Fluree Launches Verifiable Knowledge Graph Database for Agentic AI

FlureeDB acts as a secure context layer fit for autonomous systems: pull from many data sources wherever they live, answer structured queries fast and efficiently, carry citations and lineage on every result, and enforce permissions and security at the data itself. 

• FlureeDB runs as a single binary that stores data as a W3C-standard RDF knowledge graph, with every commit immutably recorded so any prior state can be reconstructed on demand. It collapses what is typically a stack of five or six services — graph store, full-text and vector search, access policy, cryptographic transaction signing, OWL/RDFS reasoning, and a bundled Model Context Protocol (MCP) server — into one intelligence layer.
  
  
• The overall Fluree platform is already in production across some of the most demanding environments in the world, including the U.S. Department of Defense, Morgan Stanley, The Associated Press, Dow Jones, Warner Bros. Discovery, Dotdash Meredith, WebMD Ignite, CBC, Canva and Arizona State University.
  

WINSTON-SALEM, N.C., June 23, 2026 (GLOBE NEWSWIRE) -- Fluree, PBC, the public benefit corporation behind the modern knowledge graph movement, today announced the general availability of FlureeDB, a semantic graph database engineered for an era in which every answer — human or machine — must be fast, defensible, and traceable to its source.

"Most databases store records. FlureeDB stores knowledge — and the proof of where that knowledge came from," said Brian Platz, Co-CEO and Co-Founder of Fluree. "Enterprises are wiring AI agents into systems of record at unprecedented speed. The question regulators, auditors, and boards will ask next is the same one they've always asked: show me the lineage. You can't defend a decision you can't reconstruct. FlureeDB makes reconstruction free."

A category built for the agentic era

Gartner projects that at least 50% of generative AI projects were abandoned, citing poor data quality, inadequate risk controls, and unclear provenance among the leading causes. The failure mode is rarely the model — it's the context layer feeding it: data that can't be unified, answers that can't be cited, and decisions that can't be audited.

A context layer fit for autonomous systems has to do all of it at once: pull from many data sources wherever they live, answer structured queries fast and efficiently, carry citations and lineage on every result, and enforce permissions and security at the data itself. FlureeDB was built to be exactly that layer.

1. Blazing fast — and proven on a public benchmark

In the SPARQLoscope DBLP evaluation — an open benchmark of 105 real-world SPARQL queries run against the DBLP bibliographic dataset (~561 million triples) — FlureeDB ranked first overall.

In the WGPB Benchmark, loading the full Wikidata dataset of 21.5 billion triples, FlureeDB posted a 43ms geometric mean across 850 queries — the fastest published results on record.

2. Verifiable, secure, and governed by default

FlureeDB's governance model is not a layer above the database — it is the database. As organizations scale AI access across more users, more agents, and more data sources, the surface area for accidental exposure grows with it. FlureeDB shrinks that surface by making security and lineage structural properties, rather than something applications are trusted to enforce.

• Cryptographic provenance. Every transaction can be signed with JWS or as a W3C Verifiable Credential. Commits are content-addressed in a tamper-evident chain, so any modification invalidates the signature — and every answer traces to its source.
  
• Policy lives with the data. Triple-level access control — attribute-based, role-based, and relationship-based — is enforced inside the query engine, not in application code or a sidecar gateway. Policies themselves are RDF: versioned in the ledger, and queryable.
  
• Time travel as a primitive. Query the graph as it existed at any past transaction, timestamp, or commit hash — no snapshots, no separate audit log, no replay infrastructure. Any decision can be reconstructed on demand.
  
• Git-like branch, rebase, and merge. Fork a dataset to test a schema change, a migration, or an agent's proposed edits in complete isolation, then merge when validated — each branch keeping its own independent commit history.

3. Built for agents and humans

FlureeDB ships with a Model Context Protocol server in the same binary, exposing semantic recall, graph query, and persistent memory tools that AI assistants like Claude Code and Cursor can invoke directly. Companion product Fluree Memory gives coding agents a queryable, versioned long-term memory — built on the same engine, governed by the same policies.

Availability and licensing

FlureeDB meets teams wherever they want to run it: fully managed and serverless via flur.ee/solo, self-hosted on your own machines, or in Fluree's hosted cloud.

The software is available today under the Business Source License 1.1, converting to Apache 2.0 three years after each version's first public distribution.

About Fluree

Fluree, PBC is a Public Benefit Corporation building the verifiable data infrastructure for the AI era. Founded by Brian Platz and Flip Filipowski, Fluree's open-source knowledge graph platform — including Fluree Core (the database), Fluree AI (GraphRAG copilots), Fluree Sense (data pipeline), Fluree CAM (content intelligence), and Fluree ITM (taxonomy and ontology management) — powers mission-critical systems across defense, finance, media, and healthcare. Fluree is a Gartner Cool Vendor in Data and AI Management.

The company operates from Winston-Salem, NC, with offices in New York, Colorado, Paris, and India. Learn more at flur.ee and the Fluree blog.

Developer Resources

• Source code: github.com/fluree/db
• Full documentation: labs.flur.ee/docs/db
• Getting started in 60 seconds: labs.flur.ee/downloads
• Quickstart tutorial: labs.flur.ee/docs/db/getting-started
• SPARQLoscope DBLP benchmark (methodology + full results): labs.flur.ee/blog/sparqloscope-dblp-benchmark-results
• Fluree Memory for AI coding assistants: labs.flur.ee/guides/fluree-memory
• Migration guides (Jena, GraphDB, Stardog, Neo4j, SQL, data lakes): labs.flur.ee/guides
• Storage backend reference (S3, DynamoDB, IPFS, filesystem): labs.flur.ee/docs/db/operations/storage

Media Contact: Kevin Doubleday, Director of Marketing, Fluree · kdoubleday@flur.ee

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.