<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
<channel>
  <title>AI 武林：OpenTelemetry</title>
  <link>https://aiwulin.itsmygo.uk/tools/opentelemetry/</link>
  <description>AI 武林收錄的內容裡，最新提到 OpenTelemetry 的 21 筆，每筆附中文摘要。</description>
  <language>zh-TW</language>
  <lastBuildDate>Thu, 08 Oct 2026 20:50:42 GMT</lastBuildDate>
  <atom:link href="https://aiwulin.itsmygo.uk/tools/opentelemetry/rss.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>谁说云原生过时了？Kubernetes 等主流项目正在为 Agent 重做“地基”</title>
    <link>https://aiwulin.itsmygo.uk/c/202ab9c211/</link>
    <guid isPermaLink="false">aiwulin-202ab9c211</guid>
    <pubDate>Thu, 08 Oct 2026 00:00:00 +0800</pubDate>
    <dc:creator>Tony Bai</dc:creator>
    <description>&lt;p&gt;指出雲原生技術如 Kubernetes 並未過時，反而正為 AI Agent 重構基礎設施。透過工作負載感知排程、GPU 動態資源分配及專用沙箱等進化，解決了 Agent 在狀態管理與算力浪費上的挑戰，並推動流量層與安全可觀測性的全面升級。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://tonybai.com/2026/10/08/who-says-cloud-native-is-outdated-agent-foundation&quot;&gt;https://tonybai.com/2026/10/08/who-says-cloud-native-is-outdated-agent-foundation&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>Using Parseable with Datasette for OpenTelemetry traces</title>
    <link>https://aiwulin.itsmygo.uk/c/4c7e094228/</link>
    <guid isPermaLink="false">aiwulin-4c7e094228</guid>
    <pubDate>Wed, 07 Oct 2026 00:00:00 +0800</pubDate>
    <dc:creator>Simon Willison's Weblog</dc:creator>
    <description>&lt;p&gt;介紹如何使用 Parseable 與 Datasette 搭配 OpenTelemetry 追蹤資料。作者透過 Codex 輔助設定流程，並展示如何在 Parseable 的網頁介面中直接檢視 Datasette 的追蹤記錄。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://simonwillison.net/2026/Oct/6/datasette-parseable-opentelemetry&quot;&gt;https://simonwillison.net/2026/Oct/6/datasette-parseable-opentelemetry&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>可觀測性與 LLMOps</category>
    <category>Codex</category>
  </item>
  <item>
    <title>From Vibes to Production: Evaluating and Shipping AI Agents That Work 101 — Laurie Voss, Arize AI</title>
    <link>https://aiwulin.itsmygo.uk/c/41464a215c/</link>
    <guid isPermaLink="false">aiwulin-41464a215c</guid>
    <pubDate>Tue, 06 Oct 2026 00:00:00 +0800</pubDate>
    <dc:creator>AI Engineer</dc:creator>
    <description>&lt;p&gt;Laurie Voss 示範如何透過追蹤（traces）與多層評估機制，解決 AI 代理在金融分析任務中因缺乏即時資料而失敗的問題。影片結合確定性檢查、忠誠度評分與自定義行動力評分，建立從失敗案例到生產環境的完整評估工作流。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=qZj7sqmidqA&quot;&gt;https://www.youtube.com/watch?v=qZj7sqmidqA&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>AI Agent 基礎</category>
    <category>AI 評測</category>
  </item>
  <item>
    <title>Advanced Workshop: Mastering AI Observability — Doug Guthrie, Braintrust</title>
    <link>https://aiwulin.itsmygo.uk/c/7d17dcc290/</link>
    <guid isPermaLink="false">aiwulin-7d17dcc290</guid>
    <pubDate>Tue, 06 Oct 2026 00:00:00 +0800</pubDate>
    <dc:creator>AI Engineer</dc:creator>
    <description>&lt;p&gt;Doug Guthrie 透過實作支援代理人的追蹤，示範如何將生產環境的資料轉化為開發迴圈中的具體改進。課程涵蓋建立追蹤、設定自動化評分與主題分類，並利用程式碼評分與 LLM 判官結合，最終透過 Pull Request 回饋開發流程。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=hfEczxdNyvU&quot;&gt;https://www.youtube.com/watch?v=hfEczxdNyvU&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>可觀測性與 LLMOps</category>
  </item>
  <item>
    <title>Why Your AI Agent Fails in Production (And How to Catch It)</title>
    <link>https://aiwulin.itsmygo.uk/c/c80507865d/</link>
    <guid isPermaLink="false">aiwulin-c80507865d</guid>
    <pubDate>Wed, 30 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Google Cloud Tech</dc:creator>
    <description>&lt;p&gt;展示如何建立端到端的 AI 代理評估管道，將本地演示中的問題轉化為生產環境的實際測試。觀眾能學習使用 Antigravity、OpenTelemetry 等工具標準化追蹤並設定自動化評分標準。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=wPdoZRbvaF4&quot;&gt;https://www.youtube.com/watch?v=wPdoZRbvaF4&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>AI Agent 基礎</category>
    <category>AI 評測</category>
  </item>
  <item>
    <title>Stop YOLO-ing AI Observability: Building Production-Ready Agents | Arize x AWS</title>
    <link>https://aiwulin.itsmygo.uk/c/5fa5c5663f/</link>
    <guid isPermaLink="false">aiwulin-5fa5c5663f</guid>
    <pubDate>Thu, 24 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Arize AI</dc:creator>
    <description>&lt;p&gt;AWS 資深經理 Nate Later 分享團隊如何從大型前沿模型原型轉向生產級 AI 系統，強調 OpenTelemetry 對非確定性軟體的重要性，並說明為何過度依賴 YOLO 式追蹤會造成技術債。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=J83LVtXgx3o&quot;&gt;https://www.youtube.com/watch?v=J83LVtXgx3o&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>可觀測性與 LLMOps</category>
  </item>
  <item>
    <title>How to Build Reliable Long-Running AI Agents | Temporal + Arize</title>
    <link>https://aiwulin.itsmygo.uk/c/92193df885/</link>
    <guid isPermaLink="false">aiwulin-92193df885</guid>
    <pubDate>Tue, 22 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Arize AI</dc:creator>
    <description>&lt;p&gt;由 Arize 與 Temporal 共同舉辦，示範如何建立能跨失敗重啟的長跑 AI 代理，並透過 Arize AX 追蹤模型與工具呼叫。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=TAXglqzS7F0&quot;&gt;https://www.youtube.com/watch?v=TAXglqzS7F0&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>What Is MLflow? Tracing AI Agents &amp; LLM Workflows</title>
    <link>https://aiwulin.itsmygo.uk/c/375e35f482/</link>
    <guid isPermaLink="false">aiwulin-375e35f482</guid>
    <pubDate>Thu, 17 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>IBM Technology</dc:creator>
    <description>&lt;p&gt;Legare Kerrison 說明傳統監控無法捕捉多代理 AI 系統的靜音失敗與非確定性問題，並介紹 MLflow 如何透過追蹤、LLM 評審與提示註冊來提升可觀測性與可靠性，提供從開發到生產部署的實作建議。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=iZX6d0OdZys&quot;&gt;https://www.youtube.com/watch?v=iZX6d0OdZys&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>可觀測性與 LLMOps</category>
    <category>MLOps</category>
  </item>
  <item>
    <title>Making Your Data Ready for Agentic AI</title>
    <link>https://aiwulin.itsmygo.uk/c/d9866ea09d/</link>
    <guid isPermaLink="false">aiwulin-d9866ea09d</guid>
    <pubDate>Thu, 27 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Martin Fowler</dc:creator>
    <description>&lt;p&gt;指出自主代理 AI 缺乏人類分析師的直覺與判斷，因此資料必須具備可信、有上下文、可追蹤、受管治與可操作五項屬性。作者提出透過資料合約、語義層與可觀測性架構，將這些隱性知識轉化為機器可理解的明確規則，讓資料真正準備好供 AI 使用。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://martinfowler.com/articles/making-data-ready-for-agentic-ai.html&quot;&gt;https://martinfowler.com/articles/making-data-ready-for-agentic-ai.html&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>AI Agent 基礎</category>
    <category>可觀測性與 LLMOps</category>
    <category>訓練資料與合成資料</category>
  </item>
  <item>
    <title>Arize AX Demo 2026: Build Self-Improving Agents with Signal</title>
    <link>https://aiwulin.itsmygo.uk/c/4f1d72987f/</link>
    <guid isPermaLink="false">aiwulin-4f1d72987f</guid>
    <pubDate>Tue, 25 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Arize AI</dc:creator>
    <description>&lt;p&gt;示範如何使用 Arize AX 自動追蹤、評估並修復 AI 代理，透過編碼代理自動配置追蹤並利用 Signal 發現未知失敗模式。觀眾能學習如何建立自我改進的代理系統，無需手寫追蹤程式碼。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=3fhQUi6yC4Y&quot;&gt;https://www.youtube.com/watch?v=3fhQUi6yC4Y&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>Browser automation with Pydantic AI + Playwright</title>
    <link>https://aiwulin.itsmygo.uk/c/3fc732858b/</link>
    <guid isPermaLink="false">aiwulin-3fc732858b</guid>
    <pubDate>Fri, 21 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>pamela fox's blog</dc:creator>
    <description>&lt;p&gt;介紹如何使用 Pydantic AI 搭配 Playwright 建立能自主瀏覽網頁的代理，並示範如何整合 Microsoft Foundry 模型與 Entra 無鑰匙認證。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://blog.pamelafox.org/2026/08/browser-automation-with-pydantic-ai.html&quot;&gt;https://blog.pamelafox.org/2026/08/browser-automation-with-pydantic-ai.html&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>工作流自動化</category>
    <category>瀏覽器與電腦操作</category>
    <category>開發框架</category>
  </item>
  <item>
    <title>How Evaluation-Driven Development (EDD) Works</title>
    <link>https://aiwulin.itsmygo.uk/c/4b8c517594/</link>
    <guid isPermaLink="false">aiwulin-4b8c517594</guid>
    <pubDate>Tue, 23 Jun 2026 00:00:00 +0800</pubDate>
    <dc:creator>Decoding AI Magazine（Paul Iusztin）</dc:creator>
    <description>&lt;p&gt;介紹 Evaluation-Driven Development (EDD) 方法，透過模擬輸入並執行真實代理來生成測試資料，用於在合併前驗證功能並檢測退步。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.decodingai.com/p/how-evaluation-driven-development-works&quot;&gt;https://www.decodingai.com/p/how-evaluation-driven-development-works&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>AI 評測</category>
  </item>
  <item>
    <title>MCP Servers Are Becoming the UI for AI Agents</title>
    <link>https://aiwulin.itsmygo.uk/c/a740443f30/</link>
    <guid isPermaLink="false">aiwulin-a740443f30</guid>
    <pubDate>Wed, 17 Jun 2026 00:00:00 +0800</pubDate>
    <dc:creator>Agentic Conversations（原 MLOps.community）</dc:creator>
    <description>&lt;p&gt;訪談 Opal 創辦人 Naseem Al-Naji，探討 MCP Server 如何成為 AI Agent 的介面，並介紹 MCPcat 如何透過分析工具呼叫與使用者目標，解決開發者與使用者脫節的問題。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://podcasters.spotify.com/pod/show/mlops/episodes/MCP-Servers-Are-Becoming-the-UI-for-AI-Agents-e3ks6n3&quot;&gt;https://podcasters.spotify.com/pod/show/mlops/episodes/MCP-Servers-Are-Becoming-the-UI-for-AI-Agents-e3ks6n3&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>MCP 模型上下文協定</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>从 GPU 到 Token：AI 基础设施的八层可观测性体系</title>
    <link>https://aiwulin.itsmygo.uk/c/2408c48866/</link>
    <guid isPermaLink="false">aiwulin-2408c48866</guid>
    <pubDate>Tue, 09 Jun 2026 00:00:00 +0800</pubDate>
    <dc:creator>Jimmy Song（宋净超）博客</dc:creator>
    <description>&lt;p&gt;提出從 GPU 硬體到業務成本的八層 AI 基礎設施可觀測性架構，指出僅看 GPU 利用率已不足以評估現代 AI 系統狀態。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://jimmysong.io/zh/blog/gpu-to-token-observability&quot;&gt;https://jimmysong.io/zh/blog/gpu-to-token-observability&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>AI 晶片與硬體</category>
    <category>可觀測性與 LLMOps</category>
  </item>
  <item>
    <title>The 2026-07-28 MCP Specification Release Candidate</title>
    <link>https://aiwulin.itsmygo.uk/c/8ababc24ac/</link>
    <guid isPermaLink="false">aiwulin-8ababc24ac</guid>
    <pubDate>Thu, 21 May 2026 00:00:00 +0800</pubDate>
    <dc:creator>Model Context Protocol Blog</dc:creator>
    <description>&lt;p&gt;Model Context Protocol 發布 2026 年 7 月 28 日規範候選版，核心變革為無狀態協議架構，移除連線握手與會話標頭，使伺服器可透過普通 HTTP 負載平衡部署。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://blog.modelcontextprotocol.io/posts/2026-07-28-release-candidate&quot;&gt;https://blog.modelcontextprotocol.io/posts/2026-07-28-release-candidate&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>MCP 模型上下文協定</category>
  </item>
  <item>
    <title>How to Find the Agent Failures Your Evals Miss with Scott Clark - #767</title>
    <link>https://aiwulin.itsmygo.uk/c/520aaa5a8a/</link>
    <guid isPermaLink="false">aiwulin-520aaa5a8a</guid>
    <pubDate>Fri, 08 May 2026 00:00:00 +0800</pubDate>
    <dc:creator>The TWIML AI Podcast</dc:creator>
    <description>&lt;p&gt;Scott Clark 分享如何透過後生產力分析發現標準評估遺漏的代理失敗，例如工具呼叫幻覺。透過將追蹤資料轉為向量指紋進行聚類與主題發現，建立反饋迴圈自動最佳化系統。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://twimlai.com/podcast/twimlai/how-find-agent-failures-your-evals-miss&quot;&gt;https://twimlai.com/podcast/twimlai/how-find-agent-failures-your-evals-miss&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>AI Agent 基礎</category>
    <category>AI 評測</category>
  </item>
  <item>
    <title>[ 概念介紹 ] ADLC 開發方式 — 透過 Agentic Development Lifecycle 開發模式，打造 Agent 服務系統</title>
    <link>https://aiwulin.itsmygo.uk/c/cb343391e1/</link>
    <guid isPermaLink="false">aiwulin-cb343391e1</guid>
    <pubDate>Tue, 28 Apr 2026 00:00:00 +0800</pubDate>
    <dc:creator>Simon Liu</dc:creator>
    <description>&lt;p&gt;介紹 ADLC（代理開發生命週期）如何解決傳統軟體開發模式無法應對 AI Agent 隨機性的問題。文章詳述了從行為契約定義、語義工具化、模型裁判評測到基礎設施自動化等核心實踐，幫助開發者將不確定性納入工程規範。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://medium.com/@simon3458/agent-adlc-intro-f5a279ca1ed7&quot;&gt;https://medium.com/@simon3458/agent-adlc-intro-f5a279ca1ed7&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>From Blind Spots to Observability: Operationalizing LLM Apps with OpenLit</title>
    <link>https://aiwulin.itsmygo.uk/c/61a649ce65/</link>
    <guid isPermaLink="false">aiwulin-61a649ce65</guid>
    <pubDate>Mon, 16 Feb 2026 00:00:00 +0800</pubDate>
    <dc:creator>AI Engineering Podcast</dc:creator>
    <description>&lt;p&gt;探討將大型語言模型應用推向生產環境時常見的盲點，包括模型行為不透明、Token 成本失控及提示詞管理困難。講者 Aman Agarwal 介紹了基於 OpenTelemetry 的 OpenLit 工具，提供視覺化追蹤、提示詞版本控制與成…&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.aiengineeringpodcast.com/openlit-open-source-llmops-episode-77&quot;&gt;https://www.aiengineeringpodcast.com/openlit-open-source-llmops-episode-77&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>可觀測性與 LLMOps</category>
  </item>
  <item>
    <title>From IBM Acquisition to AI-Native Observability | Dash0 CEO</title>
    <link>https://aiwulin.itsmygo.uk/c/304ed4417b/</link>
    <guid isPermaLink="false">aiwulin-304ed4417b</guid>
    <pubDate>Tue, 10 Feb 2026 00:00:00 +0800</pubDate>
    <dc:creator>The AI Native Dev</dc:creator>
    <description>&lt;p&gt;訪談 Dash0 CEO Mirko Novakovic，探討其公司如何基於 OpenTelemetry 標準構建 AI 原生可觀測性平台。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.buzzsprout.com/2375985/episodes/18650168-from-ibm-acquisition-to-ai-native-observability-dash0-ceo&quot;&gt;https://www.buzzsprout.com/2375985/episodes/18650168-from-ibm-acquisition-to-ai-native-observability-dash0-ceo&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>可觀測性與 LLMOps</category>
  </item>
  <item>
    <title>Watch the recordings from my Python + MCP series</title>
    <link>https://aiwulin.itsmygo.uk/c/f2bcf13935/</link>
    <guid isPermaLink="false">aiwulin-f2bcf13935</guid>
    <pubDate>Fri, 19 Dec 2025 00:00:00 +0800</pubDate>
    <dc:creator>pamela fox's blog</dc:creator>
    <description>&lt;p&gt;Pamela Fox 分享 Python 與 MCP 的三場直播串內容，涵蓋使用 FastMCP 建立伺服器、在 Azure 部署及加入認證機制。讀者可學習如何開發、部署並安全地整合 AI 代理與聊天機器人。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://blog.pamelafox.org/2025/12/watch-recordings-from-my-python-mcp.html&quot;&gt;https://blog.pamelafox.org/2025/12/watch-recordings-from-my-python-mcp.html&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>MCP 模型上下文協定</category>
  </item>
  <item>
    <title>Building the Internet of Agents: Identity, Observability, and Open Protocols</title>
    <link>https://aiwulin.itsmygo.uk/c/1b4c2db17e/</link>
    <guid isPermaLink="false">aiwulin-1b4c2db17e</guid>
    <pubDate>Mon, 10 Nov 2025 00:00:00 +0800</pubDate>
    <dc:creator>AI Engineering Podcast</dc:creator>
    <description>&lt;p&gt;探討多智慧體系統的擴展挑戰，包括身份驗證、可觀測性及開放協議。講者介紹了 Cisco Outshift 的 Agency 框架與 SLIM 通訊層，並分享 CAPE 平台工程師等實際應用案例，展示如何透過協作提升生產力。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.aiengineeringpodcast.com/outshift-multi-agent-systems-exploration-episode-68&quot;&gt;https://www.aiengineeringpodcast.com/outshift-multi-agent-systems-exploration-episode-68&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>AI Agent 基礎</category>
    <category>可觀測性與 LLMOps</category>
    <category>多代理系統</category>
  </item>
</channel>
</rss>
