🎥 Recorded live at the MLOps World | GenAI Summit 2025 — Austin, TX (October 8, 2025) Session Title: From Vectors to Agents: Managing RAG in an Agentic World Speaker: Rajiv Shah, Chief Evangelist, Contextual AI Talk Track: Agents in Production Abstract: The RAG (Retrieval-Augmented Generation) landscape has evolved at lightning speed — from simple keyword search to semantic embeddings to multi-step agentic reasoning. With this evolution comes a new challenge: context engineering. In this talk, Rajiv Shah helps attendees cut through the hype and focus on practical, production-ready RAG architectures. You’ll learn when to use Speedy Retrieval (500 ms) for low-latency tasks, Accuracy-Optimized RAG (10s) for balanced trade-offs, and Exhaustive Agentic Search for complex reasoning pipelines. Rajiv walks through three distinct architectural patterns — complete with code examples, context orchestration techniques, and decision frameworks — to help teams pick the right approach for their use case. What you’ll learn: • When to use speed-first retrieval vs. agentic search (and when not to) • How to optimize context engineering and window management in RAG systems • Why “good enough” retrieval often beats “perfect” agentic reasoning in production • How agentic search compares in cost, latency, and complexity vs. vector search • A practical framework for designing scalable RAG architectures.