Tip: select any text on this page to bring up the Ask tooltip and question the Synedrix OS assistant about it.
What Neo4j is
Neo4j's Operations Manual opens by describing an architecture designed for optimal management, storage, and traversal of nodes and relationships, taking a property graph approach beneficial for both traversal performance and operations runtime.
Two structural rules matter for a lineage model: a node can have zero to many labels, and a relationship must always have a start node, an end node, exactly one type, and a direction. A lineage edge therefore cannot exist without both endpoints and cannot be ambiguous about which way causality runs.
Cypher queries are written as ASCII-art patterns. The official SQL comparison makes the difference concrete: a two-JOIN SQL query becomes a single MATCH pattern drawn through the graph.
What it is normally used for
How Neo4j is used generally, before Synedrix OS enters the picture.
- Modelling networks: social, road, supply chain, or any interconnected structure
- Answering relationship questions rather than entity questions
- Fraud and anomaly detection through pattern matching over connections
- Recommendation engines driven by traversal rather than aggregation
- Impact and dependency analysis across systems
What it owns, and what it does not
Every platform in the stack has a non-overlapping responsibility. The boundary is as much a part of the design as the capability.
Owns
- Workflow and outcome lineage
- Agent influence and correlation networks
- Entity relationship graphs
- Cross-system dependency context
- Entity and reasoning memory: the memory fabric's graph tier
Does not own
- Canonical relational truth and audit records
- → Supabase PG17 (Layer 7)
- Time-series metrics and telemetry
- → QuestDB (Layer 7)
- Hot state, pub/sub, deduplication
- → Redis (Layer 7)
- General-purpose semantic and vector retrieval (NAMS's own index is scoped to entity resolution only)
- → Qdrant (Layer 7)
- LLM traces, prompts, cost attribution
- → Langfuse (Layer 10)
- Workflow sequencing and retries
- → n8n (Layer 3)
Why Neo4j
Tracing what caused what is what graphs are built for
Answering "what happened, why, and under whose authority" means following a chain of cause and effect that can be any number of steps long. A relational database would need increasingly complex queries to follow that chain; a graph database follows it natively, the way it's designed to.
Checking whether a path exists is a simple query
Every agent must have an escalation path that eventually reaches a specific oversight agent. Confirming that such a path exists is a reachability check: something Neo4j answers directly, and nothing else in the stack does natively.
Cause and effect can't be recorded ambiguously
Every connection in Neo4j must have a clear starting point, ending point, type, and direction. It's structurally impossible to record a cause-and-effect relationship that doesn't specify which way it runs.
New kinds of relationships don't require a migration
The full set of things that can be connected in the lineage graph is still growing. Because Neo4j doesn't require a fixed schema, new categories of nodes and relationships can be added as they come up, without a database migration.
Reasoning gets a typed home, not just a lineage record
NAMS(an open-source library, not a new platform) layers a typed entity model over this same Neo4j and records agent reasoning as nodes joined to the entities they touched by explicit edges. That's the missing link between an outcome and the business state it moved, and it runs on infrastructure the stack already operates.
Licence position
Neo4j questions, answered
Short, citable answers drawn from this page.
What role does Neo4j play in Synedrix OS?
Neo4j (Lineage graph) is mapped to Layer 7: Memory & Data in Synedrix OS. Lineage is a traversal problem. Nothing else in the stack answers reachability natively.
What does Neo4j own in Synedrix OS?
Neo4j owns: Workflow and outcome lineage; Agent influence and correlation networks; Entity relationship graphs; Cross-system dependency context; Entity and reasoning memory: the memory fabric's graph tier.
What does Neo4j not own in Synedrix OS?
Neo4j does not own: Canonical relational truth and audit records (owned by Supabase PG17 (Layer 7)); Time-series metrics and telemetry (owned by QuestDB (Layer 7)); Hot state, pub/sub, deduplication (owned by Redis (Layer 7)); General-purpose semantic and vector retrieval (NAMS's own index is scoped to entity resolution only) (owned by Qdrant (Layer 7)); LLM traces, prompts, cost attribution (owned by Langfuse (Layer 10)); Workflow sequencing and retries (owned by n8n (Layer 3)).
What licence is Neo4j under?
Neo4j is released under Community Edition GPLv3; Enterprise Edition closed-source. Community Edition is open source under GPLv3, but Enterprise Edition is not, and the Enterprise-only list is governance and observability features rather than only scale features. Community Edition has no role-based access control, so every authenticated user has full read and write on the graph: the strictly read-only constraint on the Audit and Lineage Agent has no database-level enforcement there. Community Edition also exports no metrics by any documented path, and backup is offline dump only. Vector indexes, which the memory fabric's graph-tier library (NAMS) requires at 5.20+, are confirmed available in Community Edition, verified against Neo4j's own edition feature table. Community Edition's one-database limit means tenant separation on that tier is enforced above the database, not by it.
The full reference page, with the operating model, deployment detail, and links to every canonical specification this platform touches, lives in the documentation repository, which is private while the platform is being built.