AI · AIC-92

GraphRAG and Agent Memory

GraphRAG and Agent Memory is an advanced follow-on to basic RAG, covering how to build knowledge graphs in Neo4j from documents with LLMs, use graph and Text2Cypher retrievers, and design long-term memory for AI agents with LangGraph and Graphiti. It suits AI engineers and Python developers who already run RAG systems and have hit the limits of vector search.

Updated 1 Oct 2026

From 7,110 THB / person 7,900 −10% excl. VAT 7% · group rates available

Corporate training quote Join the waiting list

PDFDownload the course outline

Duration12 hours · 2 days
FormatOnsite / live online
Next roundOn request
CertificateIncluded

Course overview

RAG systems built on vector search alone answer well when the answer sits in a single passage, but they often miss when a question has to connect several documents, such as which product models a customer uses that are affected by the latest notice, or when it asks about the whole dataset. At the same time, AI agents that work with users over time need to remember who said what and know which facts are out of date. Knowledge graphs have become an important data layer for both RAG and agent memory.

This course is the second step after RAG & Knowledge Base and shows learners how to build GraphRAG and agent memory with Neo4j and Python. Learners write Cypher, design a graph schema and build a knowledge graph from documents with an LLM using the neo4j-graphrag package, compare vector, vector plus graph and Text2Cypher retrievers, and learn the community summary approach of GraphRAG. They then move on to short-term and long-term agent memory with LangGraph and temporal knowledge graphs with Graphiti, measure the results against vector RAG, and put security and PDPA controls in place. It closes with a capstone GraphRAG agent. (2 days, 6 hours per day, 12 hours in total, Advanced level.)

What you’ll gain

Explain when GraphRAG is worth it compared with vector RAG, and when it is not needed
Write Cypher to create and query data in Neo4j, including vector indexes
Design a graph schema and build a knowledge graph from documents with an LLM
Use vector, vector plus graph and Text2Cypher retrievers to suit each question
Understand community summaries and the local and global search modes of GraphRAG
Design short-term and long-term agent memory with LangGraph
Build temporal memory that knows which facts have changed, using Graphiti
Measure GraphRAG against vector RAG and put security and PDPA controls in place

Who this course is for

AI engineers and Python developers who have built RAG systems and hit the limits of vector search
People who have completed RAG & Knowledge Base and want to take the next step
AI agent developers who want agents to remember users and context across sessions
Data engineers who need to connect several data sources for AI to use
Solution architects who must choose RAG and memory architectures for their organisation

Prerequisites

Working Python skills such as functions, classes and installing packages
An understanding of RAG, embeddings and vector search, or completion of RAG & Knowledge Base
Basic Docker, command-line and REST API skills
A laptop that can run Docker Desktop, Python and VS Code, with permission to install software

Curriculum

Course Details

The course runs for 2 days, 6 hours per day (12 hours in total, 09:00-16:00), as lectures with labs built on a single sample dataset throughout. Advanced level. It is the second step after RAG & Knowledge Base, which covers vector RAG, and adds a knowledge graph in Neo4j to answer questions that span several documents, plus memory for AI agents with LangGraph

and Graphiti. Every lab runs Neo4j Community Edition on Docker or the free tier of Neo4j Aura. Learners use their own or company Gemini API or OpenAI API accounts, which are billed by usage (the Gemini API has a free tier). Learners take home a lab guide, Python and Cypher code for every lab, a docker-compose file for Neo4j and a set of test questions.

Day 1 Knowledge Graphs and GraphRAG

Section 1: Why Vector RAG Alone Is Not Enough

Questions vector RAG misses: spanning documents, multi-level relationships and whole-dataset questions
What a knowledge graph is: nodes, relationships and properties
GraphRAG patterns: graph-enhanced vector search, Text2Cypher and community summaries
When a graph is worth it, and when vector RAG is enough
Lab: install Neo4j with Docker or use the free tier of Neo4j Aura

Section 2: Lab: Cypher for RAG

Create nodes and relationships with CREATE and MERGE
Query with MATCH, WHERE and multi-hop paths
Constraints, indexes and vector indexes in Neo4j
Common GraphRAG queries and reading results in Neo4j Browser

Section 3: Designing a Graph Schema from Documents

Define entities and relationships from the questions the business needs answered
Lexical graph: link documents and chunks to entities
Entity resolution: merge names written differently, including Thai and English names
Lab: design a schema for a sample product manual and company policies

Section 4: Lab: Building a Knowledge Graph with an LLM

Use SimpleKGPipeline from the neo4j-graphrag package to extract entities from text and PDFs
Give the LLM a schema so it extracts only what you need
Check the quality of the resulting graph and fix duplicate entities
Cost and time when building a graph from large document sets

Section 5: Lab: Different Retrievers

Vector and hybrid retrievers on Neo4j
VectorCypherRetriever: find chunks, then expand the context along relationships
Text2CypherRetriever: turn questions into Cypher, and what to watch out for
Lab: a Q&A system that picks a retriever by question type

Section 6: Whole-dataset Questions and Community Summaries

The Microsoft GraphRAG approach: community detection and per-community summaries
How local search, global search and DRIFT search differ
Lower-cost indexing approaches such as LazyGraphRAG and LightRAG
Lab: answer whole-dataset questions with community summaries

Day 2 Agent Memory and Production Use

Section 7: Agent Memory Fundamentals

Short-term and long-term memory, and semantic, episodic and procedural memory
What to remember, what not to remember, and when to forget
Vector memory compared with graph memory
A tool overview: the LangGraph store, Mem0 and Graphiti

Section 8: Lab: A LangGraph Agent That Uses the Graph as a Tool

An agent that chooses vector search, graph queries or Text2Cypher as tools
A checkpointer to remember the conversation within a thread
Long-term memory in a store, separated by user
Test the agent with questions that need several retrieval steps

Section 9: Lab: Temporal Memory with Graphiti

Store facts with the period they are valid and know which facts have been superseded
Add episodes from conversations and business events
Hybrid search across semantic, keyword and graph traversal
An agent that remembers customer needs and history across sessions

Section 10: Lab: Evaluating GraphRAG and Memory

Test question sets with single-hop, multi-hop and whole-dataset questions
Measure correctness, completeness and source citation
Compare vector RAG and GraphRAG on the same question set
Check for wrong or stale memories and how to correct them

Section 11: Security and Production Use

Prevent Cypher injection with read-only access and checking queries before they run
Per-user access control in both the graph and memory
PDPA: personal data in memory, the right to erasure and retention periods
Incremental graph updates when documents change, with latency and cost in mind

Section 12: Workshop: GraphRAG Agent Capstone

Choose a brief: a technical manual assistant, a policy assistant or a customer assistant that remembers history
Build the graph, retrievers and memory for the system
Measure it against vector RAG on the same questions
Present the work and a plan for next steps

Schedule & training options

For individuals — public rounds

No public rounds are open right now. Join the waiting list and we will contact you first when the next round opens, or ask us on LINE. Or call 02-570-8449 or 088-807-9770

Join the waiting list Ask about rounds

For organisations — in-house / private

Tailor the content to your team’s tools and projects
Your dates, at your office or live online
Quotation with tax ID for procurement

Corporate training quote

Instructors

Samit KoyomInstructor

Frequently asked questions

Who is GraphRAG and Agent Memory for, and what background is needed?

Built for AI engineers and Python developers who have built RAG systems and hit the limits of vector search · People who have completed RAG & Knowledge Base and want to take the next step · AI agent developers who want agents to remember users and context across sessions Background you should have: Working Python skills such as functions, classes and installing packages · An understanding of RAG, embeddings and vector search, or completion of RAG & Knowledge Base Not sure the fit is right? Talk to our team on LINE @itgenius or call 02-570-8449.

How much does GraphRAG and Agent Memory cost and how long does it run?

THB 7,900 (currently THB 7,110 on promotion). The course runs 12 hours. The price excludes 7% VAT (for payment in a company's name). Pay by bank transfer to the company account, confirm it on our payment page, and we can issue the receipt or tax invoice in your company's name.

Do I get a certificate?

Yes. Everyone who completes the course receives a Certificate of Completion from IT Genius Institute. Each certificate carries its own number, and anyone holding that number can verify it online on our certificate page, so you can add it to your portfolio or pass it to HR as evidence of training.

Where does the training take place, and is there an online option?

You can attend onsite at IT Genius Institute or arrange to join online, and we also run it as a private in-house session for your team. Ask about dates and venues on LINE @itgenius or call 02-570-8449.

What if I fall behind or miss a session — can I retake it?

Yes. You may retake the same course free of charge in a later round, under the institute's conditions. Tell our team which course and round you attended, and we will check it and offer you the rounds that still have seats. Ask us on LINE @itgenius or call 02-570-8449.

How do I enrol, or request a quotation for my company?

Enrol online with the registration form on this page. You can register several attendees at once and enter your tax ID and billing address for the tax invoice. Or request a company quotation straight from the quote button. For anything else call 02-570-8449 or reach us on LINE @itgenius.