Neo4j and AWS Bedrock GraphRAG Workshop

View the full workshop guide

A hands-on workshop teaching Graph Retrieval-Augmented Generation (GraphRAG) patterns using Neo4j Aura and Amazon Bedrock. You will build and query a knowledge graph of SEC 10-K financial filings, then connect AI agents that retrieve structured and unstructured data to answer questions about companies, risk factors, and institutional ownership.

Workshop Structure

Part 1: Setup & Visual Exploration with Neo4j (Labs 0-2)

Lab	Title	Description
Lab 0	Sign In	AWS Console sign-in and Bedrock access verification
Lab 1	Neo4j Aura Setup	Sign up for Neo4j Aura, load knowledge graph via Cypher, explore graph
Lab 2	Aura Agents	Build a no-code AI agent using Neo4j Aura Agent platform

Part 2: Building GraphRAG Agents (Labs 3-5)

Lab	Title	Description
Lab 3	Intro to Bedrock and Agents	SageMaker setup, Strands Agents SDK, tool binding, ReAct pattern, AgentCore deployment
Lab 4	neo4j-graphrag Library	Load chunk embeddings, vector retrieval, and vector-cypher retrieval over a knowledge graph
Lab 5	Neo4j MCP Server	Strands Agents with MCP: tool discovery, Cypher Templates, and Text2Cypher patterns

Part 3: Bonus — Build Your Own Pipeline

Lab	Title	Description
Lab 6	GraphRAG Pipeline (Bonus)	Build a GraphRAG data pipeline from scratch: data loading, embeddings, and vector-cypher retrieval

Prerequisites

• AWS Account with Bedrock access (or workshop credentials via OneBlink)
• Basic Python knowledge (for Labs 3-6)

Quick Start

Option 1: AWS SageMaker Studio (Recommended for workshops)

Follow Lab 3 to set up SageMaker Studio and clone the repository.

Option 2: Local Development

git clone https://github.com/neo4j-partners/neo4j-bedrock-graphrag-workshop.git
cd neo4j-bedrock-graphrag-workshop

# Copy and fill in your credentials
cp CONFIG.txt CONFIG.txt.local
# Edit CONFIG.txt with your Neo4j and AWS credentials

Start with Lab 0 for AWS setup instructions.

Configuration

All credentials are stored in CONFIG.txt at the project root (gitignored). The file uses dotenv format:

NEO4J_URI=neo4j+s://xxx.databases.neo4j.io
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=your_password_here
MODEL_ID=us.anthropic.claude-sonnet-4-5-20250929-v1:0
REGION=us-east-1

See CONFIG.txt for all available settings grouped by lab.

Technology Stack

Component	Technology
Knowledge Graph	Neo4j Aura
Foundation Models	Amazon Bedrock (Claude Sonnet)
Embeddings	Amazon Nova Multimodal Embeddings
Agent Frameworks	Strands Agents SDK
GraphRAG Library	neo4j-graphrag
Agent Protocol	Model Context Protocol (MCP)

Architecture

User Query → AI Agent → Tool Selection
                              ↓
        ┌─────────────────────┴─────────────────────────┐
        ↓                     ↓                         ↓
  Vector Search         Text2Cypher              Cypher Template
        ↓                     ↓                         ↓
  Nova Embeddings        Claude LLM                Direct Query
        ↓                     ↓                         ↓
        └─────────────────────┴─────────────────────────┘
                              ↓
                       Neo4j Aura
                              ↓
                    SEC 10-K Knowledge Graph

Contributing

We welcome contributions! To report bugs or suggest improvements, open an issue at: https://github.com/neo4j-partners/neo4j-bedrock-graphrag-workshop/issues

License

This project is licensed under the MIT License - see the LICENSE file for details.