Presentations

AWS + Neo4j

Presentations on grounded enterprise AI with Neo4j, from the cloud data and AI landscape to fraud investigation, semantics, and agent memory. The resources below link to starter code and a hands-on workshop.

Landscape, architecture, and use cases

Cloud Data and AI Stacks How AWS, Databricks, Google Cloud, IBM, and Microsoft build the same data and AI stack, where they differ, and where a customer-owned graph fits. Overview Multi-cloud AWS Main Services The main AWS products behind each layer of the data and AI stack, from S3 Tables and Neptune to Bedrock, AgentCore, Amazon Quick, and Kiro, and where Neo4j plugs in. AWS Bedrock AgentCore Neo4j + AWS: Grounding Generative AI in Graph Data A fraud-investigation overview of Neo4j on AWS: connected context, Virtual Graph, Amazon Quick over MCP, GraphRAG agents, and agent memory. Fraud GraphRAG MCP Fraud Data Architecture How AWS transaction evidence and Neo4j connected context work together to expose a fraud ring. Fraud Architecture

Semantics and knowledge

Neocarta: A Semantic Map for Enterprise Data How Neocarta connects metadata in Neo4j and turns it into context for agents, with a query example and optional technical detail. Semantics Metadata Neosemantics: Bringing RDF Semantics Into Neo4j How Neosemantics imports, maps, validates, reasons over, and exports RDF with Neo4j, with concrete graph examples in the technical appendix. Semantics RDF Enterprise Knowledge Layer How a shared, governed Knowledge Layer gives every agent the same business meaning, source routing, policy, and decision traces. Semantics Governance

Code and workshops

Starter code and a hands-on workshop for building grounded agents with Neo4j on AWS.

Starter kit AWS Bedrock AgentCore Starter Kit The starter kit deploys the Neo4j MCP server to Amazon Bedrock AgentCore behind an OAuth2 Gateway. Its samples show several ways for AI agents to call that server.
  • Demos End-to-end GraphRAG demos for an aircraft fleet, fraud investigation, and SEC filings.
  • Integrations Connections from Databricks Unity Catalog and Neo4j Aura Agents to the graph.
  • Patterns An OAuth2 Gateway with a Lambda interceptor that enforces role-based access.
View on GitHub →
Workshop GraphRAG with Neo4j on AWS: From Search to Grounded Agents Build a hotel booking agent that answers from a connected graph. You deploy its tools and the agent to AgentCore, then add graph memory in Neo4j.
6 modules Hands-on notebooks Bedrock AgentCore MCP Graph memory Basic Python
Open workshop → Source code →