OliverAI has announced pre-seed funding from Menlo Ventures and Unusual Ventures to grow OliverDB, an analytical data platform built specifically for agentic AI systems. The company argues that current enterprise analytics infrastructure cannot handle the continuous, machine-speed workloads that AI agents generate, creating both performance bottlenecks and rising compute costs that threaten the return on enterprise AI investment.
Co-founders Praneet Sharma, Grace Johnson, and Toby O’Brien built OliverDB to address this gap. According to the company, in tests using ClickBench query shapes, Oliver ran hundreds of times faster than ClickHouse on CPUs and thousands of times faster with GPU execution, allowing multi-petabyte environments that once required thousands of servers to run on just a few — or a single GPU.
Sharma described the improvement as a fundamental step change rather than an incremental gain. Tim Tully of Menlo Ventures compared the shift to how Snowflake and Databricks defined prior eras of data infrastructure, positioning Oliver as the analogous foundation for agentic AI. John Vrionis of Unusual Ventures said enterprises need not just capable models but governed, reliable access to data.
Beyond raw performance, OliverDB offers observability and governance across databases and MCP servers, letting enterprises set policies and track every agent action. Its “model swarm” approach runs multiple specialized models in parallel, reserving frontier models for tasks that truly require them. OliverDB is available now as a managed service or within customer VPCs.
Featured image: Credit: OliverAI
