Innovating with AWS: Next-Level Personalization with LLM-Powered Content Recommendations

In the rapidly evolving digital landscape, staying ahead means leveraging cutting-edge technology to enhance user engagement and maximize value from content assets. At Blueshift, we’re excited to introduce our latest innovation in personalized content recommendations, powered by large language model (LLM) embeddings developed in partnership with Amazon Bedrock. These new capabilities leverage generative AI technologies’ power to better understand user preferences and optimize recommendations to each user’s taste, preferences, and relevance.

The new LLM embedding algorithms go beyond traditional recommendation systems by interpreting the semantic context of content and user interactions. This advanced approach ensures tailored recommendations across various formats — whether users read articles, listen to podcasts, watch videos, or browse product catalogs. The algorithm can suggest similar types of content, such as related videos, items, or articles, and it also excels in cross-promoting different content types. For example, it might recommend a basketball training video to someone who has bought a basketball shoe or suggest basketball gear to go with their shoes. Additionally, it can effectively propose add-ons and accessories for purchased products, enhancing the user experience and boosting revenue opportunities. This ability to understand and connect user interests across formats ensures a richer, more engaging digital journey for every user.

At the core of the LLM recommendations is AWS Bedrock’s robust infrastructure, which supports a diverse array of LLM models, ensuring both flexibility and scalability for handling complex catalogs and data formats. LLM models can interpret the rich multi-modal data within a client’s catalog, such as item titles, descriptions, prices, categories, brands, tags, and images. These attributes are meticulously analyzed using LLM to discern the contextual essence of each item, creating precise embeddings.

These embeddings then enable the generation of highly personalized recommendations by identifying items with similar embeddings throughout the entire catalog. The system offers relevant and timely recommendations, coupled with insights from a user’s recent activity, significantly enhancing the user experience. The integration with AWS Bedrock not only allows the use of their native LLM models, such as Titan and Claude, but also supports custom and OpenAI models, providing a comprehensive and adaptable solution for personalized content discovery, as illustrated below.

Media and Publishing

The media and publishing industry thrives on keeping audiences engaged and subscribed. Timely content recommendations are key to achieving this, as they ensure readers are continuously presented with articles that match their interests and preferences. For example, a leading media publisher uses our LLM-based recommendations to suggest similar news articles, enhancing reader engagement by keeping content relevant and timely in their recurring newsletters. This intelligent curation helps maintain high engagement levels and increases readers’ time on the platform.

E-commerce and Retail

In e-commerce and retail, personalized recommendations transform browsing intent into buying moments. A leading beauty products retailer exemplifies the power of these 1:1 recommendations by linking skin care sales to relevant content such as blogs and how-to videos. They trigger in-the-moment campaigns to recommend relevant skincare brands and related audio and video content as users browse the site and apps. At each stage of the buying cycle, from browsing to consideration to purchase and post-purchase, each user is connected with relevant content as they embark on their skin care regimen. From eye creams and body lotions to face serums, users enjoy receiving tips and tricks and the right application tools to complete their skin care regimen and make addressing their skin care needs a rewarding journey.

A leading automotive parts retailer utilizes these new LLM-powered recommendations to connect DIY enthusiasts, pro shops, and everyday car owners to the right fitment replacement parts or add-ons, along with articles and videos. As these users browse through a vast catalog of parts and accessories, they are presented with only relevant items and content suited to their vehicle needs throughout their browse, consider, buy, and build journey.

Travel and Leisure

The travel and leisure sector requires a keen understanding of cross-catalog content to effectively engage customers. Rest Less, a UK-based digital platform tailored to people over 50 with over 1 million members, harnesses the strength of LLM-based recommendations to offer readers personalized travel deals and bespoke vacation packages related to the content they are reading. For example, it suggests island vacation packages to someone exploring a blog about Mediterranean destinations. This ability to recommend relevant cross-catalog content boosts engagement and drives conversions by aligning offers with user interests.

Deals and Promotional Platforms

A coupon and deal publishing leader had a platform specializing in deals and promotions. This company leverages LLM-based recommendations to recommend deals that are not just relevant but are priced similarly, catering to users’ budget preferences. This strategic recommendation approach helps maximize the relevance of offers, increasing user satisfaction and the likelihood of coupon redemption. These platforms can drive more focused and effective promotional campaigns by ensuring users see deals matching their price expectations and content preferences.

Security and Compliance

Our seamless integration with AWS services ensures your data is constantly managed and used securely with state-of-the-art encryption and data handling practices. Your data is never used for training or shared with any other third party.

Designed for Marketing Teams

Along with 1:1 recommendations, our platform empowers your marketing team with tools to:

• Segment users effectively, creating tailored campaigns that resonate on a personal level.
• Launch and manage campaigns across multiple channels, all from a single integrated platform.
• Track performance and iterate with comprehensive analytics that provide insights into what strategies work best.

Adopting Blueshift’s LLM-powered recommendation engine means providing your users with uniquely personalized experiences that drive engagement, value and loyalty. Join leading brands utilizing the most advanced technology to understand better and serve your customers.

Blueshift is not just a platform but a partnership in your brand’s journey towards unparalleled digital engagement. We are an AWS Marketplace Partner and have achieved the AWS Advertising and Marketing Technology Competency. Contact us today to see how our innovative solutions can transform your customer interactions.

Ready to make the most of your data? Create your free account now.