Falkordb
Founded 2023 Β· 20 employees on LinkedIn Β· 5 known investors
FalkorDB is an open-source, Redis-based property graph database that uses sparse matrices and linear algebra for low-latency graph queries. It targets developers and enterprises building GraphRAG, agentic AI, chatbots, fraud detection, and security graph applications, with multi-tenant SaaS and on-prem support.
Also known as FalkorDB
Investors Β· 5
Also in the syndicate Β· 4
Company profile
researched Aug 2026FalkorDB develops a property graph database aimed at generative AI and machine learning workloads, particularly GraphRAG (retrieval-augmented generation grounded in knowledge graphs) and agent memory. The engine follows the openCypher property graph model with proprietary query extensions, stores graphs as sparse adjacency matrices, and uses linear algebra for query execution. It supports full-text, vector and range indexes in one engine, and connections over the RESP and Bolt protocols. It runs as a Redis module and can be launched via Docker (ports 6379 for the server and 3000 for the FalkorDB Browser web UI) or as a managed cloud instance.
The company markets the database for five principal use cases: GraphRAG (combining LLMs with domain-specific knowledge graphs, with ontology auto-detection and built-in agent orchestration), agentic AI (graph traversal combined with vector search for personalized, explainable recommendations with persistent conversation history), chatbots (entity extraction, fact linking and relationship mapping for context-aware conversation), fraud detection (analyzing relationships between IPs, devices, transactions and accounts to surface fraud rings), and security graphs (schemaless storage and near real-time querying of findings, vulnerabilities and assets for cyber and cloud security vendors deploying multi-tenant SaaS or on-prem).
Supporting components include the open-source FalkorDB Browser visualization UI (graph creation and querying, inline node property editing, query history, search across schema, nodes and metadata, node style customization per label, CSV upload with local or S3-compatible storage backends, and Helm-based Kubernetes deployment), official client libraries in Java, Python, Node.js, Rust, Go and C#, and documented migration paths from Neo4j.
Founding story
FalkorDB was founded in 2023 by three alumni of Redis. Roi Lipman, co-founder and CTO, originated the idea for the database more than eight years earlier; Avi Avni, co-founder and chief architect, is a database specialist who had contributed to the development of the C# and F# programming languages and had built a similar database independently before joining; Dr. Guy Kurland, co-founder and CEO, had previously worked on high-speed databases and led development teams in the field. The founders have said the GraphRAG direction came from observing that enterprises struggled to deploy LLM applications reliably β citing a September 2023 Microsoft report in which a hybrid plus semantic ranker approach reached roughly 75% accuracy β combined with 2023 academic work pointing to knowledge graphs as an improved retrieval substrate.
Business model
FalkorDB distributes an open-source graph database engine (licensed under the Server Side Public License v1) alongside a commercial managed offering, FalkorDB Cloud, which can be started for free and runs on GCP, AWS and Azure. The commercial tier is positioned around enterprise features including multi-graph/multi-tenancy, graph access control, TLS, VPC, cluster deployment, high availability, multi-zone deployment, continuous persistence, automated backups every 12 hours, and 24/7 support. Self-hosted and on-prem deployment via Docker images is also supported.
Revenue derives from the managed FalkorDB Cloud service and enterprise deployments with support and advanced operational features, layered on a free open-source core and a free cloud tier.
Traction
The main FalkorDB GitHub repository shows about 5,700 stars and 436 forks, with the browser UI repository at 129 stars. The company reported seven employees and $3 million raised in mid-2024, and is hiring across R&D, developer relations and sales in Tel Aviv and the Bay Area. It maintains official client libraries across six languages and integrations with Graphiti, Cognee, LangChain, LlamaIndex and g.v().
Latest developments
Company materials reference release 4.14.10 of the database, a June 2025 refresh of the FalkorDB Browser UI, and continued content and community activity, including sponsorship of the August 2026 'Memory Meets Motion' hackathon in San Francisco where FalkorDB served as the memory layer. The main engine repository is described as a Rust implementation of the FalkorDB Redis module built on GraphBLAS.
βΈFull profile β market position, technology, go-to-market, geography, history
Market position
FalkorDB positions itself in the graph database market β described by the company as valued globally at over $3 billion in 2023 and growing at a 21.9% CAGR through 2030 β while targeting the larger generative AI market, which it cited as $200 billion in 2023 and projected to exceed $1 trillion by 2030. It competes directly with established graph databases including Neo4j, AWS Neptune, TigerGraph and ArangoDB.
The company describes FalkorDB as the first queryable property graph database to use sparse matrices for the adjacency matrix representation and linear algebra for querying, which it links to compact storage and low-latency traversal. Marketing materials position it against Neo4j, AWS Neptune, TigerGraph and ArangoDB on multi-tenancy, latency, horizontal scaling, vector search, Cypher support and memory efficiency, citing company benchmarks of P50/P95/P99 latencies of 36ms/74ms/83ms versus 469ms/13,969ms/41,157ms for unnamed competition and 100MB versus 600MB memory usage. Multi-tenancy at the level of 10,000+ graphs per deployment and Redis-based architecture are also cited as distinguishing traits.
Technology
The engine implements the openCypher property graph model with proprietary extensions, stores graphs as sparse adjacency matrices using GraphBLAS, and executes queries via linear algebra. It provides full-text, vector and range indexing, RESP and Bolt protocol support, and a built-in GraphRAG capability that turns unstructured sources into a knowledge graph. It ships as a Redis module (the main repository is a Rust implementation producing libfalkordb) with a Docker image bundling the server on port 6379 and a web browser UI on port 3000. Enterprise-oriented capabilities include multi-graph/multi-tenancy, graph access control, TLS, VPC, clustering, high availability, multi-zone deployment, continuous persistence and 12-hourly automated backups.
Go-to-market
Go-to-market combines open-source distribution and developer community building β GitHub repositories, Discord, documentation, benchmarks and migration guides from Neo4j β with a free-tier cloud instance and a direct enterprise sales motion. Open roles include an account executive and a technical sales development representative targeting infrastructure, cybersecurity, AI and data platform accounts, plus a developer relations advocate. Integrations with AI frameworks such as Graphiti (by Zep), Cognee, LangChain, LlamaIndex and the g.v() graph client serve as additional distribution channels.
Developers and enterprise teams building GenAI and agentic applications on complex, interconnected data, including GraphRAG systems, chatbots and agent memory layers, plus fraud-detection teams and cyber and cloud security vendors that need multi-tenant SaaS or on-prem graph infrastructure.
Geography
The company lists R&D and other roles in Tel Aviv, Israel (hybrid) and commercial and developer-relations roles in the San Francisco Bay Area, including remote/hybrid arrangements. The cloud service is offered on GCP, AWS and Azure.
History
The company was founded in 2023 by three alumni of Redis. Co-founder and CTO Roi Lipman had conceived of the database concept more than eight years before the 2024 account. A $3 million seed round dated January 2023 was reported, with the company employing seven people as of mid-2024. The product has since expanded to include FalkorDB Cloud, a browser-based visualization UI (updated in June 2025), a GraphRAG SDK, and clients for Python, JavaScript/Node.js, Java, Rust, Go and C#. The engine repository describes a Rust implementation building a FalkorDB Redis module using GraphBLAS sparse matrices.
Compiled by commissioned research from 8 cited public sources β announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed β not independently audited.
Timeline Β· 3
launches, deals, and filingsFalkorDB was a sponsor technology at the eight-hour 'Memory Meets Motion' hackathon held at Frontier Tower in San Francisco, where projects were required to use FalkorDB as the memory layer.
An update to FalkorDB Browser, the visualization UI for FalkorDB, added faster graph creation and querying, inline node property editing, a query history panel, improved search, expanded settings and user controls, and per-label node style customization.
FalkorDB reported a $3 million seed round dated January 2023, bringing total investment to date to $3 million.
$3M source β
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
βΈResearch sources Β· 8
primary sources listed
- Falkordbfalkordb.com Β· web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Falkordb do?
- FalkorDB is an open-source, Redis-based property graph database using sparse matrices and linear algebra for GraphRAG and AI workloads.
- Who are Falkordb's investors?
- Falkordb's investors include Angular Ventures.