Analysis
CedarDB used TechCrunch Disrupt's Startup Battlefield stage on October 14 to launch CedarDB Cloud in limited availability, betting that AI agents writing and querying data will force a rethink of the decades-old split between transactional and analytical databases, TechCrunch reported.
The technical bet
CedarDB is a Munich-based spinout of the Technical University of Munich, built on Umbra, a research database engine designed from scratch to run fast on modern hardware. Five Umbra researchers co-founded the company, led by CEO Moritz Sichert, with Lukas Vogel, Philipp Fent, Dominik Durner and Christian Winter. The pitch is that most databases force companies to run separate systems for transactions (OLTP) and analytics (OLAP), copying data between them -- CedarDB claims to handle both on one engine, without extra CPUs or memory, while staying PostgreSQL-compatible so switching costs stay low.
“The company raised a modest 5.3 million euro seed round in May 2025 and has not disclosed new funding alongside this product launch.”
Why AI agents raise the stakes
Sichert's argument is that AI agents querying a database don't behave like dashboards or batch reports -- they generate unpredictable, high-frequency queries that mix transactional writes with analytical lookups in real time, exactly the workload split traditional databases handle worst. PostgreSQL compatibility also matters for a different reason: it's the pattern most large language models were trained on, so agents writing database queries are more likely to generate correct SQL against a Postgres-compatible system than against a more obscure query dialect.
Competitive landscape
CedarDB's most direct large competitor is Databricks, which launched its own unified transaction-and-analytics architecture, LTAP, in June 2026, built on a single copy of data in its lake. CedarDB is a 14-person team competing against a company valued in the hundreds of billions -- a scale mismatch common in infrastructure, but one that still means CedarDB's PostgreSQL-compatibility argument needs to translate into actual enterprise migrations, not just developer interest. The company raised a modest 5.3 million euro seed round in May 2025 and has not disclosed new funding alongside this product launch.
What to watch: whether CedarDB can land its first non-pilot enterprise customer on CedarDB Cloud before Databricks' LTAP, or a similarly resourced incumbent, absorbs the same one-database-for-AI-agents pitch into a product most enterprises already pay for.
