See what your AI sees
Browse your graph, build queries with the Cypher Designer and check what your agent actually stored. Profile, save and share what you find.
Read more about GDB ExplorerNewConnect Claude, Cursor or Codex to your database→
Managed Neo4j for AI apps and agents
Get a production-ready graph database in minutes. We take care of backups, scaling and security, so you can focus on what you're building.
Our customers
Trusted by thousands of teams since 2015
Why GrapheneDB
Neo4j Community, production-ready
Community is the free edition of Neo4j. On its own it leaves out what production needs. We add it and run it for you.
See how it compares to AuraDB- No Enterprise license feesSame Cypher and driversNo lock-in
- See what your database is doingMetricsSlow query logsGrafana endpoint
- Built for productionSnapshotsRead replicas24×7 on-call
- Dedicated and privateDedicated serversVPC peeringIP allowlisting
Built for AI
Graph and vectors, one database
Neo4j stores embeddings next to your nodes, with built-in vector indexes. Find similar items, then follow their relationships, in a single Cypher query. No separate vector store to keep in sync.
- GraphRAG
Retrieval that pulls in the related facts, not only text that looks similar, so answers are complete and traceable.
- Agent memory
What your agent learns, kept as nodes and relationships, so it remembers people, tasks and decisions across sessions.
- Graph ML
Centrality, communities and node embeddings as features for your fraud, churn and recommendation models.
① Vector search1CALL db.index.vector.queryNodes('doc_embedding', 5, $q)2YIELD node AS doc, score② Follow relationships3MATCH (doc)-[:MENTIONS]->(p:Product)-[:HAS_POLICY]->(pol)4RETURN doc.title, p.name, pol.text, score
The platform
Everything Neo4j needs in production
Backups, replicas, private networking, metrics and logs, handled for you on dedicated AWS servers.
Run Neo4j Community in production with the tools it needs: backups, read replicas, database metrics, query logs and more.
- Hot snapshots
- Read replicas
- Read-only users
- Database metrics
- Slow queries
Explorer and MCP
Your graph, in front of you and your AI
Explore your graph visually in the browser, or let Claude, Codex and Cursor work with it directly through GrapheneDB's MCP server.
Why is the orders query slow on production?
Checking database metricsReading the slow query logProfiling the query
It scans every Order node before filtering by status. An index on that property should fix it. Want to see the query plan?
Your database, inside Claude, Codex and Cursor
GrapheneDB runs a hosted MCP server. Sign in from any MCP client and your assistant can check metrics and logs, profile a slow query and run Cypher. It only sees what your user can see.
Read more about AI access through MCPUse cases
Not only AI
The same graph database runs the rest of your business: anywhere the connections between your data matter as much as the data itself.
Fraud detection
Spot fraud rings through shared devices, cards and addresses.
Read more about fraud detectionRecommendations
"Customers like you also bought" in one query, in real time.
Read more about recommendationsKnowledge graphs
Connect the facts from your documents and databases so anyone can explore them.
Read more about knowledge graphsIdentity and access
Who can reach what, through every role and nested group.
Read more about identity and accessNetwork and IT ops
Know which services go down when a component fails.
Read more about network and it opsSupply chain
See every supplier behind a product, tiers deep.
Read more about supply chain
Compare
GrapheneDB vs AuraDB
The same Neo4j, on your own server, and yours to take with you. Here's how two typical setups compare with both AuraDB plans.
| Feature | GrapheneDB | AuraDB Professional | AuraDB Business Critical |
|---|---|---|---|
| 16 GB database, per month | $359.55 | About $1,050 ($0.09 per GB per hour) | About $2,340 ($0.20 per GB per hour) |
| 8 GB on 3 nodes (1 primary + 2 read replicas), per month | $551.37 | ✕Read replicas only on Business Critical | About $3,500 (each replica billed like the primary) |
| Dedicated server, not shared |  | ✕Only on Virtual Dedicated Cloud | ✕Only on Virtual Dedicated Cloud |
| Backups | Up to every 8 hours, 2 years retention | Daily, 7 days retention | Hourly, 30 days retention |
| 24x7 on-call team for incidents | On every plan | ✕Best-effort support | One-hour response |
| Export your data and load it into Neo4j Community | | ✕Backups use an Enterprise-only format | ✕Backups use an Enterprise-only format |
| Run Neo4j Community yourself later | | ✕Not for 36 months after the contract (terms) | ✕Not for 36 months after the contract, unless your order form says otherwise (terms) |
| Vector search and MCP | | | |
| Graph algorithms (GDS) | GDS plugin | Separate Graph Analytics product | Separate Graph Analytics product |
| Your own plugins | Upload any .jar | ✕Built-in procedures and part of APOC only | ✕Built-in procedures and part of APOC only |
Neo4j, Neo4j AuraDB and related marks are trademarks of Neo4j, Inc. GrapheneDB is not affiliated with, endorsed by or sponsored by Neo4j, Inc. AuraDB details come from Neo4j's public pricing page and self-serve terms and Cloud Agreement as of November 2026, and may have changed since. AuraDB prices assume 730 hours a month, with read replicas billed as in Neo4j's billing docs.
Empowering businesses of all sizes, since 2015
Since with GrapheneDB you can rapidly prototype and easily deploy without any hand-holding, the NGP VAN team has started thinking about many more problems they can solve with a graph model and, consequently, many more features they can deliver to the end-user.
FAQ
Questions, answered
Anything else about running Neo4j with us? Our team is happy to help.
We run Neo4j for you on dedicated AWS instances: deploys, upgrades, backups, monitoring and scaling. You get a connection URL and use the standard Neo4j drivers.