Keewano launches event-series database KeewanoDB on Google Cloud
Summary
Keewano has launched KeewanoDB, an event-series database designed specifically for AI agents that need to maintain a detailed sequence of events rather than relying on traditional flattened database structures. By keeping event history intact, this system allows agents to read and analyze data more efficiently, addressing a common issue faced by enterprises where confident but incorrect answers arise from missing business context. The importance of this development is underscored by recent research indicating that a significant number of organizations struggle with inconsistent data, highlighting the need for a solution that provides a coherent event narrative for agents to draw upon.
Tokens
$12M
Analysis
Keewano: Keewano is a startup that developed KeewanoDB, a specialized event-series database built to preserve full sequences of events for AI agents rather than flattening them into traditional tables or aggregates. The company recently announced general availability of KeewanoDB as a managed service on Google Cloud, along with plans for additional cloud and self-managed options. Co-founders Mark Kardashov and Pavel Bibergal created the system after identifying gaps in existing data infrastructure for supporting agent reasoning about event history and causality. Mark Kardashov: Mark Kardashov is the co-founder and CEO of Keewano, where he leads product framing and market positioning for the event-series database. He has highlighted how agents and machines need different data capabilities than traditional users, driving the decision to maintain complete event sequences instead of discarding context at write time. Kardashov collaborates with co-founder Pavel Bibergal on an architecture that supports direct agent connections via SDK and the Model Context Protocol. Pavel Bibergal: Pavel Bibergal is the co-founder and CTO of Keewano, responsible for the technical design of KeewanoDB after his prior experience running data infrastructure at Plarium. He began prototyping the event-series approach after standard warehouses proved inadequate for real-time answers to complex business questions involving agents and large language models. Bibergal's contributions include the storage format and semantic layer that attach evolving context to each entity's event sequence without joins or data loss. ML Capabilities: Many existing database systems offer machine learning functionality that is typically retrofitted onto relational foundations, often requiring data transformation and external processing rather than native integration. Database Vendor Trends: Major database providers including Oracle and Couchbase have updated their platforms to better support persistent agent memory and context across transactional and analytical workloads. Agent Context Challenges: Enterprises face recurring issues where AI agents generate confident but incorrect answers because business context is missing or inconsistent in their data systems.
Categories
techai_agentsaimachine_learningvirtuals