Fluree DB - JSON-LD Graph Database Built for AI Agents

TL;DR

TL;DR: Fluree DB is an open-source graph database that runs SPARQL, JSON-LD, and openCypher queries on the same store — with built-in vector search, time travel, git-like branching, and a dedicated AI agent memory layer. One Docker command gets you started.

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Project page: flur.ee ← verified
• Source repository: github.com/fluree/db ← verified README read
• License: BUSL-1.1 (Business Source License) ← verified via LICENSE file
• HN launch thread: news.ycombinator.com/item?id=38479017 ← verified
• Documentation: docs.flur.ee ← verified
• License is BUSL-1.1 — source is open but commercial use requires a Fluree license after a grace period

What Is Fluree DB?

Fluree DB is a graph database built around W3C standards: RDF 1.1/1.2 with native SPARQL, JSON-LD, and openCypher query support — all running on the same underlying store. The README describes it as “a graph database for data that matters” with a focus on temporal, verifiable, standards-compliant data and git-like branching and merging.

Key claims from the official README:

• Over 2M facts per second bulk import
• 10.4x faster than the next engine on SPARQL benchmark (DBLP, 105 queries)
• Built-in HNSW vector search, full-text search, and geo search — no external service required
• Fine-grained access control with no external dependencies
• Fluree Memory: persistent, searchable memory layer for AI coding assistants (Claude Code, Cursor, etc.)

# Docker — fastest way to try it
docker run -p 8090:8090 fluree/server:latest

Setup Workflow

Step 1: Install the CLI

Homebrew (macOS / Linux):

brew install fluree/tap/fluree

Shell installer:

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/fluree/db/releases/latest/download/fluree-db-cli-installer.sh | sh

Windows (PowerShell):

irm https://github.com/fluree/db/releases/latest/download/fluree-db-cli-installer.ps1 | iex

Pre-built binaries for all platforms are on the GitHub Releases page.

Step 2: Zero to Graph in 60 Seconds

fluree init
fluree create movies

fluree insert '
@prefix schema: <http://schema.org/> .
@prefix ex:     <http://example.org/> .

ex:blade-runner  a schema:Movie ;
  schema:name        "Blade Runner" ;
  schema:dateCreated "1982-06-25"^^<http://www.w3.org/2001/XMLSchema#date> ;
  schema:director    ex:ridley-scott .

ex:ridley-scott  a schema:Person ;
  schema:name "Ridley Scott" .
'

# Query with SPARQL
fluree query --format table 'SELECT ?title ?date WHERE {
  ?movie a <http://schema.org/Movie> ;
         <http://schema.org/name> ?title ;
         <http://schema.org/dateCreated> ?date .
} ORDER BY ?date'

Step 3: Query in JSON-LD

The same data can be queried in JSON-LD syntax:

fluree query --jsonld '{
  "@context": { "schema": "http://schema.org/" },
  "select": ["?title", "?date"],
  "where": [
    { "@id": "?movie", "@type": "schema:Movie",
      "schema:name": "?title", "schema:dateCreated": "?date" }
  ],
  "orderBy": "?date"
}'

Step 4: Query with openCypher

fluree context set movies -e '{"@vocab": "http://schema.org/"}'

fluree query --cypher -e 'MATCH (m:Movie)
  RETURN m.name AS title, m.dateCreated AS date
  ORDER BY date'

All three compile to the same engine — same features, same performance.

Step 5: Time Travel

Query the state of the data at any point in the past:

# What did the data look like before that update?
fluree query --at 1 'SELECT ?title ?genre WHERE {
  ?movie a <http://schema.org/Movie> ;
         <http://schema.org/genre> ?genre .
}'

Step 6: Fluree Memory (AI Agent Memory)

Fluree Memory is a persistent memory layer for AI coding assistants built into the CLI:

# Give an AI tool long-term project memory
fluree memory init --repo ./my-project

Facts, decisions, and preferences persist across sessions in a Fluree ledger you control — scoped per-repo or per-user, shareable via git. Docs at labs.flur.ee/docs.

Deeper Analysis

Why Three Query Languages?

Fluree’s architecture compiles SPARQL, JSON-LD, and openCypher to a common intermediate representation before executing against the same graph store. This means you can use whichever query syntax fits your team’s skills or use case:

• SPARQL — best for RDF veterans and semantic web standards work
• JSON-LD — best for JSON-native developers and API integration
• openCypher — best for Neo4j migrants or property-graph familiarity

Vector Search Without External Services

Unlike most graph databases that require a separate vector database (Pinecone, Weaviate, etc.) for AI embeddings, Fluree includes HNSW-based vector search built directly into the database. The README confirms: “Integrated vector, text and geo search, and fine-grained access control with no external dependencies.”

Benchmark Claims

The Fluree Labs benchmark page claims Fluree is 10.4x faster than the next engine on the SPARQLoscope DBLP evaluation (105 SPARQL queries, all 850 WGPB graph-pattern queries on a 21.5-billion-triple Wikidata dump at 43ms geometric mean). The key claim is that Fluree completed all 105 queries while some other engines failed or timed out.

Verify the benchmarks at labs.flur.ee.

License Consideration

Fluree uses the Business Source License 1.1 (BUSL-1.1). This is not a permissive open-source license like MIT or Apache 2. The source is visible and the project is open, but commercial deployment may require a commercial license from Fluree PBC after the grace period. For personal or internal use, the license is fine — but factor this into your evaluation for commercial products.

Practical Evaluation Checklist

• Can run locally in Docker with no external dependencies
• Homebrew, shell installer, and Windows PowerShell support
• SPARQL, JSON-LD, and openCypher all work on the same data
• Built-in vector search (HNSW) — no separate vector DB needed
• Time travel queries allow querying historical data states
• Git-like branching and merging for data
• Fluree Memory provides AI agent session persistence
• Benchmarks claim top performance but verify against your own workload
• BUSL-1.1 license — review commercial use terms before product deployment

Security Notes

• Fine-grained access control built into the database engine
• No external service dependencies reduces attack surface
• Immutable ledger provides audit trail of all data changes
• For production deployments, follow the docs at docs.flur.ee for security hardening

FAQ

Q: Does Fluree require a JVM? A: No. The fluree binary bundles everything needed. The Docker image also ships a pre-configured HTTP server ready to accept queries on port 8090.

Q: Can I query existing Neo4j or RDF data in Fluree? A: Yes. Fluree supports full SPARQL 1.1 and openCypher 9. If you have existing RDF data (Turtle/N-Triples), you can load it directly. For Neo4j Cypher queries, they run largely unchanged in Fluree.

Q: What is the difference between Fluree and a regular graph database like Neo4j? A: Key differences: Fluree is temporal by default (every change is recorded and queryable by time), uses W3C standards (RDF, SPARQL, JSON-LD) rather than a proprietary property graph model, includes built-in vector search without external services, and has a dedicated AI agent memory layer (Fluree Memory).

Q: Is Fluree free to use? A: The source is open (BUSL-1.1 license). For personal, educational, and internal use, it functions as open-source. Commercial products may need a commercial license. There is also a free cloud tier with usage limits at flur.ee/solo.

Conclusion

Fluree DB stands out by converging three things that are usually separate: a standards-compliant RDF graph store, a property-graph interface (via openCypher), and an AI-ready vector search layer — all in one zero-dependency binary. The time travel and git-like branching features fill real gaps for teams that need auditable data history and parallel data branches.

The Docker one-liner makes local evaluation trivial, and the Homebrew install handles day-to-day CLI use on macOS and Linux. The main caveats are the BUSL-1.1 license (review before commercial use) and the fact that some enterprise features may require Fluree’s commercial licensing beyond the open-source tier.

For AI agent builders specifically, Fluree Memory is worth a closer look — it brings structured, queryable, long-term memory to coding assistants without needing a separate vector DB or Notion-style document store.

Getting started:

docker run -p 8090:8090 fluree/server:latest

Docs at docs.flur.ee | GitHub at github.com/fluree/db