New Relic 推出 Autopilot Kubernetes 智能体：90 多个诊断检测器定位集群故障根源，现开放公开预览
New Relic 公司 11 月 11 日宣布推出 New Relic Autopilot Kubernetes Agent，现开放公开预览。该工具面向站点可靠性工程师与平台工程师，是 New Relic Autopilot 智能体系列中专门针对 Kubernetes 环境的成员。
公司称，该智能体可自动定位故障根源，并生成带上下文的修复代码，交由工程师审核后执行，使团队能在故障影响终端用户之前处理异常。
New Relic 援引云原生计算基金会的数据称，运行生成式超级智能（SI）的组织中有 66% 依赖 Kubernetes。公司同时表示，多集群扩张正在放大专业人才缺口，复杂集群故障的诊断需要深厚的 Kubernetes 知识。
按公司发布，该智能体在 Kubernetes 事件、对象描述、日志与应用性能监控轨迹之上运行 90 多个专用诊断检测器，隔离根源并交付可直接运行的修复代码。其底层为 New Relic Ground Truth 上下文引擎，基于 OpenTelemetry 评估高度关联的运行数据。
该智能体针对的故障类型包括噪声邻居与计算资源问题、调度失败、网络错误，以及应用负载均衡器与入口等云配置错误。公司称，工具内置人工介入选项，工程师可在调查过程中保持监督与控制。
New Relic 首席产品官布莱恩·埃默森在发布中说，关键 Kubernetes 基础设施出故障时，工程师需要的是明确答案，而不只是更多告警；该智能体把专用智能体超级智能（SI）直接嵌入工程师的容器化工作流，在让工程师保持控制的同时消除复杂集群故障中的猜测。
公司表示，参观者可在 11 月 9 日至 12 日于犹他州盐湖城举行的 KubeCon North America 2026 大会上，在公司展位观看该智能体的现场演示。
New Relic Launches Autopilot Kubernetes Agent in Public Preview: More Than 90 Diagnostic Detectors to Pinpoint Cluster Root Causes
New Relic announced the New Relic Autopilot Kubernetes Agent on November 11, now available in public preview. The tool is designed for site reliability engineers and platform engineers as a specialized member of the New Relic Autopilot fleet of agents.
The company said the agent automatically pinpoints root causes and generates context-aware fix code for engineers to review and execute, allowing teams to address anomalies before disruptions affect end users.
Citing the Cloud Native Computing Foundation, New Relic said 66 percent of organizations running generative super intelligence (SI) rely on Kubernetes, while multi-cluster sprawl is exposing an expertise gap in diagnosing complex cluster incidents.
The agent runs more than 90 specialized diagnostic detectors across Kubernetes events, object descriptions, logs and application performance monitoring traces, the company said. It is powered by the New Relic Ground Truth context engine, built on OpenTelemetry, to evaluate highly correlated operational data.
Failure types targeted include noisy neighbors and compute resourcing problems, scheduling failures, network errors and cloud misconfigurations such as application load balancer and ingress issues, according to the release. Human-in-the-loop options let engineers keep oversight during investigations.
New Relic Chief Product Officer Brian Emerson said in the release that when critical Kubernetes infrastructure fails, engineers need clear answers, not just more alerts, and that the agent embeds specialized agentic super intelligence (SI) into containerized workflows while keeping the engineer in control.
The company said attendees can see live demonstrations at KubeCon North America 2026 in Salt Lake City, Utah, from November 9 to 12.
发布：GSM·美国（发布链接）
图片来源：图：Kubernetes 的官方标志（船舵造型）。Google, Inc. 供图，公共领域，经 Wikimedia Commons / Photo: Kubernetes logo without workmark. By Google, Inc., public domain, via Wikimedia Commons
GSM Global · 本篇英文版 English edition of this story
