AI Engineer
← All organizations

Graph databases, analytics, and AI infrastructure

Neo4j

Neo4j builds graph databases, analytics, and AI tools that let developers and enterprises query connected data, investigate relationships, and supply context to AI applications. Neo4j Aura offers managed database, analytics, and agent tools; self-managed Graph Database and Graph Data Science products run within customers’ infrastructure. Native vector search complements knowledge graphs for AI retrieval. Neo4j completed its acquisition of GraphAware on August 5, 2026, adding Hume, an intelligence-analysis platform built on Neo4j for government agencies and other organizations.

Co-founded by Emil Eifrem, Johan Svensson, and Peter Neubauer, Neo4j remains led by CEO Eifrem. Its engineering roots lie in a content-management system whose connected data required complicated SQL queries. The team replaced its relational backend with native graph storage to address the cost of recursive joins. Its Cypher query language made connected data easier to create and query; the openCypher project opened the language to adoption across multiple graph database vendors in 2015.

In 2024, the company reported more than $200 million in annual recurring revenue, double its level three years earlier, and use by 84% of Fortune 100 companies and 58% of the Fortune 500. The same announcement disclosed $50 million in capital from Noteus Partners, reaffirming a $2 billion company valuation.

neo4j.com ↗

17 talks

Newest first

11 speakers at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Stephen ChinVP of Developer RelationsZach BlumenfeldAI Research EngineerBen SquireSenior developer advocateRyan KnightSenior Partner ArchitectAndreas KolleggerDirector of Applied AI ResearchZaid ZaimAI developer advocateEmil EifremCEOAlex GilmoreMichael HungerNeo4j’s Head of Product Innovation and Developer StrategyJesús BarrasaAI Field CTOPhilip RathleNeo4j’s chief technology officer and an influential advocate for GraphRAG

Start here

  1. GraphRAG: The Marriage of Knowledge Graphs and RAG

    Start with the essay based on Emil Eifrem's talk to understand the basic GraphRAG retrieval pattern and why constructing the graph remains a central implementation challenge.

    Emil EifremAI Engineer World's Fair 2024

  2. CrabRAG: Why Automated Assistants Need Graph Memory, Not More Tokens

    Stephen Chin’s home-network demonstration shows how vector-seeded traversal and Cypher queries uncover outdated software and exposed management interfaces that similarity search misses.

    Stephen ChinAI Engineer World's Fair 2026

  3. Understanding AI Stakes to Break Production Code

    Philip Rathle’s interactive roundtable examines how the consequences of incorrect answers change deployment requirements, including human oversight, interface design, and model cost tradeoffs.

    Philip RathleAI Engineer World's Fair 2024

  4. Why your agents need decision traces, not just documents — Zach Blumenfeld, Neo4j

    Neo4j’s Zach Blumenfeld explains how context graphs extend document-oriented RAG with connected decision traces and short-term, long-term, and reasoning memory, helping agents act on prior decisions rather than merely retrieve facts.

    Zach BlumenfeldAI Engineer Europe 2026

Messages from the stage

Retrieval depends on graph construction

Emil Eifrem presents graph traversal as complementary to vector search, with development benefits conditional on having a usable knowledge graph. Jesús Barrasa shows how domain entities and relationships guide extraction and enrich retrieval.

Memory includes decisions and their context

Zach Blumenfeld examines connected decision traces and questions about assessing their quality. Andreas Kollegger and Zaid Zaim discuss reasoning memory alongside the risks and costs of autonomous actions.

Reusable structure across enterprise systems

A lakehouse workshop led by Zach Blumenfeld builds document outlines, thematic communities, and connection paths linking documents to warehouse records. Emil Eifrem proposes a shared semantic layer that maps business concepts to technical systems and learns from agent execution feedback.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-27