge roles for people who speak both languages and partner with outside specialists when you need to learn fast.
Start small and focus on ROI by beginning with simple projects that show quick results. Pick AI initiatives that directly support business goals and plan a path from basic to advanced applications.
Manage the cultural shift by remembering that changing mindsets matters as much as technology. Find enthusiastic early adopters to show others the value and train thoroughly to build trust and confidence in AI tools.
What's Next for AI in Fragrance
The fragrance industry's AI journey is just beginning. Looking ahead 5-10 years, we can expect:
Advanced Generative Systems AI models will likely advance to proposing fully realized formulas requiring minimal tweaking, and may even suggest entirely new molecules that don't yet exist.
Digital Scent Communication Companies like Moodify envision being able to digitally communicate scent through devices similar to speakers, potentially creating a new medium for multisensory experiences.
Hyper-Personalization at Scale Soon, consumers might expect every perfume purchase to include some degree of personalization—perhaps an AI-tweaked formula based on their profile, effectively making each bottle unique.
Neuro-Responsive Fragrances The intersection of AI and neuroscience (as in YSL's EEG experiments) could lead to scents designed to trigger specific emotional or cognitive responses.
New Business Models "Fragrance-as-a-service" platforms might emerge where independent creators use AI tools to craft and sell niche perfumes without owning a lab.
As Dr. A.K. Pradeep, CEO of a neuro-scent AI company, notes: "We're entering an era where AI enables hyper-personalization, unlocking consumers' ability to articulate what they want and ensuring truly unique blends."
Your Turn
The fragrance industry's AI transformation offers a blueprint that applies across sectors:
Identify your creative bottlenecks: What processes in your field take months that could be compressed to days?
Map your data assets: What institutional knowledge exists only in your experts' heads that could be structured for AI?
Start small but think big: Begin with a focused use case while developing a comprehensive AI roadmap.
Build cross-functional teams: Combine technical expertise with domain knowledge from day one.
Remember: The goal isn't to replace human creativity with algorithms, but to unlock new possibilities by combining the best of both. As the fragrance industry demonstrates, even the most traditional, artisanal fields can be enhanced—not diminished—by thoughtful AI implementation.
Adapt & Create
Kamil
This case study is based on research into multiple fragrance companies including Givaudan's Carto platform, Symrise's partnership with IBM, DSM-Firmenich's Scentmate, and IFF's digital innovations, examining their approaches to AI adoption in fragrance creation. For more information on implementing AI strategies in your business, contact our team.
Great story and framework Kamil!