Introduction
GraphRAG combines community summarization and multi-hop reasoning so the agent gets richer context than raw node properties alone.
In this lesson, you will learn how to use GraphRAG for advanced retrieval in Aura Agents.
What is GraphRAG?
GraphRAG combines graph structure with retrieval-augmented generation:
-
Community summarization: Summaries of subgraphs for richer context
-
Multi-hop reasoning: Follow relationships across the graph
-
Hybrid retrieval: Combine vector search with graph traversal
Enable GraphRAG when creating or editing your agent to give it access to these capabilities. See Aura Agent documentation for GraphRAG and tool configuration.
Why GraphRAG Helps
Without GraphRAG, the agent returns raw node data. With it:
-
Summaries instead of lists: The LLM gets condensed context about subgraphs, not just node properties
-
Multi-hop traversals: Queries like "customers → orders → products → suppliers" become practical
-
Vector + graph together: Semantic similarity finds starting points, graph traversal finds connections
Example Use Cases
GraphRAG works well for questions like:
-
"Which products are often bought together by customers in the same region?"
-
"What are the key themes in documents connected to this topic?"
-
"Summarize the relationships between these entities"
Deeper dive: Vector and text-to-Cypher retrievers
The Neo4j and Generative AI Workshop covers vector RAG, vector + Cypher retrievers, and text-to-Cypher retrievers in Python, plus building a conversational agent with LangChain.
Check Your Understanding
GraphRAG Capabilities
What does GraphRAG add to an Aura Agent?
-
❏ Only vector search
-
❏ Only Cypher query execution
-
✓ Community summarization and multi-hop reasoning over the graph
-
❏ Static documentation retrieval only
Hint
GraphRAG combines graph structure with retrieval. It can summarize communities and traverse relationships.
Solution
Community summarization and multi-hop reasoning over the graph.
GraphRAG provides richer context by summarizing subgraphs and communities and enabling multi-hop reasoning. It can be combined with vector search for hybrid retrieval.
Summary
In this lesson, you learned how to leverage GraphRAG for community summarization, multi-hop reasoning, and hybrid retrieval.