An embedded graph database for the on-device AI era.

Kùzu built an embedded, open-source graph database designed for query speed, scalability, and ease of use, letting developers model and query complex relationships in their data without running a separate database server.

Kùzu was founded in 2023 by Semih Salihoğlu, an associate professor of computer science at the University of Waterloo, and Ardan Araç, previously a Senior Director at Google and SVP at Celonis, together with several members of Salihoğlu's research group.

As AI systems move from stateless copilots to agentic workflows, memory becomes a core infrastructure problem. Models need to understand entities, relationships, context, and history across time, not just retrieve chunks of text. Kùzu sits directly in that gap: an embedded graph database that gives developers a lightweight, production-ready way to model relationship-rich memory inside AI-native applications. The team combined deep database systems research with a developer-first product mindset, which is exactly the kind of technical wedge we look for.