Agent 與自動化

模組詳細介紹

目標：本模組旨在讓學員從傳統的「跟 AI 對話」模式，進化到「派遣 AI 工作」的更高層次。學員將學習如何運用 AI Agent（如 Claude Cowork、GPT-CodeX）實現桌面自動化、遠端操控，以及如何透過外掛與資料庫整合，讓 AI 真正成為你工作的得力分身。

課程承諾：你將學會如何設計並部署你的 AI Agent，讓它能夠自動執行重複性任務、管理檔案、甚至在背景進行複雜操作，大幅解放你的時間與精力。

反差敘事：當別人還在親自操作電腦時，你已能透過手機指令讓 AI 遠端控制你的電腦完成任務；當別人還在為昂貴的 AI 工具猶豫時，你已找到高性價比的替代方案並熟練應用。你將成為 AI Agent 的指揮家，而非被動的用戶。

教學風格定位：本課程重實作，並注入思維深度，不僅教授工具操作，更傳遞「為什麼」要這樣做的底層邏輯。

作業：用手機 Dispatch 出一張考卷／報告／投影片到自己電腦。

為什麼是這個順序？

本模組從 Claude 的桌面自動化功能（Cowork）和遠端操控能力（Dispatch）開始，讓學員直觀感受 Agent 的便捷性。接著，引入 GPT-CodeX 作為另一選擇，並進行功能與成本效益的比較，拓寬學員的工具視野。隨後，深入介紹 Codex 的外掛、與 Obsidian 的連接以及資料庫實戰，逐步建構更完整的 Agent 工作流。這樣的順序設計，旨在讓學員從單點應用到全面集成，掌握 AI Agent 的核心能力。

教學設計重點

1. 「金句驅動」開場法：每個章節都會使用一句從精華影片中摘取的金句作為開場引導，迅速抓住學員注意力。
2. 「原作者式」的反差教學：課程設計借鑒了「原作者」頻道影片的標題模式，透過反差敘事（如「別人在…你卻在…」）來凸顯學習價值與差異性。
3. 教學節奏：注重示範、學員實際操作、再進行改進的循環節奏，確保學員能真正動手實踐，而非僅限於觀看演示。

---en---

Module Introduction

Goal: This module aims to elevate learners from the traditional “chatting with AI” model to the more advanced level of “dispatching AI tasks.” Learners will acquire skills in using AI Agents (such as Claude Cowork, GPT-CodeX) to achieve desktop automation, remote control, and integrate plugins and databases, thereby enabling AI to truly become an effective extension of their work.

Course Commitment: You will learn how to design and deploy your AI Agent, empowering it to automatically perform repetitive tasks, manage files, and even execute complex operations in the background, significantly freeing up your time and energy.

Contrastive Narrative: While others are still physically operating computers, you will be able to remotely control your computer with AI via mobile commands to complete tasks. While others are hesitant about expensive AI tools, you will have found cost-effective alternatives and applied them proficiently. You will become the conductor of AI Agents, not just a passive user.

Teaching Style Positioning: This course emphasizes hands-on practice while instilling deep analytical thinking. It goes beyond mere tool operation to convey the “why” behind effective AI usage.

Assignment: Use your mobile phone to dispatch a test paper, report, or presentation to your computer.

Why This Order?

This module begins with Claude’s desktop automation (Cowork) and remote control (Dispatch) capabilities, allowing learners to intuitively experience the convenience of Agents. Subsequently, GPT-CodeX is introduced as an alternative, with a comparison of its functionalities and cost-effectiveness, broadening learners’ tool perspectives. Following this, an in-depth introduction to Codex’s plugins, integration with Obsidian, and practical database applications gradually builds a more comprehensive Agent workflow. This sequential design aims to guide learners from single-point applications to comprehensive integration, mastering the core capabilities of AI Agents.

Key Teaching Design Points

1. “Quote-Driven” Opening: Each chapter uses a powerful quote extracted from highlight videos to quickly capture learner attention.
2. “Original Creator-Style” Contrastive Teaching: The course design draws inspiration from the “Original Creator” channel’s video title patterns, using contrastive narratives (e.g., “While others are… you are…”) to emphasize learning value and differentiation.
3. Teaching Rhythm: Focuses on a cyclical rhythm of demonstration, hands-on learner practice, and refinement, ensuring learners actively engage in practical application rather than just watching demonstrations.