Neo4j & GenerativeAI Fundamentals
Learn how knowledge graphs, vector search, and GraphRAG ground LLMs in facts. Build RAG pipelines and retrievers in Python to stop hallucination.
In this 2-hour course, you will learn
Neo4j & GenerativeAI Fundamentals teaches you how to ground large language models (LLMsA model trained on text to predict the next token, and so to generate language.) in facts using knowledge graphs, vector searchFinding the records whose vectors lie closest to a query vector., and GraphRAGRetrieval-augmented generation whose context comes from a knowledge graph, so the model can follow the relationships between facts.. In around two hours you will move from the core concepts of generative AIModels that produce new content rather than classifying or scoring content that already exists. to writing Python code that retrieves trusted answers from a Neo4j database.
LLMs are trained on public data with a fixed cutoff date, so they know nothing about your business, your documents, or anything that happened after training. When they lack information, they can respond with confident but false answers, known as hallucinationA false statement made confidently by a language model.. Retrieval Augmented Generation (RAGFetching relevant context from an external data source as additional context to inform a language model's response.) addresses this by supplying the model with relevant context at question time, and GraphRAG strengthens it further by drawing that context from a knowledge graphA representation of real-world entities and their relationships, stored according to organizing principles, typically in a graph database.. This course explains how the pieces fit together and gives you hands-on practice with each one.
Generative AI Fundamentals
Understand how generative AI models produce text, where hallucination originates, and why grounding responses in factual data is essential for production applications.
Large Language Models
Learn how LLMs process prompts, generate tokens, and use context windows — and apply that knowledge to craft effective prompts for graph-backed AI applications.
RAG
Build retrieval-augmented generation pipelines in Python that retrieve relevant documents from Neo4j and supply them as grounded context to an LLM at query time.
GraphRAG
Apply GraphRAG techniques to combine vector similarity search with graph traversal, retrieving richer, relationship-aware context that flat vector stores cannot provide.
Integrating Neo4j with Generative AI
Configure a Neo4j vector index, store and query embeddings, and wire a LangChain retriever to a Neo4j knowledge graph to serve accurate answers from structured data.
4 modules, 2 hours.
Who this course is for
This course is for developers, data scientists, and architects who want to build generative AIModels that produce new content rather than classifying or scoring content that already exists. applications that give accurate, explainable answers. If you have experimented with an LLMA model trained on text to predict the next token, and so to generate language. and wondered how to connect it to your own data, this course fills in the missing layer between the model and your database. You do not need any prior experience with generative AI, embeddingsInformation represented as a numerical vector, positioned so that similar information sits close together., or vector searchFinding the records whose vectors lie closest to a query vector. — every concept is introduced from the ground up. You should have completed Neo4j Fundamentals, be able to read simple Python programs, and have an OpenAI API key for the hands-on tasks, although you can adapt the code to another provider.
What you'll build and do
This is a practical course. You get a Neo4j sandbox pre-loaded with a movie recommendations dataset, complete with embeddingsInformation represented as a numerical vector, positioned so that similar information sits close together. and a vector indexA structure over a vector property. The database searches it to find the vectors nearest a given one, rather than comparing every vector stored., so you can concentrate on the retrieval techniques rather than the setup.
You will use the
neo4j-graphragPython package to run vector similarity searches against the movie plot embeddings and find films by meaning rather than keywords. You will assemble those searches into a complete RAGFetching relevant context from an external data source as additional context to inform a language model's response. pipeline that retrieves context, passes it to an LLMA model trained on text to predict the next token, and so to generate language., and returns a grounded answer. You will then upgrade the retrieverA component that searches a data source and returns the information relevant to a query. Often used to provide context for a language model. twice: first with a graph-enhanced vector retriever that traverses relationshipsA named, directed connection between two nodes. Every relationship has a type, a start node and an end node. to add connected facts to the context, and then with a text-to-Cypher retriever that translates natural language questions into CypherNeo4j's implementation of GQL, the ISO standard query language for graph databases. It is declarative: you describe the pattern to find, and the database decides how to find it. queries. Each step includes checks so you can confirm your code works before moving on.Where to go next
Once you have completed this course, Constructing Knowledge Graphs with Neo4j GraphRAG for Python is the natural next step: it shows you how to build knowledge graphs from your own unstructured data using the same Python package. If you work with LangChain, Using Neo4j with LangChain applies these retrieval patterns inside LangChain applications, and Introduction to Vector Indexes and Unstructured Data goes deeper on embeddingsInformation represented as a numerical vector, positioned so that similar information sits close together. and semantic search. This course is also the starting point for the Neo4j & Generative AI Certification, which validates your ability to build GraphRAGRetrieval-augmented generation whose context comes from a knowledge graph, so the model can follow the relationships between facts. applications with Neo4j.