Why not Neo4j for this path?

Neo4j is an excellent general-purpose graph database. JetGraph is built for a different job: sub-millisecond fraud and risk signals on every event, on a single in-memory node.

Different products

Calling JetGraph “a Neo4j replacement” is incorrect. JetGraph speaks an openCypher subset and Bolt so teams can reuse some drivers and query literacy. It does not implement the full Neo4j Cypher surface, GDS, Fabric, or Neo4j’s clustered database architecture.

JetGraph is optimized for

  • Decision-path latency (O(1) velocity, novelty, neighbor risk)
  • Real-time graph signals at authorize time
  • Streaming ingest (BatchMutate) plus score-on-the-same-path
  • Automatic risk propagation
  • Single-node deterministic performance

Neo4j is optimized for

  • Richer graph traversal and a much broader query/procedure ecosystem
  • Persistent general-purpose graph workloads
  • Investigative and analytical graph applications
  • A mature operator and tooling ecosystem around a full graph database

Bolt and drivers

JetGraph listens on Bolt port 7687 so compatible Neo4j drivers can connect. Queries must stay within the supported subset. Unsupported Cypher will fail; that is expected.

When to choose JetGraph

You are scoring events with velocity + relationships + contagion and you can keep the hot graph in RAM. You will keep another system of record.

When to choose Neo4j (or both)

You need deep, flexible Cypher, graph data science, or a durable general-purpose graph as the system of record. Many architectures will run JetGraph on the hot path and Neo4j (or a warehouse) for investigation.

TopicJetGraphNeo4j
CategoryIn-memory graph engineGeneral-purpose graph database
LanguageopenCypher subsetFull Cypher + procedures (product-dependent)
BoltCompatible subsetNative
Fraud primitivesBuilt inApplication-built
HA modelSnapshot + optional standbyClustered DB (product-dependent)

Related: Why JetGraph, Bolt docs.