AI News AI资讯 19h ago Updated 14h ago 更新于 14小时前 46

OliverAI Raises Pre-Seed Funding to Scale Agentic AI Database Platform OliverAI 获得种子前融资,加速扩展 Agentic AI 数据库平台

OliverAI secured pre-seed funding from Menlo Ventures and Unusual Ventures to scale OliverDB, an analytical data platform purpose-built for agentic AI systems OliverDB claims hundreds of times faster performance than ClickHouse on CPUs and thousands of times faster with GPU execution, based on ClickBench benchmarks The platform addresses a critical gap: enterprise analytics infrastructure cannot handle the continuous, machine-speed workloads generated by AI agents OliverDB introduces a "model sw OliverAI获Menlo Ventures和Unusual Ventures pre-seed融资,用于发展OliverDB——专为agentic AI系统构建的分析数据平台 当前企业分析基础设施无法处理AI agent产生的持续、机器速度工作负载,导致性能瓶颈和计算成本上升 性能测试显示OliverDB在ClickBench查询中比ClickHouse快数百倍(CPU)至数千倍(GPU),可将多PB环境从数千服务器缩减至单GPU运行 OliverDB提供跨数据库和MCP服务器的可观测性与治理功能,支持策略设置和agent行为追踪 采用"模型群"方法并行运行多个专用模型,将前沿模型保留给真

65
Hot 热度
70
Quality 质量
65
Impact 影响力

Analysis 深度分析

TL;DR

  • OliverAI secured pre-seed funding from Menlo Ventures and Unusual Ventures to scale OliverDB, an analytical data platform purpose-built for agentic AI systems
  • OliverDB claims hundreds of times faster performance than ClickHouse on CPUs and thousands of times faster with GPU execution, based on ClickBench benchmarks
  • The platform addresses a critical gap: enterprise analytics infrastructure cannot handle the continuous, machine-speed workloads generated by AI agents
  • OliverDB introduces a "model swarm" architecture that runs multiple specialized models in parallel, reserving frontier models only for tasks that truly require them
  • The platform provides observability and governance across databases and MCP servers, enabling enterprises to set policies and track every agent action

Why It Matters

This funding and product announcement signals that the AI infrastructure layer is evolving to meet the unique demands of agentic AI, which generates fundamentally different workload patterns than traditional enterprise analytics. For AI practitioners, it highlights that data access and governance—not just model capability—are becoming the critical bottleneck for enterprise AI deployment. The performance claims, if validated independently, could dramatically reduce the compute costs that currently threaten the ROI of enterprise AI investments.

Technical Details

  • Performance: OliverDB reportedly runs hundreds of times faster than ClickHouse on CPUs and thousands of times faster with GPU execution on ClickBench query shapes, enabling multi-petabyte environments to run on a few servers or a single GPU
  • Architecture: The "model swarm" approach deploys multiple specialized models in parallel, intelligently routing tasks and reserving frontier-tier models only for workloads that genuinely require their capability
  • Governance & Observability: OliverDB provides policy enforcement and action tracking across databases and MCP (Model Context Protocol) servers, addressing the enterprise need for governed, reliable data access
  • Deployment: Available as a managed service or within customer VPCs, supporting both cloud-native and on-premises enterprise environments
  • Founders: Praneet Sharma, Grace Johnson, and Toby O'Brien built the platform to specifically address the mismatch between traditional analytics infrastructure and agentic AI workloads

Industry Insight

  • The agentic AI infrastructure market is rapidly taking shape, with investors drawing parallels to the Snowflake/Databricks eras—early movers in specialized AI data layers could capture significant enterprise value
  • Enterprises should prioritize data governance and observability solutions alongside model selection, as uncontrolled agent data access poses both cost and compliance risks
  • The performance claims around GPU-accelerated analytical workloads for AI agents could reshape cost models for enterprise AI, potentially making previously uneconomical agent-scale deployments viable

TL;DR

  • OliverAI获Menlo Ventures和Unusual Ventures pre-seed融资,用于发展OliverDB——专为agentic AI系统构建的分析数据平台
  • 当前企业分析基础设施无法处理AI agent产生的持续、机器速度工作负载,导致性能瓶颈和计算成本上升
  • 性能测试显示OliverDB在ClickBench查询中比ClickHouse快数百倍(CPU)至数千倍(GPU),可将多PB环境从数千服务器缩减至单GPU运行
  • OliverDB提供跨数据库和MCP服务器的可观测性与治理功能,支持策略设置和agent行为追踪
  • 采用"模型群"方法并行运行多个专用模型,将前沿模型保留给真正需要的任务,现已提供托管服务或客户VPC内部署

为什么值得看

这篇文章揭示了agentic AI时代数据基础设施的关键变革方向。OliverDB针对AI agent特有的持续、高速工作负载进行了架构优化,解决了企业级AI部署中日益突出的性能与成本瓶颈问题。

技术解析

  • 性能突破:基于ClickBench基准测试,CPU场景下比ClickHouse快数百倍,GPU场景下快数千倍,实现了从数千服务器到单GPU的规模化压缩,创始人称之为"根本性跃迁"而非增量改进
  • 架构创新:采用"模型群"(model swarm)并行架构,多个专用模型同时运行,前沿模型仅用于真正需要的高复杂度任务,实现性能与成本的平衡
  • 治理与可观测性:提供跨数据库和MCP服务器的统一治理框架,支持策略配置和agent行为追踪,满足企业级合规需求
  • 部署灵活性:支持托管服务和客户VPC内部署两种模式,适应不同企业的安全和合规要求

行业启示

  • 数据基础设施需要重新设计以适应AI agent的持续、高速工作负载,传统批处理架构已无法满足agentic AI时代的需求
  • 企业AI投资回报的关键不仅在于模型能力,更在于数据访问的治理、可靠性和成本效率
  • 新一代数据平台有望像Snowflake和Databricks定义前一个时代一样,成为agentic AI基础设施的核心基石

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

Agent Agent Funding 融资 GPU GPU LLM 大模型 Deployment 部署