As Generative AI apps evolve from simple prompts to complex, multi-agent systems, the need for a unified, cost-effective, and highly scalable data foundation has become critical. Azure Cosmos DB, available both in Azure and natively within Microsoft Fabric, stands out as the most scalable and cost-efficient database to power agentic applications.
In this session, we’ll explore data modelling and design patterns for Azure Cosmos DB to support all core data needs for AI apps: from chat history and agentic memory to knowledge retrieval across unstructured, structured, and vector data. We’ll also explore powerful new capabilities in Azure Cosmos DB including vector search, full-text search, and semantic reranking, that transform the database into a high-performance retrieval engine for building intelligent, context-aware AI agents.
You’ll learn how to design end-to-end agent architectures that take advantage of Azure Cosmos DB’s capabilities:
- Agent Memory & Chat History: Persist multi-turn conversations and long-term memory using scalable, globally distributed storage.
- Knowledge Retrieval: Enhance agent responses with Retrieval-Augmented Generation (RAG), combining low-latency vector search (DiskANN) with filtering, full-text search, and semantic reranking for maximizing relevance.
- Function Calling & State Management: Enable agents to execute structured queries and maintain continuity across tasks.
- Multitenancy & Cost Efficiency: Build tenant-aware systems optimized for tenant-specific workloads that scale instantly and dynamically.
Whether you’re a seasoned AI developer, data scientist, solution architect, or just getting started with agentic applications, this session will equip you with practical, scalable patterns for using Azure Cosmos DB as the foundational database to power your AI applications.
