Introduction
Now that you understand the three types of retrievers, let’s learn how to make them conversational by wrapping them as agent tools.
Agents = Retrievers + Conversation Framework
What is an Agent?
An agent is a conversational wrapper around the retrievers you already know:
-
Vector Retriever → Vector Tool
-
Vector + Cypher Retriever → Hybrid Tool
-
Text2Cypher Retriever → Structured Query Tool
The agent adds:
-
User query analysis
-
Natural language conversation interface
-
Automatic tool selection based on question type
-
Flexible context management
The agent’s job is to choose the right retriever tool:
-
"Tell me about cybersecurity threats" → Use Vector for semantic exploration
-
"Which companies are affected by banking regulations?" → Use Vector+Cypher for contextual relationships
-
"What risks does Apple face?" → Use Text2Cypher for precise company data
Agent Intelligence: Agent Reasoning
Agent reasoning:
-
Analyze the user’s question
-
Determine what type of information is needed
-
Select the appropriate retriever tool
-
Execute the tool and format the response
-
Maintain conversation context
Progressive Learning Path
You’ll build the agent incrementally:
First Agent: Basic agent with schema introspection tool
Enhanced Agent: Add document retrieval tool
Complete Agent: Add text-to-Cypher tool
Each step adds one more tool - building complexity step by step.
The retrievers do the heavy lifting - the agent makes them conversational and smart about when to use each one.
Getting Started
In the next lessons, you’ll progressively build an agent by adding retriever tools one by one:
-
Schema tool for data model exploration
-
Vector + Cypher tool for semantic exploration and contextual relationships
-
Text2Cypher tool for precise queries
Each lesson shows how to wrap a retriever as a tool and how the agent chooses between available tools.
Summary
In this lesson, you learned that agents are conversational wrappers around the retrievers you already know:
Key Concept:
-
Retrievers = The core functionality (Vector, Vector+Cypher, Text2Cypher)
-
Agents = Conversational interface + intelligent tool selection
-
Tools = Retrievers wrapped for agent use
What Agents Add:
-
User query analysis
-
Natural language conversation interface
-
Automatic tool selection based on question type
-
Flexible context management
Progressive Learning:
-
First Agent: Schema tool → Database exploration
-
Enhanced Agent: Add document retrieval → Semantic search with context
-
Complete Agent: Add text2cypher → Structured queries
The retrievers do the work - the agent makes them conversational and intelligent about when to use each one.
In the next lesson, you will start building your agent with schema introspection as your first tool.