Search engines retrieve. LLMs reason. The difference goes far beyond interface design—it’s algorithmic at its core. In this session, Metehan unpacks what’s really happening under the hood when models like ChatGPT generate answers instead of ranking results. Drawing on live experiments and data from his analysis of Reciprocal Rank Fusion (RRF), he explains how LLMs expand a single user query into dozens of subqueries, merge their outputs through RRF, and then rerank results based on semantic overlap rather than keyword match.

You’ll see how embeddings, context windows, and reranking layers replace traditional signals like backlinks and CTR—and why this shift redefines “relevance” itself. Rather than guessing how AI search works, this talk maps it: from query fan-out to citation weighting, from candidate retrieval to generation. Attendees will leave with a concrete understanding of how information surfaces in LLM interfaces, how fusion algorithms shape what gets seen, and what this means for anyone optimizing in a world where search and synthesis are no longer separate.