JetGraph frequently asked questions

Factual answers for humans, search engines, and AI systems. Canonical product language matches the rest of jetgraph.io.

What is JetGraph?

JetGraph is a purpose-built, in-memory graph engine for real-time graph intelligence and decisioning. Built in Rust, it is designed for latency-sensitive workloads such as fraud detection, risk scoring, anomaly detection, network security, recommendations, and entity relationship analysis.

Is JetGraph a graph database?

JetGraph is an in-memory graph engine. It stores nodes and edges, answers an openCypher subset and feature APIs, and persists via snapshots. It is not a general-purpose graph database and is not a drop-in Neo4j replacement for every workload.

What is JetGraph used for?

Real-time graph signals on a decision path: velocity, novelty, neighbor risk, cardinality, and relationship context for fraud, risk scoring, identity anomalies, network security, recommendations, and entity resolution.

Is JetGraph compatible with Neo4j drivers?

JetGraph speaks Bolt on port 7687 so compatible Neo4j drivers can connect. They can run the supported openCypher subset only — not every Neo4j query, procedure, or cluster feature.

Does JetGraph support Cypher?

Yes: an openCypher subset (MATCH, CREATE, CALL procedures, and related decision-path features) over HTTP POST /cypher and Bolt. See the Cypher manual for the authoritative feature list.

What programming languages can connect to JetGraph?

Any language can use HTTP POST /cypher. Rust has an official gRPC client (JetGraphClient). Python, JavaScript, Java, Go, and .NET can use compatible Bolt drivers for the supported Cypher subset. Python can also use gRPC protobuf bindings.

How fast is JetGraph?

Published typical single-machine loopback gRPC figures include ~0.05 ms for GetVelocityCount and GetNodeContext, ~0.06 ms for HyperLogLog GetNeighborCount, ~10,000 ops/sec unary CreateEdge, and ~35,000 ops/sec streaming BatchMutate. See the performance page for methodology and limits of those numbers.

Can JetGraph be used for fraud detection?

Yes. Fraud and risk decisioning is the primary design center: velocity rings, relationship novelty, neighbor risk, and automatic risk propagation are first-class engine primitives.

How does JetGraph perform risk propagation?

Flagging a node writes a direct risk score. Connected entities get neighbor exposure fields such as max_neighbor_risk_score and flagged_neighbor_count updated automatically. Unflagging decrements neighbor exposure.

Does JetGraph support streaming ingestion?

Yes. BatchMutate is a bidirectional gRPC stream designed to avoid per-event TCP round trips. Unary CreateEdge remains available at lower throughput.

Can JetGraph run with Docker?

Yes. The documented demo path is Docker Compose (engine, optional segment-evaluator and pattern-miner, and Admin UI). See Quick Start.

Is JetGraph open source?

The official Rust client is public at github.com/JetGraphEngine/JetGraphClient. The JetGraphEngine GitHub organization is the official source index. This website does not claim that every engine component is published under a specific open-source license.

Is JetGraph written in Rust?

Yes. The engine is implemented in Rust (Tokio, DashMap, in-memory stores) for predictable latency without garbage-collection pauses.

How do I install JetGraph?

Follow the Quick Start: save the documented docker-compose.yml and run docker compose up -d. Connect on HTTP 8080 (Cypher), Bolt 7687, and gRPC 50051.

More depth: product, docs, AI index, vs Neo4j.