GraphAcademy
Course

Neo4j Agent Memory Workshop

Build an AI agent with persistent, explainable memory backed by Neo4j - short-term, long-term, and reasoning - using the neo4j-agent-memory package.

3 hours 20 minutes19 lessons across 5 modules
About this course

In this 3-hour course, you will learn

In this workshop, you will build an AI agent with persistent, explainable memory backed by Neo4j using the neo4j-agent-memory package.

You will work through the three layers of agent memory - short-term conversation, long-term knowledge, and reasoning traces - and see how they connect as a single context graph in Neo4j. Each lesson pairs a short explanation with code you run in your own environment.

By the end of the workshop, you will have built a Pydantic AI learning assistant that records what a learner understands, grounds its recommendations in a course knowledge graph, and writes its reasoning trace into Neo4j.

  • Agent Memory

  • neo4j-agent-memory

  • Context Graphs

  • Reasoning Traces

  • POLE+O

Course content

5 modules, 3 hours 20 minutes.

  1. 4 lessons
    1. Installation and Configuration6 min
    2. The Starting AgentOptional8 min
    3. An Agent That Forgets9 min
    4. The Same Agent, With Memory10 min
  2. 3 lessons
    1. Configure the Memory Client7 min
    2. Add Short-term Memory to Your Agent12 min
    3. The Short-term Memory APIOptional8 min
  3. 6 lessons
    1. Add Long-term Memory to Your Agent7 min
    2. Entities and RelationshipsOptional13 min
    3. DeduplicationOptional7 min
    4. Choosing Your Extractor11 min
    5. Facts and Preferences6 min
    6. The Facts and Preferences APIOptional5 min
  4. 4 lessons
    1. Record the Reasoning6 min
    2. Querying Traces3 min
    3. The Reasoning APIOptional13 min
    4. Combined ContextOptional5 min
  5. 2 lessons
    1. Write Your Own Tool8 min
    2. Make It YoursOptional45 min