- Published on
CedarDB wants one database to handle both transactions and analytics, as AI agents raise the stakes
- Authors
- Name
- DeepHorus
Munich-based database startup CedarDB is betting one engine can handle both operational transactions and analytical queries, a combination it says is increasingly urgent for agentic AI. The Technical University of Munich spinout announced limited availability of CedarDB Cloud on Wednesday at TechCrunch Disrupt. CEO Moritz Sichert said the rise of AI agents has made the dual-purpose approach more relevant.
Traditionally, databases split these workloads. Operational databases store daily records like orders and payments, while analytical databases help companies spot trends. CedarDB argues enterprises now need both capabilities in one system, especially for natural-language questions about years of transaction data.
CedarDB grew out of Umbra, a research project built from scratch to outperform legacy database engines by exploiting modern hardware. Five Umbra researchers co-founded the company, which has spent a decade developing its technology. The team prioritized reliability, fault tolerance, and compatibility with popular tools, co-founder Christian Winter wrote.
A key choice was PostgreSQL compatibility. According to Sichert, it keeps switching costs low because users need not rebuild applications, and it aligns with the logic large language models know best from training. CedarDB launched CedarDB Enterprise as an on-premise offering and CedarDB Community Edition in May last year, alongside a âŹ5.3 million seed round.
The cloud offering was delayed, not abandoned. After beta testing with waitlist users, CedarDB is now serving limited cloud users with a team of 14 people. Sichert expects the company will need a larger team as traction grows, and that is when it would raise its next round.
Competition is intensifying. Databricks cited the âagentic eraâ in June when it launched LTAP, a data architecture aiming to unify transactions, analytics, streaming, and operational data on one copy of storage. Sichert sees Databricksâ move as validation of demand, though it is too early to judge the shiftâs disruptiveness.
