{
  "title": "AI Observability with Grafana",
  "video_id": "FbQTLxFdb6I",
  "published": "2026-07-29T07:04:56+0000",
  "complexity": "Full Setup",
  "old_url": "https://workflows-of-ai.com/workflow/ai-observability-grafana",
  "description": "Send Drupal AI calls over OpenTelemetry to Grafana, Tempo, Mimir and Loki to see latency, tokens, cost and agent traces. Set up in one command.",
  "subtitle": "See which model ran, how long it took, what it cost and what your agents said to each other, in Grafana dashboards on your own server.",
  "summary": "The AI Observability module logs every AI call, and the OpenTelemetry recipe sends it through Alloy to Tempo, Mimir and Loki, shown in Grafana. The One Line Installer builds the whole stack from nothing, with a demo setup that adds the API Explorer, an OpenAI key and a parent agent with a sub-agent. The dashboards show request rate and latency, tokens and cost priced from models.dev, and agent sessions with the messages between parent and sub-agent.",
  "categories": [
    "Drupal AI",
    "Agents"
  ],
  "intro_html": "<p>Ever added AI to a Drupal site and wondered which model actually ran, how slow it was, how many tokens it used and what that cost? And when an AI agent calls sub-agents, what really happened inside?</p><p>This video shows a complete, self-hosted AI observability stack for Drupal that answers all of it - built and running in one command.</p><p>Using the One Line Installer, we spin up a Drupal site wired to Grafana, Tempo, Mimir, Loki and Alloy, then watch every AI call show up live: request rate and latency, token usage, real cost priced from models.dev, and a full parent to sub-agent trace where you can read the actual prompts and responses flowing between agents.</p><p>Everything is open source, provider-agnostic (OpenAI, Anthropic, Gemini, Mistral, Fireworks and more) and self-hosted - your prompts and data never leave the box.</p><p>What's covered:</p><p>One command to build Drupal plus a full observability stack and demo AI agents. How the pieces fit together: the AI Observability module sends data over OpenTelemetry to Alloy, which routes traces to Tempo, metrics to Mimir and logs to Loki, all visualized in Grafana. The requests overview with rate, errors and latency. The token and cost dashboard with live pricing from models.dev. And the AI Agent Sessions view, with a nested agent to sub-agent tree and the prompt-and-response message flow between them.</p><p>Links:</p><p>AI Observability OpenTelemetry recipe: <a href=\"https://www.drupal.org/project/ai_observability_opentel_recipe\">https://www.drupal.org/project/ai_observability_opentel_recipe</a></p><p>One Line Installer: <a href=\"https://www.drupal.org/project/one_line_installer\">https://www.drupal.org/project/one_line_installer</a></p><p>Drupal AI module: <a href=\"https://www.drupal.org/project/ai\">https://www.drupal.org/project/ai</a></p><p>Model pricing data: <a href=\"https://models.dev\">https://models.dev</a></p>",
  "intro_added": false,
  "chapters": [
    "0:00 What AI Observability does",
    "1:38 How the stack fits together",
    "3:18 The One Line Installer",
    "4:16 Installing the demo setup",
    "7:15 Keys and the API Explorer",
    "8:45 The Grafana dashboards",
    "9:08 Tokens and cost from models.dev",
    "9:29 Request rate and latency",
    "11:03 Parent and sub-agent traces"
  ],
  "corrections": [
    [
      "Freely Give",
      "FreelyGive"
    ],
    [
      "open telemetry",
      "OpenTelemetry"
    ],
    [
      "Open Telemetry",
      "OpenTelemetry"
    ],
    [
      "Cloud Code",
      "Claude Code"
    ],
    [
      "open AI API",
      "OpenAI API"
    ],
    [
      "brush",
      "Drush"
    ],
    [
      "pseudo",
      "sudo"
    ]
  ],
  "ai_note": "The video and the introduction are Marcus's own, first published on workflows-of-ai.com. AI wrote the subtitle, the summary and the video chapters from the transcript, and corrected misheard names in YouTube's automatic captions."
}