Introduction
You will complete your agent by adding a Text2Cypher query tool as the third tool.
The agent automatically chooses the best tool for each question type:
Schema Tool:
Document Retrieval Tool:
Database Query Tool:
-
Precise queries and counts
-
"How many risk factors does Apple face?"
-
"What stock has Microsoft issued?"
Open the notebook: 02_03_text2cypher_agent.ipynb
Your complete agent will now have:
-
Tool 1: Schema Tool (database structure exploration)
-
Tool 2: Document Retrieval Tool (vector search + graph context)
-
Tool 3: Database Query Tool (text-to-Cypher for precise queries) ← NEW!
This creates a comprehensive GraphRAG agent that can handle any type of question intelligently.
Try These Questions
Try these questions to see all three tools in action:
-
"What products does Microsoft mention in its financial documents?"
-
"How many risk factors does Apple face and what are the top ones?"
-
"What stock has Microsoft issued?"
-
"What are the main risk factors mentioned in the documents?"
-
"Summarize Apple’s risk factors and how they relate to other companies"
The agent will choose from the 3 tools available:
-
Graph Exploration (Schema Tool)
-
Contextual Search (Vector + Cypher Tool)
-
Precise Queries (Text2Cypher Tool)
Notice: The agent intelligently selects the right tool(s) for each question type! Complex questions may invoke multiple tools.
Summary
In this lesson, you completed your GraphRAG agent by adding the Text2Cypher Retriever as the third tool:
Key Concepts:
-
Complete tool suite: All three retrievers now available as conversational tools
-
Intelligent routing: Agent automatically selects best tool(s) for each question
-
Progressive capability: From simple search to complex multi-tool reasoning
What You Built:
-
Complete GraphRAG agent with three retriever tools
-
Conversational interface to all retriever capabilities from previous modules
-
Intelligent tool selection for optimal answers
Your Journey:
-
✅ Knowledge Graph Creation: PDF to Knowledge Graph pipeline
-
✅ Retriever Development: Built three different retrievers
-
✅ Agent Tools: Converted retrievers to conversational agent tools
Final Result: A complete GraphRAG agent that can answer any question using the most appropriate retrieval strategy automatically!