AI Observability: Dashboards

The AI Observability module includes 8 dashboards, each covering a separate slice of the AI infrastructure.

Dashboard	Purpose
Infrastructure Overview	Summary operational picture of all AI components
Agents	AI agent behavior: requests, responses, tokens, durations
Traces	Spans, statuses, and trace distribution
LLM Costs	Cost by models and providers
Service Metrics	Inference and AI service metrics
GPU Status	Load, memory, temperature, errors
Claude Code	Local Claude Code activity: dialogs, tokens, events, projects
Codex	Local Codex activity: requests, TTFT, tokens, sessions, projects

Infrastructure Overview

Summary AI Observability dashboard for operational control of the entire AI infrastructure.

Agents

AI Observability dashboard for analyzing AI agent behavior - requests, responses, tokens, and durations.

Traces

AI Observability dashboard for analyzing AI request traces - spans, trace groups, statuses.

LLM Costs

AI Observability dashboard for analyzing LLM costs - tokens, cost, models, providers.

Service Metrics

AI Observability dashboard for analyzing AI service runtime metrics - vLLM, queue, latency, KV cache.

GPU Status

AI Observability dashboard for monitoring GPUs - load, memory, temperature, ECC errors.

Claude Code

AI Observability dashboard for monitoring local Claude Code - dialogs, tokens, events, users, projects.

Codex

AI Observability dashboard for monitoring local Codex - requests, tokens, TTFT, durations, projects.