OliverAI today announced pre-seed funding from Menlo Ventures and Unusual Ventures to scale OliverDB, its high-performance analytical data platform purpose-built for agentic AI.
SAN FRANCISCO, Aug. 26, 2026 /PRNewswire/ — Today’s enterprise analytics infrastructure wasn’t built for agentic AI. Unlike humans, AI agents create continuous, unpredictable, machine-speed analytical workloads, which existing systems weren’t designed to handle. As AI deployments scale, infrastructure can become both a performance bottleneck and a cost problem—limiting how quickly agents can work, while driving compute consumption up faster than the value they create. That puts the performance and promised ROI of enterprise AI at risk.
OliverDB is designed for this new reality. Its analytical engine enables AI agents to investigate enterprise data in real time while using a fraction of the compute required by existing systems. In tests using query shapes from ClickBench, the open-source analytical database benchmark developed by ClickHouse, Oliver ran hundreds of times faster than ClickHouse on CPUs and thousands of times faster with GPU execution.*
“This is not a 25% or 50% improvement to existing analytics infrastructure. It’s a complete step change,” said Praneet Sharma, co-founder of OliverAI. “A multi-petabyte analytics environment which requires thousands of servers today can collapse down to just a couple of servers—or even a single GPU—on Oliver, while simultaneously delivering dramatically faster performance. That changes what AI agents can do, how quickly they can do it, and the economics of operating them at enterprise scale.”
“Snowflake and Databricks helped define new categories of data infrastructure for the cloud era,” said Tim Tully, partner at Menlo Ventures. “Agentic AI represents another fundamental shift. Oliver represents a new kind of infrastructure designed specifically for the needs of agentic AI—combining the performance, efficiency, and control enterprises need to deploy AI agents at scale.”
Govern and observe agents without constraining them
Performance alone is not enough. Enterprises also need precise control over which systems agents can access, what information they can use, and which actions they can take.
OliverDB provides agent observability and governance across enterprise databases and Model Context Protocol (MCP) servers. Enterprises can define policies for each agent, limit its available data and actions, and maintain an attributable record of every interaction—giving agents room to operate while keeping every action within policy.
“The challenge in enterprise AI is no longer simply whether a model is capable enough,” said John Vrionis, founder at Unusual Ventures. “The challenge is giving AI systems meaningful access to valuable enterprise data without sacrificing control, reliability, or trust. Oliver combines the performance, governance, and grounded intelligence required to deploy those systems in the real world.”
Better outputs with fewer tokens.
Oliver’s model swarm is designed to produce deeper, more reliable outputs while lowering token spend and reducing dependence on expensive frontier models. Instead of relying on one model to generate a single answer, Oliver runs many smaller, specialized models in parallel to test competing hypotheses, examine alternative explanations, and build on useful findings from prior investigations. A conductor model weighs the results and returns the conclusion best supported by the available data—or no conclusion when the data is insufficient. This gives enterprises more thorough analysis while reserving frontier models for work that actually requires them.
OliverDB is available today for enterprise deployments, either as a managed service or inside a customer’s VPC. Enterprises can test the platform against their own demanding analytical workloads, using their own data and queries, to measure its performance and infrastructure impact directly.
Learn more and start using Oliver today, at oliverdb.ai.
*Benchmark results are based on internal Oliver testing using ClickBench analytical workloads against ClickHouse 24.8. Results are workload- and configuration-dependent and have not been independently audited. View the complete methodology, configurations, and results.
For media or general inquiries: info@oliverdb.ai
View original content to download multimedia:https://www.prnewswire.com/news-releases/menlo-ventures-and-unusual-ventures-back-oliver-ai-to-redefine-data-infrastructure-for-agentic-ai-302860322.html
SOURCE Oliver AI, Inc.
