In this webinar, participants learn how to use LlamaIndex property graph abstractions in collaboration with Neo4j to build advanced Retrieval Augmented Generation (RAG) systems. The session covers several key topics: constructing and querying knowledge graphs, low-level details on graph construction and retrieval, and utilizing knowledge graph agents for complex reasoning. The presenters, including Tomaz from Neo4j and Logan from LlamaIndex, explain the differences between property graphs and traditional triple stores, and demonstrate various graph constructors and retrievers available in LlamaIndex. They also discuss the importance of entity disambiguation and provide practical examples using GPT-4 and other LLMs. The session concludes with a demonstration of custom retrieval methods and encourages participants to explore knowledge graphs for enhancing RAG systems.