Agents as Search Engineers - Building Smarter Search Systems with LLM Agents

Learn how to deploy AI agents as search engineers to transform brittle keyword-based search systems into intelligent, adaptive solutions in this conference talk from the Coding Agents Conference at the Computer History Museum. Discover practical patterns for implementing agent-based search improvements including query rewriting, hybrid retrieval, and agent-based reranking that can be deployed in real production systems. Explore how large language models enable agents to act like search engineers by rewriting queries, disambiguating user intent, and judging relevance dynamically, moving beyond traditional hand-engineering and expensive labeling approaches. Gain insights from real-world deployment experiences including both successes and challenges when implementing these systems at scale. Understand the wins, pitfalls, and open questions surrounding agent-based search systems while learning how to make search smarter, faster, and more adaptive without creating black box solutions. Walk away with a practical playbook for leveraging agents to enhance search functionality in digital products where search serves as the primary user interface.