Neo4j is the world's leading graph database. Instead of storing data in rows and tables, it stores it as nodes (things) and relationships (the connections between them). A lot of legal data already looks like this: cases cite cases, statutes amend statutes, clauses reference definitions, entities own entities. A graph lets you follow those connections directly instead of rebuilding them from table joins every time.
<aside> <img src="notion://custom_emoji/9067698e-8c48-4c31-9814-0f22ae0539d7/378c01c2-41b2-802a-941e-007a5d32fd71" alt="notion://custom_emoji/9067698e-8c48-4c31-9814-0f22ae0539d7/378c01c2-41b2-802a-941e-007a5d32fd71" width="40px" />
</aside>
Neo4j is the connective layer for projects where the relationships between documents matter as much as the documents themselves. Some ideas on where it fits:
Semantic web data, kept as a graph. Neo4j works natively with the RDF/SPARQL data the Cellar API returns, so you can query EU legal data without flattening it into tables first.
Useful in combination with: Cellar API.
Citation and case law networks. Map how cases reference, distinguish, and overrule each other, then walk that structure with queries to answer questions a keyword search cannot.
Useful in combination with: Claude Code, Perplexity (live search).
GraphRAG. Pair graph traversal with LLM retrieval so answers are grounded in explicit relationships, not just text similarity, with the option of returning the retrieval path alongside the answer.
Useful in combination with: NVIDIA / Nemotron (reasoning models), GCP (hosting), Lovable or Momen (frontend).
Credits are applied to a registered Neo4j Aura account; the email associated with that account is needed in order to credit it. Credits cannot be shared.
Neo4j GraphAcademy: https://graphacademy.neo4j.com/