{"id":29969,"date":"2025-12-17T11:13:12","date_gmt":"2025-12-17T04:13:12","guid":{"rendered":"https:\/\/renovacloud.com\/?post_type=story-pt&p=29969"},"modified":"2026-03-11T13:39:13","modified_gmt":"2026-03-11T06:39:13","slug":"pnj-enhancing-jewelry-shopping-with-genai-powered-intent-visual-search-on-aws","status":"publish","type":"story-pt","link":"https:\/\/renovacloud.com\/en\/success-stories\/pnj-enhancing-jewelry-shopping-with-genai-powered-intent-visual-search-on-aws\/","title":{"rendered":"PNJ: Enhancing Jewelry Shopping with GenAI-Powered Intent & Visual Search on AWS"},"content":{"rendered":"

OVERVIEW<\/b><\/h2>\n

As one of Vietnam\u2019s most iconic jewelry brands, PNJ has built a reputation for craftsmanship, sophistication, and exceptional customer experience. With the rapid shift toward digital shopping, PNJ sought to redefine how customers discover products online\u2014moving beyond traditional text search toward a more intuitive, image-driven, AI-powered search experience.<\/span><\/p>\n

To achieve this vision, PNJ partnered with <\/span>Renova Cloud<\/b> to build a <\/span>GenAI-powered Intent & Visual Search<\/b> on AWS (<\/strong><\/span>Amazon Web Services)<\/b>. This modern, multimodal system empowers customers to find jewelry using natural-language descriptions, uploaded photos, or a combination of both, unlocking a new level of convenience and personalization.<\/span><\/p>\n

 <\/p>\n

KEY CHALLENGES<\/b><\/h2>\n
    \n
  • Product scale and complexity outpacing discovery capabilities:<\/strong> PNJ\u2019s rapidly expanding catalog, with frequent new collection launches and thousands of visually nuanced SKUs, made it difficult for both customers and staff to effectively navigate and differentiate products using traditional tools.<\/li>\n
  • Slow product knowledge adoption for sales associates:<\/strong> The volume and visual subtlety of products extended onboarding time for new hires, delaying their ability to confidently understand, recommend, and sell products.<\/li>\n
  • Ineffective search and browsing experience for digital customers:<\/strong> Conventional keyword-based search and category browsing failed to capture the subjective, visual nature of jewelry preferences, limiting relevance and usability.<\/li>\n
  • High friction in product discovery and decision-making:<\/strong> Customers were forced to rely on extensive scrolling and manual comparison, resulting in engagement fatigue, longer decision cycles, and difficulty narrowing options by design style, stone shape, or aesthetic intent.<\/li>\n
  • Lost engagement and conversion opportunities:<\/strong> The absence of intent-aware and visual discovery mechanisms increased session abandonment and reduced the overall efficiency of the purchase journey.<\/li>\n<\/ul>\n

     <\/p>\n

    SOLUTION<\/b><\/h2>\n

    To address these challenges, PNJ and Renova Cloud designed a <\/span>production-ready, multimodal GenAI search platform<\/b> capable of delivering fast, intelligent, and personalized product discovery.<\/span><\/p>\n

    1. Architecture & Deployment<\/b><\/h3>\n

    The solution integrates multiple AWS-managed services into a unified search pipeline:<\/span><\/p>\n

    \"Architectures
    Image 1. PNJ enhances jewelry shopping experience with GenAI-powered intent & visual search on AWS.<\/figcaption><\/figure>\n

     <\/p>\n

      \n
    • Amazon Bedrock:<\/b> Generates embeddings for images and text, enriches product data with AI-generated visual descriptions, and powers multimodal intent matching.<\/span><\/li>\n
    • Amazon OpenSearch Service (vector database):<\/b> Stores and indexes embeddings for large-scale similarity search across PNJ\u2019s extensive product catalog.<\/span><\/li>\n
    • Amazon S3:<\/b> Serves as the central storage for product images and uploaded customer images, as well as product metadata files.<\/span><\/li>\n
    • AWS Lambda & Step Functions:<\/b>.Core computation and orchestration, coordinates preprocessing workflow for new data ingestion and query recommendations<\/span><\/li>\n
    • Amazon API Gateway:<\/b> Provides an entry point for secured API traffic between client and server side.<\/span><\/li>\n
    • Amazon CloudWatch + AWS X-Ray:<\/b> Thorough monitoring and observability capability for PNJ teams <\/span><\/li>\n
    • Amazon DynamoDB:<\/strong> Storing system\u2019s metadata for each processing phase for tracking and monitoring<\/span><\/li>\n
    • AWS Secrets Manager, IAM & KMS:<\/strong> IAM controls access, KMS encrypts data, and Secrets Manager protects credentials\u2014together forming the security foundation of the architecture.<\/li>\n<\/ul>\n

       <\/p>\n

      2. Search Capabilities Delivered<\/b><\/h3>\n
      \"Solving
      Image 2. Solving complex customer intent with GenAI-powered intent & visual search on AWS<\/figcaption><\/figure>\n
        \n
      • Image Search:<\/b> Customers upload a jewelry photo to find similar styles instantly.<\/span><\/li>\n
      • Intent Search:<\/b> Natural-language queries interpret style, material, shape, occasion, or design preferences.<\/span><\/li>\n
      • Hybrid Search:<\/b> Combines image + text for best-match results (e.g., <\/span>\u201cSimilar ring but in rose gold\u201d<\/span><\/i>).<\/span><\/li>\n
      • Automated Metadata Enrichment:<\/b> GenAI creates standardized visual descriptions that improve catalog accuracy and downstream analytics.<\/span><\/li>\n<\/ul>\n

        BENEFITS<\/b><\/h2>\n

        1. Financial Impacts<\/b><\/h3>\n

        Accelerated Sales Readiness & Reduced Training Costs:<\/b>
        \n<\/b>In traditional retail environments, sales associates are expected to memorize extensive product catalogs and complex product attributes. As PNJ continuously introduces new jewelry collections and hundreds of SKUs, this approach leads to prolonged onboarding cycles, higher training costs, and inconsistent product consultations. Traditionally, new sales associates require <\/span>approximately 4\u20136 weeks<\/b> to become familiar with the product catalog and confidently assist customers.<\/span><\/p>\n

        By implementing a <\/span>GenAI-powered Intent Search and Visual Search platform<\/b>, sales associates can instantly retrieve relevant products using natural language descriptions or reference images. The system achieves <\/span>88.5% Top-3 search accuracy<\/b>, allowing both experienced and newly hired staff to quickly identify suitable products without manually navigating large catalogs. Acting as an <\/span>AI-assisted knowledge augmentation tool<\/b>, the solution reduces reliance on memorization and enables employees to access product knowledge in real time, significantly accelerating the path to sales readiness.<\/span><\/p>\n

        Business Outcomes:<\/b><\/p>\n