Beyond GraphRAG: Building Neuro-Symbolic AI Systems with Neo4j and Ontologies

Session track: AI Engineering

Session time: 12:30 am - 01:00 am UTC November 13

Session description:

GraphRAG has become one of the most popular patterns for grounding LLMs with enterprise data. Yet, many GraphRAG architectures still rely heavily on embeddings and graph traversal while overlooking a fundamental ingredient of knowledge representation: ontologies. In this session, we explore how neuro-symbolic AI combines the strengths of neural models and symbolic reasoning to overcome limitations such as hallucinations, inconsistent answers, and lack of explainability. Rather than treating Neo4j as a simple retrieval layer, we use it as the foundation for representing domain knowledge, semantic relationships, constraints, and business rules through ontological models. Attendees will learn how ontologies can guide retrieval, improve contextual understanding, enable semantic validation, and support reasoning workflows that go far beyond traditional RAG and GraphRAG architectures. Through real-world examples and practical design patterns, we will demonstrate how Neo4j, knowledge graphs, and LLMs can work together to create AI systems that are not only knowledgeable but also semantically grounded and easier to trust. If GraphRAG is the bridge between LLMs and enterprise knowledge, neuro-symbolic AI is the next step toward truly intelligent systems.