Build Daily

Tinley Park · August 23, 2026
Neo4jAug 14, 2026

The Graph Gathering 2026 with Emil Eifrem

Open on YouTube ↗← Back to the feed
What it covers

Emil Eifrem introduces The Graph Gathering 2026, an annual un-conference exploring the intersection of graphs and AI with leading theorists and practitioners. He highlights key discussions on ontologies, agentic memory, context graphs, graph RAG, and enterprise-wide knowledge layers.

The outline

Tap a timestamp to jump there in the video.

  1. 0:00

    Event Introduction

    Emil Eifrem introduces The Graph Gathering as an annual Neo4j event focused on the frontier of graphs and AI.

  2. 0:18

    Participant Profile

    The event invites about 20 sharp individuals who combine theoretical expertise with real-world practical experience in graphs and AI.

  3. 0:52

    Un-conference Format

    The gathering operates as an un-conference where specific topics are generated bottom-up from the room rather than being pre-planned.

  4. 1:08

    Key Discussion Themes

    Participants converged on major discussions around ontologies, agentic memory, and context graphs and their intersections.

  5. 1:41

    Ontology Approaches

    The group debated top-down versus bottom-up ontology derivation, concluding that a hybrid approach with human-in-the-loop is best for unstructured data.

  6. 2:12

    Agentic Memory and RAG

    Discussion explored how agentic memory overlaps with RAG corpus state and context graphs in this early-stage space.

  7. 2:27

    Graphs Across the AI Stack

    Emil outlines several areas where graphs overlap with AI, starting with the retrieval layer known as graph RAG.

  8. 2:44

    Graph RAG Success

    Retrieval augmented generation using knowledge graphs is identified as a successful early application of graphs in AI.

  9. 2:59

    Agentic Memory Potential

    Graph-based implementations are expected to thrive significantly in the area of agentic memory.

  10. 3:14

    Enterprise Knowledge Layer

    A major emerging pattern involves consolidating data into a unified enterprise-wide knowledge or context layer for agents.

Small rooms, sharp edges

A twenty-person un-conference keeps the signal dense enough for solo builders to spot which graph-AI patterns are ready to ship. Ontologies and context graphs matter most when they reduce integration work rather than add ceremony.