Watch more from .local Bengaluru: https://www.youtube.com/playlist?list=PLdZHbuVOoAyM Subscribe to the MongoDB for Developers YouTube Channel: https://www.youtube.com/@MongoDBDevelopers?sub_confirmation=1 Most LLM applications today remain stateless, treating every interaction as if it were the first. This creates a fundamental barrier to building AI systems that can learn from feedback, retain context, and improve over time. In this session, explore how developers can move beyond ephemeral context windows to build stateful, memory-rich agentic systems using MongoDB and Voyage AI. Through a practical walkthrough, you’ll see how hybrid search, reranking, and agent memory work together to improve retrieval quality, preserve context across interactions, and enable more reliable agent behavior. We’ll show how MongoDB Atlas can serve as the unified data layer for modern AI applications—combining operational data, vector search, knowledge retrieval, and agent memory in a single platform. The session will also explore the core memory patterns behind production-grade agents and how agents can use external tools and data sources to adapt to changing context over time. The result is a practical framework for building AI systems that can reason, remember, and deliver more accurate, personalized experiences. Chapters: 00:00:05 Introduction: Why Most AI Projects Fail 00:01:44 The MIT AI Study & The Stateless Problem 00:02:44 Limitations of Basic RAG vs. Agentic AI 00:04:10 Multi-Agent Architecture Overview 00:05:00 Optimizing Latency: Natural Language vs. Slash Commands 00:08:05 Designing a 5-Tier Human-Like AI Memory Layer 00:11:34 Context Engineering & 83% Token Reduction 00:13:00 Advanced Hybrid Search and Native Reranking in MongoDB Atlas 00:18:18 Action-Based Learning & Model Context Protocol (MCP) Loops 00:22:19 Key Architecture Takeaways & Wrap Up Visit Mongodb.com → https://mdb.link/MongoDB Read the MongoDB Blog → https://mdb.link/Blog Read the Developer Blog → https://mdb.link/developerblog MongoDB for Developers YouTube Channel → https://www.youtube.com/@MongoDBDevelopers

 145      5