This post was written with Jake Friedman from Wildlife.
Businesses are seeking innovative ways to differentiate themselves through hyper-personalization and enhanced customer experiences. At the Cannes Lions International Festival of Creativity 2025, AWS showcased The Fragrance Lab, an interactive and inspiring experience that demonstrates how generative AI can support the development of hyper-personalized consumer goods and accelerate advertising creative concept and campaign assets development. Following Cannes Lions 2025, The Fragrance Lab received a Gold and Silver Stevie Award from the International Business Awards in the Brand & Experiences category.
Built using Amazon Nova in Amazon Bedrock, The Fragrance Lab represents a comprehensive end-to-end application that illustrates the transformative power of generative AI in retail, consumer goods, advertising, and marketing. While our activation at Cannes Lions focused on personalized fragrance development and ad campaign creation, the underlying architecture and methodology can be adapted across diverse categories, from fashion to food and beverage, opening endless possibilities for customized customer experiences.
Introducing The Fragrance Lab
In this post, we explore the development of The Fragrance Lab. Our vision was to craft a unique blend of physical and digital experiences that would celebrate creativity, advertising, and consumer goods while capturing the spirit of the French Riviera. To bring this vision to life, we collaborated with Wildlife, a company that is exceptional at transforming AWS generative AI services into compelling physical experiences. Wildlife was fundamental in brainstorming ideas that would inspire customers and showcase novel use cases that AI makes possible.
Crafting the fragrance
As the first step, the experience used Amazon Nova Sonic, a speech-to-speech model that engages in intuitive dialogues with attendees to understand their personality and preferences. Nova Sonic extends its capabilities through tool integration, allowing it to manage user traits and interface actions through specialized tools such as addTraitTool, removeTraitTool, and uiActionIntentTool. These tools help maintain conversation state and a consistent flow throughout the experience. The collected conversation data and trait information are then processed through a custom Retrieval Augmented Generation (RAG) system built with Amazon Nova Pro, a highly capable multimodal model that offers our best combination of accuracy, speed, and cost. Nova Pro serves as the intelligence engine for analyzing interactions and extracting essential keywords to determine the perfect fragrance notes and composition. The application also used Amazon Bedrock Guardrails, which offers customizable safeguards and responsible AI policies to block undesirable topics—such as allergens or harmful content—to offer a seamless customer experience.
For example, a customer might share with Nova Sonic that they are interested in travel. Nova Pro picked up that exploring new places often “brings a sense of freshness and excitement,” which resulted in a fragrance that feels fresh and invigorating, featuring “a burst of citrus or a floral breeze.” The customer might also share that they enjoy early morning walks across spring fields, which Nova Pro translates into a top note of fresh bergamot, a middle note featuring floral honey, and a base of lavender. The customers’ inputs guide the selection of fragrance notes—from base, to heart, to top notes—which were then expertly mixed by on-site perfumers to create truly personalized scents. Perfumers were able to customize and craft hundreds of unique fragrances per day, aided by AI. A process that would normally take hours for a perfumer was accelerated to minutes, empowering both the customer and the fragrance expert.
Creating the campaign
After the personalized fragrance formula was created and sent to the perfumer queue, Amazon Nova Canvas generated customized marketing creative, including the fragrance name, tagline, and imagery that captured the essence of the formula. Attendees were able to further customize the campaign assets using guest inputs such as moody, beachy, or playful. The resulting fragrance image was then transformed into dynamic video content through Amazon Nova Reel, which customers could further customize to meet their creative vision and download to save or share. To match the Cannes Lions atmosphere, the campaign videos were generated with a French-accented female voice using Amazon Polly. The entire experience is built in Amazon Bedrock, a fully managed service to build and scale generative AI applications with AI models.
The following data flow diagram shows how multiple Amazon Nova models can be combined for a rich, cohesive, and personalized customer experience.
Best practices for implementation
The Fragrance Lab centers around interactions with Amazon Nova Sonic, providing users with a natural language interface to express their preferences for a custom scent. Through its tool integration capabilities, Nova Sonic orchestrates the entire experience by managing user traits and triggering appropriate workflows. These workflows seamlessly guide the experience from initial conversation to fragrance development and ultimately to campaign asset creation, driving both the visual elements and progression of the experience. The model’s ability to maintain a conversational state, while defining clear conversational flows, helps ensure a consistent and pleasant experience for every user.
A well-defined workflow and conversational assistant are pivotal in guiding these conversations to uncover the qualities that are most important to each user. And the system prompt determines the personality, style, and content of your conversational assistant.
Prompt example:
Additional contextual information within the prompt also plays a key role in Amazon Nova Sonic effectively maintaining state, while defining the conversational flow helps ensure consistent, pleasant, and concise experiences for every user.
Prompt example:
With direct references to our tools in the conversational flow, the user interface feels reactive and connected to the user’s input while providing opportunities for the assistant to demonstrate its expertise on this subject, which comes into the spotlight when user traits and preferences are later mapped to a set of available ingredients and raw fragrance materials.
This complex fragrance recipe development is handled by Nova Pro, using its accuracy and speed to generate consistently high-quality scents. To draw from a wealth of fragrance knowledge in real time, RAG was implemented to extend Nova Pro capabilities beyond pre-trained knowledge with access to knowledge sources that include essential scent design principles, a deep understanding of each available ingredient, their profiles and potential roles within the fragrance, and their possible connections to users’ aromatic identities.
The resulting fragrances are then visualized using Nova Canvas and Nova Reel. The creative models generate original compositions that reveal the fragrance name, ingredients, and a visual identity within a high-end creative campaign asset. A set of conditioning images featuring unbranded fragrance bottles help to anchor each image (as shown in the following image).
Prompt example:
Results
Attendees at Cannes Lions took away a physical fragrance mixed by on-site perfumers. While developing hyper-personalized consumer goods might not be scalable across all use cases, brands can innovate with artificial intelligence and achieve manufacturing outcomes that weren’t previously possible. The advertising campaign concept and asset development use case is easy to implement for brands, agencies, and media networks, allowing users to iterate and optimize campaign creative quickly. Using Amazon Bedrock, additional features could be added like translations and sizes, depending on requirements.
You can watch a video walk through of The Fragrance Lab onsite at Cannes Lions 2025, and check out the following example campaign outputs.
Conclusion
The Fragrance Lab demonstrates the power of Amazon Nova in Amazon Bedrock and how customers can create fully personalized consumer experiences. This use case can be replicated across various retail and consumer goods categories including skincare and cosmetics, fashion and accessories, food and beverage, home goods, and wellness products—all benefiting from natural conversation interaction, AI-powered product development, product identity, and creative marketing campaign generation. Get started with Amazon Nova in Amazon Bedrock today.
About the authors
Raechel Frick is a Sr Product Marketing Manager at AWS. With over 20 years of experience in the tech industry, she brings a customer-first approach and growth mindset to building integrated marketing programs. Based in the greater Seattle area, Raechel balances her professional life with being a soccer mom and after-school carpool manager, demonstrating her ability to excel both in the corporate world and family life.
Gaby Ferreres is the Head of Industry Marketing for Media & Entertainment, Sports, Games, Advertising & Marketing at AWS, where she works with technology and industry leaders to accelerate innovation on behalf of customers. She is a global marketing leader and creator of experiences that elevate customer journeys. Before AWS, she held different positions at Microsoft, Telefonica, and more.
Ashley Weston is Sr. Marketing Event Manager for Global Third-Party Programs at AWS, where she partners with industry marketing to deliver the highest visibility and most business-critical events for AWS.
Tiffany Pfremmer is Sr. Industry Marketing Manager at Amazon Web Services (AWS) where she leads strategic integrated marketing initiatives across the Media & Entertainment, Games, and Sports verticals to deliver marketing campaigns that connect AWS cloud solutions with customer opportunities.
Jake Friedman is the President and Co-founder at Wildlife, where he leads a team launching interactive experiences and content campaigns for global brands. His work has been recognized with the Titanium Grand Prix at the Cannes Lions International Festival of Creativity for “boundary-busting, envy-inspiring work that marks a new direction for the industry and moves it forward”. You can find him on LinkedIn.
About Wildlife
Wildlife fuses a digitally born skillset with a future proof mindset to deliver breakthrough products, experiences and campaigns for daring partners. We live by a motto: Technology changes, story doesn’t.