// GRAPHS
Your data is connected.
Load it as a graph.
Run it in production.
Apache Hop reads from and writes to the graph databases you already use — Neo4j, Memgraph, FalkorDB, Apache AGE, Amazon Neptune and Gremlin servers — with one connection type and one set of transforms. Putki runs those pipelines in production, and uses graphs itself to show how your runs and projects connect.
● Cypher and Gremlin, one set of transforms ● Vector search and GraphRAG with standard Hop transforms ● Open source graph plugins, Apache-licensed
// APACHE HOP
One way to work with every graph
Apache Hop has graph support for any database that speaks Cypher or Gremlin. Build the pipeline once; switching databases means switching the connection.
Graph database connections
A single graph database connection, with a type per database: Bolt for Neo4j, Memgraph and Amazon Neptune, the Redis protocol for FalkorDB, PostgreSQL for Apache AGE, and Gremlin for Gremlin Server, JanusGraph and Neptune.
- The same transforms and actions work against every database
- Each connection knows what its database supports, so you see what works where
Model, write and query
Map rows to nodes and relationships once in a graph model, then load, query and maintain the graph from pipelines and workflows.
- Graph output writes rows through a graph model; Graph query runs Cypher statements or Gremlin traversals with parameters from your rows
- Get graph schema, Graph index, Graph constraint and Graph script manage the database in its own syntax
- Bulk loads with Generate graph CSVs and Neo4j import
Vectors in the graph
Embeddings travel through a pipeline as Vector fields and are stored on nodes and relationships, next to the data they describe.
- Create vector indexes on Neo4j, Memgraph and FalkorDB with Graph index
- Graph vector search finds the nearest nodes or relationships, with the same similarity score on every database
GraphRAG from standard transforms
Answer questions from a knowledge graph: chunk and embed your documents, let a language model extract entities and relationships, and give it the connected facts when it answers.
- Text chunker, Embed text, Extract graph and Language model chat — no custom code
- Samples for Neo4j, Memgraph and FalkorDB, and for Apache AGE with pgvector: a fully Apache-licensed stack
// DATABASES
Bring the graph database you have
The graph plugins are optional Apache Hop plugins: install them from the marketplace, or use the full client.
| DATABASE | LANGUAGE | CONNECTION |
|---|---|---|
| Neo4j | Cypher | Bolt |
| Memgraph | Cypher | Bolt |
| Amazon Neptune | openCypher, Gremlin | Bolt or Gremlin |
| FalkorDB | Cypher | Redis protocol |
| Apache AGE | Cypher | PostgreSQL |
| Gremlin Server, JanusGraph | Gremlin | Gremlin |
// PUTKI
Graphs in production, with Putki
Putki runs your graph pipelines like every other Apache Hop pipeline: scanned images, monitoring and support. And it puts its own operational data in a graph, because the questions you ask about runs and projects are questions about connections.
Impact analysis
The graph of your projects, read from their files into Apache AGE: what reads and writes a table, what a change to it can reach, and which references are broken. Nothing has to run.
Professional ● Enterprise
Running graph pipelines
Graph loads are pipelines like any other. Putki adds what production needs around them: hardened, scanned images, execution monitoring with Grafana dashboards, alerts and support.
All plans
Execution information
Every run, its transforms and actions, their row counts and logs, and which run started which — stored as a knowledge graph in Neo4j that you can access and query.
Professional ● Enterprise
// LEARN MORE
Do it yourself
The Apache Hop manual, the Putki docs and our blog cover graph databases from the first load to a knowledge graph.
Connected data, still in tables?
Customers, products, suppliers, assets: the relationships are already in your relational data. We help you design the graph model, load it with Apache Hop, and keep it in sync — on the graph database you choose.
The Apache Hop manual, the Putki docs and our blog cover graph databases from the first load to a knowledge graph.
// COMMON QUESTIONS
Things people ask
Databases that speak Cypher or Gremlin: Neo4j, Memgraph, Amazon Neptune, FalkorDB, Apache AGE, Gremlin Server and JanusGraph. The graph databases page of the manual lists which transforms and actions work with which database.
Not with Neo4j, Memgraph or FalkorDB: the embeddings live on the nodes, and Graph vector search searches their vector index. Apache AGE has no vector index, so the GraphRAG samples keep the embeddings in pgvector next to the graph, in the same PostgreSQL.
Yes. The graph plugins are part of Apache Hop, under the Apache License. Putki adds the production layer around them — images, monitoring, execution information, impact analysis and support. Note that the FalkorDB server itself is licensed under the SSPL; Hop ships only its client.
Nothing has to: existing Neo4j connections keep working as they are, and Cypher output and Cypher builder are still there. Hop GUI can convert them to graph database connections when you are ready. What you gain is the choice: the same pipelines can target Memgraph, FalkorDB, Apache AGE or a Gremlin server, plus vector indexes, vector search and the GraphRAG transforms.
Put your connected data to work
Tell us which graph database you use, or want to use, and what you need to answer from it. We'll help you design the model and the pipelines that keep it current.