GRAPH DATABASE
PostgreSQL with graph
and vector superpowers
Every Graph instance runs PostgreSQL with Apache AGE for Cypher queries and pgvector for embeddings. Schema validation ensures data integrity.
PostgreSQL
The world's most advanced open-source relational database. ACID compliant, battle-tested, with decades of reliability. Your data is stored in standard Postgres tables.
Apache AGE
Graph extension for PostgreSQL. Execute Cypher queries directly on your data without leaving Postgres. Traverse relationships with native graph performance.
pgvector
Vector similarity search for embeddings. Store vectors as entity properties and query with cosine or L2 distance. True hybrid graph + vector queries.
Schema-enforced entities
Define DTDL models that specify exactly what properties and relationships are allowed. Every write is validated against your schema before it touches the database.
{
"@id": "dtmi:code:Module;1",
"@type": "Interface",
"displayName": { "en": "Code Module" },
"contents": [
{
"@type": "Property",
"name": "path",
"schema": "string",
"description": { "en": "File path relative to repo root" }
},
{
"@type": "Property",
"name": "embedding",
"schema": { "@type": "Array", "elementSchema": "double" }
},
{
"@type": "Relationship",
"name": "imports",
"target": "dtmi:code:Module;1",
"description": { "en": "Module import dependency" }
}
],
"@context": "dtmi:dtdl:context;4"
}// Agent attempts to write invalid data
await session.call_tool("create_or_replace_digital_twin", {
"twin_id": "module-auth",
"model_id": "dtmi:code:Module;1",
"properties": {
"path": "src/auth.ts",
"temprature": 72.5 // Typo + wrong property
}
});
// Error: Property 'temprature' is not defined in model
// 'dtmi:code:Module;1'. Did you mean 'path' or 'embedding'?
// Transaction rolled back.Prevent bad data at write time
When an agent tries to write invalid data, the transaction is rejected with a clear error message. No corrupted state, no contradictory information.
Why this matters: Vector databases accept any JSON. Over time, your knowledge base fills with inconsistent data that confuses agents. Schema validation prevents this from happening.
Powerful query capabilities
Combine graph traversal with vector similarity in a single query.
Hybrid graph + vector
Traverse relationships with Cypher, then rank results by embedding similarity. Structure AND semantics in one query.
MATCH (m:Twin)-[:imports]->(dep:Twin)
WHERE dep.name = 'auth-service'
AND m.`$metadata`.`$model` = 'dtmi:code:Module;1'
RETURN m.path, m.description
ORDER BY l2_distance(m.embedding, $query_vector)
LIMIT 10
-- Find modules that import auth-service
-- Ranked by semantic similarity to querySpatial queries with PostGIS
Full PostGIS support for geospatial data. Find entities near a location, within a polygon, or along a path.
MATCH (sensor:Twin)-[:locatedIn]->(zone:Twin)
WHERE sensor.`$metadata`.`$model` = 'dtmi:facility:Sensor;1'
AND ST_DWithin(
sensor.location::geometry,
ST_MakePoint(-122.4194, 37.7749)::geometry,
1000 -- meters
)
RETURN sensor.`$dtId`, sensor.reading, zone.nameHow it compares
Different databases for different problems.
| Capability | Konnektr Graph | Vector DBs | SQL Databases |
|---|---|---|---|
| Schema validation | DTDL models | None | Rigid tables |
| Relationship traversal | Native Cypher | Not supported | Slow JOINs |
| Vector similarity | pgvector | Primary feature | Extension only |
| Hybrid queries | Graph + Vector | Vector only | Keyword only |
| ACID transactions | PostgreSQL | Eventually consistent | Full support |
AGENT INTEGRATION
Every graph includes an MCP server
Connect Claude, Cursor, or custom agents directly to your graph. 16 tools for creating, querying, and managing entities.
Simple, predictable pricing
Start free with 500 entities. Pay for capacity as you grow. Upgrade to Standard ($49/mo) or Enterprise when you need more.