Neo4j: The Definitive Guide

Foreword

Graphs are everywhere you look: information is connected and doesn’t exist in isolation. Especially now, in the age of smart agentic systems, having a reliable and performant full-stack database engine that is built from the ground up to deal with richly connected information is critical to grounding your LLM’s language skills in trusted, contextual facts. Knowledge graphs are digital twins of your business that allow you to ask and answer more comprehensive questions and find the deeper insights hidden in your data. GraphRAG lets you use advanced retrieval augmented generation (RAG) patterns to make LLM output explainable and contextually grounded.

My own journey with graphs started in the 1990s, when I accidentally reinvented the Dijkstra pathfinding algorithm while building client-side tooling for a multiuser dungeon online text adventure. Later, in 2008, I met Emil Eifrem, one of the founders of Neo4j, at a geek-cruise conference on the Baltic Sea. I was intrigued to hear about Neo4j for the first time, as I was working in retail applications. I wanted to know more about the applications of graph models and queries in the complex hierarchies of data. I started building open source integrations for the Neo4j database (which was back then only a small Java library, hence the “4j” name) and joined the small Swedish startup in 2010 as employee number 10. I worked on all parts of the platform, contributing to everything from the kernel to Cypher (the world’s best query ...