Features and tutorials

Every screen in Constitutional AIOps, walked through with real screenshots and step by step. Each page covers what the feature is, what you see on it, and how to use it. If a feature degrades without an optional dependency (Neo4j, the LGTM stack, an LLM endpoint), the page says what you see instead.

Observe

• Dashboard: the live system view covering health, agent latency, recent activity, and the event-driven widgets.
• Generative UI: the computed widgets and the opt-in LLM insight layer, explained in full.
• Incidents and RCA: the incident list, confidence scoring, root-cause detail, and export.
• Metrics: measured agent latency and request metrics.
• Telemetry: logs, traces, and where each panel's data came from.

Investigate

• Graph Explorer: the episodic knowledge graph of incidents, root causes, actions, services, and entities.
• Console: the cockpit that puts the platform topology and the assistant side by side.
• Chat and Assistant: ask questions about your system with the constitutional gate in the loop.

Operate

• Infrastructure: servers and containers, health, and remediation actions.
• MCP tools: the tool registry every action is drawn from.
• Agent Hub: the fast and reasoning agents and their configured endpoints.
• Settings: safety, models, notifications, BYOK, and the AI-widget opt-in.

Evaluate

• Benchmark: run the RCA benchmark against a dataset and export the results.

Bring your own endpoint first

Chat, RCA explanations, and the LLM insight widgets need a configured OpenAI-compatible endpoint. If you have not set one up yet, start with Bring Your Own Endpoint.