Create customized merchandise and advertising and marketing campaigns utilizing Amazon Nova in Amazon Bedrock

This put up was written with Jake Friedman from Wildlife.

Companies are in search of modern methods to distinguish themselves by way of hyper-personalization and enhanced buyer experiences. On the Cannes Lions Worldwide Pageant of Creativity 2025, AWS showcased The Perfume Lab, an interactive and provoking expertise that demonstrates how generative AI can help the event of hyper-personalized shopper items and speed up promoting inventive idea and marketing campaign property improvement. Following Cannes Lions 2025, The Perfume Lab acquired a Gold and Silver Stevie Award from the Worldwide Enterprise Awards within the Model & Experiences class.

Constructed utilizing Amazon Nova in Amazon Bedrock, The Perfume Lab represents a complete end-to-end utility that illustrates the transformative energy of generative AI in retail, shopper items, promoting, and advertising and marketing. Whereas our activation at Cannes Lions centered on customized perfume improvement and advert marketing campaign creation, the underlying structure and methodology will be tailored throughout numerous classes, from trend to meals and beverage, opening countless prospects for custom-made buyer experiences.

Introducing The Perfume Lab

On this put up, we discover the event of The Perfume Lab. Our imaginative and prescient was to craft a novel mix of bodily and digital experiences that may have fun creativity, promoting, and shopper items whereas capturing the spirit of the French Riviera. To convey this imaginative and prescient to life, we collaborated with Wildlife, an organization that’s distinctive at remodeling AWS generative AI providers into compelling bodily experiences. Wildlife was elementary in brainstorming concepts that may encourage clients and showcase novel use circumstances that AI makes attainable.

Crafting the perfume

As step one, the expertise used Amazon Nova Sonic, a speech-to-speech mannequin that engages in intuitive dialogues with attendees to grasp their persona and preferences. Nova Sonic extends its capabilities by way of device integration, permitting it to handle person traits and interface actions by way of specialised instruments equivalent to addTraitTool, removeTraitTool, and uiActionIntentTool. These instruments assist keep dialog state and a constant stream all through the expertise. The collected dialog information and trait data are then processed by way of a customized Retrieval Augmented Era (RAG) system constructed with Amazon Nova Professional, a extremely succesful multimodal mannequin that provides our greatest mixture of accuracy, velocity, and value. Nova Professional serves because the intelligence engine for analyzing interactions and extracting important key phrases to find out the right perfume notes and composition. The appliance additionally used Amazon Bedrock Guardrails, which affords customizable safeguards and accountable AI insurance policies to dam undesirable matters—equivalent to allergens or dangerous content material—to supply a seamless buyer expertise.

For instance, a buyer would possibly share with Nova Sonic that they’re curious about journey. Nova Professional picked up that exploring new locations typically “brings a way of freshness and pleasure,” which resulted in a perfume that feels recent and invigorating, that includes “a burst of citrus or a floral breeze.” The shopper may also share that they take pleasure in early morning walks throughout spring fields, which Nova Professional interprets right into a prime notice of recent bergamot, a center notice that includes floral honey, and a base of lavender. The shoppers’ inputs information the number of perfume notes—from base, to coronary heart, to prime notes—which have been then expertly combined by on-site perfumers to create actually customized scents. Perfumers have been capable of customise and craft a whole lot of distinctive fragrances per day, aided by AI. A course of that may usually take hours for a perfumer was accelerated to minutes, empowering each the shopper and the perfume knowledgeable.

Creating the marketing campaign

After the customized perfume formulation was created and despatched to the perfumer queue, Amazon Nova Canvas generated custom-made advertising and marketing inventive, together with the perfume title, tagline, and imagery that captured the essence of the formulation. Attendees have been capable of additional customise the marketing campaign property utilizing visitor inputs equivalent to moody, beachy, or playful. The ensuing perfume picture was then reworked into dynamic video content material by way of Amazon Nova Reel, which clients might additional customise to fulfill their inventive imaginative and prescient and obtain to avoid wasting or share. To match the Cannes Lions ambiance, the marketing campaign movies have been generated with a French-accented feminine voice utilizing Amazon Polly. All the expertise is inbuilt Amazon Bedrock, a completely managed service to construct and scale generative AI functions with AI fashions.

The next information stream diagram exhibits how a number of Amazon Nova fashions will be mixed for a wealthy, cohesive, and customized buyer expertise.

Finest practices for implementation

The Perfume Lab facilities round interactions with Amazon Nova Sonic, offering customers with a pure language interface to precise their preferences for a customized scent. By way of its device integration capabilities, Nova Sonic orchestrates the complete expertise by managing person traits and triggering applicable workflows. These workflows seamlessly information the expertise from preliminary dialog to perfume improvement and in the end to marketing campaign asset creation, driving each the visible components and development of the expertise. The mannequin’s potential to keep up a conversational state, whereas defining clear conversational flows, helps guarantee a constant and nice expertise for each person.

A well-defined workflow and conversational assistant are pivotal in guiding these conversations to uncover the qualities which can be most essential to every person. And the system immediate determines the persona, type, and content material of your conversational assistant.

Immediate instance:

You're an AI assistant designed to assist the person discover their persona and 
emotional panorama within the context of making a novel perfume. You interact in heat, 
free-flowing, playful dialog with the person to attract out their character, 
preferences, moods, and needs. Your finish aim is to derive a set of three to five persona 
traits that greatest describe the person. These traits will later be utilized in a separate 
course of to match applicable perfume elements. Your tone is heat, stylish, and subtly 
playful.

Further contextual data inside the immediate additionally performs a key position in Amazon Nova Sonic successfully sustaining state, whereas defining the conversational stream helps guarantee constant, nice, and concise experiences for each person.

Immediate instance:

1. **Welcoming Customers**
    Welcome the person to the applying expertise with a short overview of the
    course of and ask if they're able to proceed.
2. **Assistant Turns** 
    Ask brief and distinctive open ended inquiries to the person and select a persona trait 
    that you simply suppose would swimsuit the person greatest.
3. **Dealing with Person Turns**
    Acknowledge the person's solutions briefly and warmly.
    Deal with one trait per flip.
    Name the "addTraitTool", "removeTraitTool", "replaceTraitTool", or "clearTraitsTool" 
    instruments to handle traits.
    If the person says to return, skip, customise, or affirm/submit it means it's best to 
    name the "uiActionIntentTool" 

With direct references to our instruments within the conversational stream, the person interface feels reactive and linked to the person’s enter whereas offering alternatives for the assistant to exhibit its experience on this topic, which comes into the highlight when person traits and preferences are later mapped to a set of obtainable elements and uncooked perfume supplies.

This complicated perfume recipe improvement is dealt with by Nova Professional, utilizing its accuracy and velocity to generate persistently high-quality scents. To attract from a wealth of perfume information in actual time, RAG was applied to increase Nova Professional capabilities past pre-trained information with entry to information sources that embody important scent design ideas, a deep understanding of every accessible ingredient, their profiles and potential roles inside the perfume, and their attainable connections to customers’ fragrant identities.

The ensuing fragrances are then visualized utilizing Nova Canvas and Nova Reel. The inventive fashions generate unique compositions that reveal the perfume title, elements, and a visible id inside a high-end inventive marketing campaign asset. A set of conditioning pictures that includes unbranded perfume bottles assist to anchor every picture (as proven within the following picture).

Immediate instance:

A high-end perfume advert surroundings impressed by a [persona description]. A transparent, 
unbranded fragrance bottle is visually centered and tightly framed. Key elements [top 
note ingredient], [middle note ingredient], [base note ingredient], and [booster 
ingredient] are organized to encompass the bottle in a balanced composition, showing 
behind, apart from, and partially in entrance of the bottom. The scene evokes [atmospheric/mood 
descriptors] utilizing [light/color language]. The setting ought to really feel [stylistic direction],
like a [reference style (e.g., fashion editorial, lifestyle spread, luxury campaign)].

Outcomes

Attendees at Cannes Lions took away a bodily perfume combined by on-site perfumers. Whereas growing hyper-personalized shopper items won’t be scalable throughout all use circumstances, manufacturers can innovate with synthetic intelligence and obtain manufacturing outcomes that weren’t beforehand attainable. The promoting marketing campaign idea and asset improvement use case is straightforward to implement for manufacturers, businesses, and media networks, permitting customers to iterate and optimize marketing campaign inventive rapidly. Utilizing Amazon Bedrock, further options might be added like translations and sizes, relying on necessities.

You’ll be able to watch a video stroll by way of of The Perfume Lab onsite at Cannes Lions 2025, and take a look at the next instance marketing campaign outputs.

Conclusion

The Perfume Lab demonstrates the ability of Amazon Nova in Amazon Bedrock and the way clients can create absolutely customized shopper experiences. This use case will be replicated throughout varied retail and shopper items classes together with skincare and cosmetics, trend and equipment, meals and beverage, residence items, and wellness merchandise—all benefiting from pure dialog interplay, AI-powered product improvement, product id, and artistic advertising and marketing marketing campaign technology. Get began with Amazon Nova in Amazon Bedrock at this time.

Concerning the authors

Raechel Frick is a Sr Product Advertising Supervisor at AWS. With over 20 years of expertise within the tech trade, she brings a customer-first strategy and progress mindset to constructing built-in advertising and marketing packages. Based mostly within the better Seattle space, Raechel balances her skilled life with being a soccer mother and after-school carpool supervisor, demonstrating her potential to excel each within the company world and household life.

Gaby Ferreres is the Head of Business Advertising for Media & Leisure, Sports activities, Video games, Promoting & Advertising at AWS, the place she works with expertise and trade leaders to speed up innovation on behalf of consumers. She is a worldwide advertising and marketing chief and creator of experiences that elevate buyer journeys. Earlier than AWS, she held totally different positions at Microsoft, Telefonica, and extra.

Ashley Weston is Sr. Advertising Occasion Supervisor for World Third-Social gathering Applications at AWS, the place she companions with trade advertising and marketing to ship the very best visibility and most business-critical occasions for AWS.

Tiffany Pfremmer is Sr. Business Advertising Supervisor at Amazon Net Companies (AWS) the place she leads strategic built-in advertising and marketing initiatives throughout the Media & Leisure, Video games, and Sports activities verticals to ship advertising and marketing campaigns that join AWS cloud options with buyer alternatives.

Jake Friedman is the President and Co-founder at Wildlife, the place he leads a staff launching interactive experiences and content material campaigns for world manufacturers. His work has been acknowledged with the Titanium Grand Prix on the Cannes Lions Worldwide Pageant of Creativity for “boundary-busting, envy-inspiring work that marks a brand new course for the trade and strikes it ahead”. Yow will discover him on LinkedIn.

About Wildlife

Wildlife fuses a digitally born skillset with a future proof mindset to ship breakthrough merchandise, experiences and campaigns for daring companions. We stay by a motto: Know-how modifications, story doesn’t.