Artificial intelligence news.
Straight from the original announcements.

Red Hat OpenShift 5 adds supervised AI for incident diagnosis

November 11, 2026By the AI News Bites Editorial TeamSource: Red Hat
Red Hat OpenShift 5 adds supervised AI for incident diagnosis
Image: Red Hat

Red Hat OpenShift 5 adds supervised AI capabilities designed to help platform teams investigate incidents, create observability dashboards and review remediation plans. OpenShift Lightspeed can organize evidence from metrics, logs and traces, while an opt-in agentic workflow stages proposed actions for human approval.

Key points
  • OpenShift Lightspeed structures evidence from metrics, logs and traces.
  • AI-assisted Perses panels can be generated from natural-language requests.
  • Agentic troubleshooting is an opt-in OpenShift 5 technology preview.
  • Every cluster state change requires explicit administrator approval.
  • Organizations can bring compatible model endpoints and control processing.

Lightspeed turns telemetry into an investigation draft

Red Hat says modern enterprise environments already contain extensive Prometheus metrics, traces and logs, but engineers can spend significant time moving between consoles and queries before deciding whether an alert represents a genuine emergency. OpenShift observability and OpenShift Lightspeed are intended to reduce that diagnostic work by gathering evidence, investigating likely causes and presenting proposed remediation for review.

Starting from an alert or anomaly, OpenShift Lightspeed can inspect current cluster state and configured telemetry sources, including Prometheus metrics, pod logs and OpenTelemetry distributed traces. Through the observability Model Context Protocol server, it can help with notification triage and deeper telemetry inspection.

The resulting investigation draft contains three elements: Source, which identifies where the evidence came from; Signal, which highlights the detected anomaly; and Finding, which provides a plain-language inference. Red Hat gives the example of slow database calls in shared-services causing connection timeouts in payments-api. The engineer remains responsible for deciding whether the finding is valid, particularly when noisy errors appear alongside a genuine underlying issue.

This conversational troubleshooting mode is available on OpenShift 4.22 and later when using the OpenShift Lightspeed operator 1.1.5 or later.

Natural-language requests create Perses dashboard panels

OpenShift Lightspeed can also work with the cluster observability operator and the Red Hat build of Perses to generate real-time metrics panels inside the OpenShift console. Engineers can ask questions in ordinary language instead of starting with a blank Prometheus Query Language editor.

Examples in the announcement include asking which pods were crashlooping in the last 5 minutes, which pods use the most CPU, how memory is being used across the payment-service namespace, or which pods have the highest network ingress. Lightspeed then generates PatternFly-themed panels.

Red Hat has built a verification step into this workflow. With cluster observability operator 1.6, each AI-assisted panel displays an AI-assisted badge. Users can inspect the underlying PromQL, check the time range and validate the source metrics before pinning a panel to a production Perses dashboard.

Agentic troubleshooting is an opt-in technology preview

OpenShift 5 introduces an opt-in agentic troubleshooting workflow as a technology preview. It requires OpenShift 5, OpenShift Lightspeed operator 2 and a configured tool-calling large language model. No default model is shipped.

The workflow uses an adapter that watches Alertmanager and respects existing receivers such as Critical or Default. When an allowed alert fires, the agent can run standard oc and kubectl diagnostic commands, draft a root-cause diagnosis and stage a proposed remediation plan in a decision queue. The plan is then available for review in the OpenShift web console.

If an approved remediation fails to clear the triggering condition, the agent stops and escalates the incident to a human operator rather than repeatedly attempting other fixes.

Guardrails and model choice remain central

Red Hat describes three mandatory controls for the troubleshooting framework: human approval for every change to cluster state, strict role-based access control sandboxing through ephemeral secure execution environments, and complete auditability of observations, proposed actions and execution outcomes.

Users can track agentic runs with Tempo, with direct links to audit traces available in the Agentic Runs interface in the OpenShift web console. The traces are intended to support organizational AI compliance needs.

OpenShift Lightspeed uses an open provider architecture rather than shipping a pre-packaged default large language model. Supported endpoints include hosted commercial APIs, IBM watsonx, Red Hat OpenShift AI, Red Hat Enterprise Linux AI and other compatible endpoints. Organizations can choose where model requests are processed, disable conversation telemetry collection and use locally hosted models in fully disconnected, air-gapped environments. The source says cluster data processing depends on the configured model endpoint, while administrative identity aligns with console RBAC.

For users, the announcement positions OpenShift 5 as a way to move from scattered telemetry toward structured, human-verified decisions. The capabilities can assist with evidence gathering and preparation, but the stated controls keep operational changes behind administrator approval.

Red HatOpenShift 5OpenShift LightspeedPersesenterprise AIdata centers

Related news

Huawei unveils AI Data Platform for European businesses
Chips and infrastructureNovember 16, 2026, 09:56

Huawei unveils AI Data Platform for European businesses

Huawei has introduced a new AI Data Platform in Florence, positioning it as infrastructure for improving data management and the performance of artificial intelligence…

Red Hat OpenShift Lightspeed 2 adds live and autonomous troubleshooting
Products and toolsNovember 12, 2026

Red Hat OpenShift Lightspeed 2 adds live and autonomous troubleshooting

Red Hat OpenShift Lightspeed 2 can now investigate OpenShift cluster problems directly, either through a live troubleshooting chat or an autonomous agentic run. The…

Red Hat Makes RHEL MCP Server Generally Available With PCP Diagnostics
Products and toolsNovember 11, 2026

Red Hat Makes RHEL MCP Server Generally Available With PCP Diagnostics

Red Hat has made its Model Context Protocol server for Red Hat Enterprise Linux generally available, giving AI assistants supported, read-only access to system…