GraphRAG and Explainable AI: Building Trustworthy LLM Outputs - Rohit Bhardwaj

This session explores the concept of trust failures in enterprise large language models (LLMs), focusing on issues like hallucination and lack of traceable provenance in regulated industries. The speaker introduces GraphRAG, a novel approach that integrates knowledge graphs using Neo4j with retrieval-augmented generation to provide transparent and auditable AI outputs. Attendees will learn how to design, evaluate, and implement GraphRAG architectures in compliance with frameworks such as the EU AI Act and NIST AI Risk Management Framework. Through practical examples, the session demonstrates how GraphRAG connects entities and source documents to enhance evidence-based reasoning and boost confidence in AI applications.