Plongez dans les coulisses de l’indexation vectorielle et son impact sur l'accuracy. Démystification des algorithmes HNSW et DiskANN, comparaison de leurs forces et cas d’usage; présentation de JVector, une approche hybride pour des recherches précises et évolutives. Au programme: concepts clés, retours d’expérience, démo live et bonnes pratiques pour intégrer efficacement la vector search dans vos applications.

For today’s Gen-AI apps, fast performance, instant scalability, and cost-effectiveness are more critical than ever. This session will delve into the importance of these factors when building RAG pattern apps while maintaining low costs. We will explore the capabilities of Azure Cosmos DB and its new vector database capabilities using DiskANN, a technology developed by Microsoft Research. With DiskANN, users can achieve low latency, high-recall vector search at any scale. Combined with Azure Cosmos DB’s unique scale-out architecture and instant autoscale, it provides enormous value with a cost profile unmatched by any vector database in the market today. This allows for the development of large-scale applications that are not only powerful and reliable but also economical. Join us and discover how to architect high accuracy, low latency, and cost-effective RAG pattern applications at any scale using Azure Cosmos DB and DiskANN. Regardless of your role, this session will provide valuable insights into bringing this new generation of applications to your business.