Why the Old Split Between Databases Is Breaking Down
For decades, enterprise data infrastructure has been divided into two camps. Analytical databases help businesses spot trends and answer big questions, while operational databases quietly record the daily flow of orders, payments, and customer activity. Engineers have long accepted that these jobs require separate systems, with data copied and moved between them through pipelines that are expensive to build and fragile to maintain. Munich-based CedarDB is betting that this division is no longer necessary, and its timing may be better than it first appears.
The company grew out of Umbra, a research project at the Technical University of Munich that was built from scratch to outperform legacy engines by taking full advantage of modern hardware. CedarDB’s cofounder and CEO, Moritz Sichert, says the platform handles both transactional and analytical work without the usual tradeoffs. Its current momentum comes largely from large enterprises whose existing stacks were designed to store years of transactions efficiently, but struggle when staff or software ask fast, natural-language questions about that same history.
Why Agentic AI Raises the Stakes
The argument becomes sharper once AI agents enter the picture. A conventional analyst might run a handful of reports each week. An autonomous agent, by contrast, may query live operational data continuously, pulling inventory levels, checking payment status, and reasoning about trends in the same moment a customer is placing an order. That pattern collapses the time gap that separated analytics from operations, and it punishes architectures that rely on nightly batch copies.
Databricks reached a similar conclusion this year. At its Data and AI Summit in June, the company cited “the agentic era” as the reason for launching LTAP, a data architecture meant to unify transactions, analytics, streaming, and operational data on a single copy of storage. When two data companies with very different histories arrive at the same architectural idea, it usually signals genuine market pressure rather than a passing trend. Sichert reads the move as validation that demand exists for what CedarDB has been building.
Compatibility, Cloud Plans, and the Road Ahead
CedarDB’s strategy also hinges on pragmatism. Its engine is compatible with PostgreSQL, the widely used open-source database, which keeps switching costs low because customers do not need to rewrite their applications. The company also made a deliberate choice to launch an on-premise enterprise product and a free community edition before a managed cloud service. Its CedarDB Cloud entered limited availability this week, following beta testing with waitlisted users. A hosted version should broaden the customer base beyond large enterprises, though it will face established competition from well-funded data platforms.
The company’s team of 14 is small by comparison, and its cofounders have openly said they expect to raise another round once traction grows. Whether a German spinout can challenge larger American rivals will depend on execution, pricing, and whether enterprises trust a newer engine with mission-critical workloads. Still, the underlying question CedarDB raises is one that every data-driven organization is now asking.
For consumers and business buyers, this shift has practical implications. When databases can answer analytical questions in real time against live transactions, the software products built on top of them can recommend, price, and personalize with far less lag. Shoppers may see more responsive product suggestions, companies may adopt AI-driven tools faster, and procurement teams will increasingly weigh data architecture as a purchase decision in its own right. Tracking which unified platforms gain enterprise traction is therefore a useful signal of where technology adoption, and buying intent, is heading next.
