Iceland-based Videntifier Technologies ehf. has announced that its Lithuanian subsidiary, VectorTree, has received funding from the European Union for a project titled “Building an Advanced Vector Database for Artificial Intelligence”.
The funding is part of the European Regional Development Fund under the EU Funds Investment Programme 2021–2027. The total project budget is €943,052.28. Of that, €565,831.38 will be provided by the European Union.
Developing a scalable vector database
VectorTree is creating an Advanced Vector Database (AVD) as a cloud-based service to support artificial intelligence systems. This database is built on Videntifier’s NV-tree algorithm, which is known for processing large volumes of visual data efficiently.
Ari Kristinn Jónsson, CEO of VectorTree, says, “As AI systems increasingly process vast volumes of unstructured data—such as images, video, text, audio, and sensor input—they require scalable infrastructure. With the EU’s support, we’re building a vector database that not only scales, but also changes how data for AI is retrieved, increasing accuracy without relying solely on exact matches.”
AI systems often work with unstructured data. To make this data useful, AI models convert it into vector embeddings—numerical formats that represent key features. VectorTree’s AVD will store these embeddings and allow fast similarity searches, enabling AI to compare new inputs with stored data and retrieve results instantly.
A key feature of VectorTree’s AVD is vector clustering. Unlike typical databases that return single matches, this system can handle groups of related vectors.
For example, on video platforms, it can group and match scenes with similar visual patterns and timing, helping AI suggest content that aligns with user preferences. In legal technology, it can match past cases that use different wording but convey similar meaning, supporting more comprehensive research. In healthcare, it can surface prior cases with comparable symptoms or histories, helping AI assist with clinical decisions.
Another major strength is scalability. While most vector databases manage a few billion vectors, AVD is being developed to handle tens to hundreds of billions without losing speed or accuracy. This capability comes from the foundation of Videntifier’s NV-tree technology, which is already in use by organisations like INTERPOL and Meta for large-scale visual search.
VectorTree’s AVD is being developed with support from the European Union and is expected to be available for initial sales in 2025, with full development continuing through early 2027. The company aims to serve AI researchers, developers, and enterprises with a platform built for scale and performance.
Brief about Videntifier Technologies
Videntifier Technologies Ehf, founded in 2008 and based in Iceland, develops video identification tools to assist organisations in detecting illegal content online.
The company supports law enforcement, online platforms, and CSAM hotlines in managing large volumes of multimedia data. One of its key partners is the National Center for Missing and Exploited Children (NCMEC), which uses Videntifier’s technology to reduce reliance on manual review of harmful media.
The company originated from research at Reykjavik University and IRISA-CNRS and continues to apply advances in multimedia databases and computer vision to its products.
