FalkorDB, a Startup by Redis Veterans, Raises $3 Million to Enhance Language Models

TEL AVIV, Israel - June 20, 2024 - PRLog -- FalkorDB, a startup developing technology to enhance language models, has completed a $3 million seed funding round. The round was led by Angular Ventures, with participation from K5 Tokyo Black, and private investors including Aryeh Mergi, co-founder of M-Systems, XtreamIO, Jerry Dischler, President of Cloud Applications at Google and Eldad Farkash and Saar Bitner, founders of Firebolt.
Currently participating in the ninth Cohort of the Intel Ignite accelerator program for deep tech startups, FalkorDB provides GraphRAG technology to enhance the performance of large language models by integrating external knowledge sources in the form of a Knowledge Graph. This allows the model to retrieve relevant information before generating a response, improving its ability to provide accurate and context-aware answers. The combination of large language models with networked information bridges the gap between the generative language capabilities of the models and the structured, detailed data stored in knowledge bases.
To streamline the process, FalkorDB is developing a solution that fully automates the conversion of organizational information into a Knowledge Graph, a process that can otherwise be complex and challenging. FalkorDB's integration as GraphRAG offers an innovative solution that leverages the advantages of both worlds: efficiently managing internal organizational information through graphs and incorporating relevant external knowledge to enhance decision-making and interactions with large language models. FalkorDB's technological uniqueness compared to the rest of the market lies in its extremely low latency Graph Database, which uses a unique approach based on representing information in sparse matrices and querying the data by computing algebraic expressions.
Founded in 2023 by Dr. Guy Korland (CEO), Roy Lipman (CTO), and Avi Avni (Chief Architect), all former Redis employees, the raised capital will be used to accelerate product development, hire personnel, and initiate marketing and sales.
FalkorDB targets the new market of Large Language Models (LLMs) based applications, one of the fastest-growing sectors in technology, with significant growth in the use of these models for content creation, customer service, and more.
A major challenge in the field of large language models is the adoption of the technology by large organizations so that the models can help make quick, highly reliable business decisions based on the organization's internal information. Language models trained on internet data cannot meet this need because they are not exposed to the organization's internal information and suffer from "hallucinations" that undermine reliability and the adoption of the technology by organizations. The solution to this need is to make current and relevant organizational information accessible to the model during its operation and base the received answers on this information. Solutions addressing this need are referred to as RAG (Retrieval-Augmented Generation).