This Week in Neo4j: MCP, Deployment, Vector, GraphRAG and more

Photo of Alexander Erdl

Alexander Erdl

Senior Developer Marketing Manager

Vaibhava Lakshmi Ravideshik

Welcome to This Week in Neo4j, your fix for news from the world of graph databases!

We are celebrating the general availability of the official Neo4j MCP Server, making it easier than ever to connect Claude, VS Code and other MCP hosts to any Neo4j deployment.
You’ll also learn how to deploy Neo4j securely on a VPS, explore the new native VECTOR data type and build Python-based GraphRAG knowledge graphs with our latest GraphAcademy course.

Only a few more days until the Call for Papers for NODES AI closes: Submit now and be on stage April 15, when practitioners from all over the world will gather virtually to explore AI, context engineering, and intelligent agents.

Happy Graphing,

Alexander Erdl

 

COMING UP!

Vaibhava Lakshmi Ravideshik is an AI engineer, specializing in RAG pipelines, knowledge graphs, and AI-driven automation. She designs intelligent AI agents that use graph-based reasoning, retrieval optimisation, and multi-agent workflows.

Connect with her on LinkedIn.

Vaibhava delivered one of the top-rated sessions at NODES2025: “Adaptive Knowledge Graphs for Dynamic RAG Systems”. She showed how to build self-improving GraphRAG architectures using Neo4j, Cypher, vector search, and adaptive ML techniques, enabling AI systems to think more contextually, retrieve more precisely, and evolve.


Vaibhava Lakshmi Ravideshik

 

MCP: Neo4j MCP


Our official Neo4j MCP Server is now available in version 1.0.1 and works with common MCP hosts, such as Claude Desktop, VS Code and others, for all types of Neo4j deployments: Neo4j Aura, Self-Managed, Docker, Neo4j Desktop and Neo4j Sandbox. Get it from Github and explore our MCP Developer pages for more info.

 

DEPLOYMENT: Deploying Single Instance Neo4j On A VPS With Custom Domain


Will Lyon provides a solid reference for anyone looking to run a secure, production-ready Neo4j instance outside of managed services. In the article, he deploys Neo4j on a VPS, covering everything from Java installation and APOC setup to domain configuration, HTTPS via Let’s Encrypt, firewall rules, and restoring from backups.

 

VECTOR: Introducing Neo4j’s Native Vector Data Type


David Pond gives us the latest about the new built-in VECTOR data type. Embedding vectors can be stored as fixed-length, typed properties rather than generic lists. This change simplifies application code, enforces vector shape/dtype constraints for safety, and unlocks future vector-specific optimisations in storage, indexing and GenAI workflows.

 

GraphRAG: Constructing Knowledge Graphs with Neo4j GraphRAG for Python


This newest course from GraphAcademy shows you how to use the official neo4j-graphrag Python package to build knowledge graphs from unstructured text, enrich them with structured data, and create retrievers that let LLMs query graphs via Cypher or hybrid vector-graph search.

 

 

POST OF THE WEEK: rmoff

A fun few hours with Cursor, neo4j, and Cytoscape analysing astroturfing and sock-puppet accounts on Reddit.

Common pattern of a handful of accounts posting 'questions', and then another comments on it.

Maybe an attempt at seeding LLMs, which scrape Reddit?

[image or embed]

— rmoff 🏃‍♂️🫖🥓 (@rmoff.net) 13. November 2025 um 19:44



Please share it if you like it!