(圖說:AWS Summit Hong Kong 2025 Dev Lounge。圖片來源:Dorothy。)
緣起
很榮幸收到投稿邀請被選上。非常感謝大家都在默默地督促著我前進 :) 另一個榮幸是躬逢其盛,遇上 AWS Summit Hong Kong 十週年,回想從 2014 年與 Lenie 與 Locarno 一起邀請 Jeff Barr 來台灣給 COSCUP 一場主題演講 (keynote)、一路到鼓勵參加 AWS Hero program、再到今天有這個機會來到香港在 AWS Summit Hong Kong 的 Dev Lounge 跟一票開發者與技術管理者一起交流,真心覺得緣分與迭代進步的幸運交叉,心中充滿各種感激。
分享的內容來自於近一年多來我們與傳統產業互動、梳理流程、物件狀態化的經驗,以及我們自身產品與技術整合團隊 (PTI) 的工作流程與 AI 工具協作的調整(踩坑之旅)。搭配 Amazon Q CLI 以及從現有流程來延伸出 AI 協作的思維,邀請 AI 成爲我們團隊的新成員,練習分派任務給 AI 做,跟所有新同事一樣,都會有磨合期,但是只要一天不邀請這位(這群?!)新同事,那磨合就不會開始。
(Photo credits to Dorothy)
簡報下載
大綱
- About Me
- Agenda
- 📋 1 AI Adoption Landscape - The AI Transformation
- 🛠 2 DLC Through AI Lens - Reimagining Developer Workflows
- 👨💻 3 Humbled Transformation Framework - People, Workflow, System
- 📋 AI: What & Why?
- AI Usage in Organizations
- Investment Trends
- Workflow Integration
- McKinsey 和 Accenture 在 2025 年初發布的報告裡頭帶出以下幾個數字:
- 100%: 我們在座每一個人
- 90%: 拿到或增加 Gen AI 資源或預算
- 80%: 使用 AI 在至少一個功能單位
- 40%: 導入 AI 時遇到阻礙
- 20%: 重新設計工作流程以導入 Gen AI (這裡應該是重點!)
- 1%: 組織覺得有達到完整的 AI 成熟度
- 🛠 Developer Workflows
- Quick SDLC Recap
- Plan & Design
- Implement & Test
- Deploy & Maintain
- Amazon Q CLI Demo 1: Vibe Coding
- Amazon Q CLI Demo 2: Architecture Analysis
- Demo 2: Generated Architecture Diagram
- Amazon Q CLI
- SDLC Task Scenarios & AI Tool Mapping Examples
- Quick SDLC Recap
- 👨💻 Rethinking Your AI-Powered Framework
- AI == Software, in a System
- AI == Software
- Systems Solve Problems
- LLMs Are Only ~15%
- Average Model + Awesome Workflow
- Three Pillars of Humbled Transformation Framework
- People - 以人為本、解決問題、發展策略、當責團隊
- Workflow - 工作日常、探究細節、拆解步驟、基於事實
- System - 退後一步、綜觀全局、整合流程、創造價值
- 這只是個範例模板框架,你可以依照你的場景、你的經驗,修改成你覺得舒服的模樣。
- Systems Thinking & Strategy
- Feedback Loop
- Q&A
- Thank You & Stay Connected
- Reference & Materials
- Do / Delegate / Develop Framework
- RAG & Knowledge
- RAG & Knowledge Integration: Core Concepts
- RAG Architecture Flow
- RAG Performance Metrics (Example)
- CI/CD & Observability
- CI/CD & Observability: Core Concepts
- CI/CD Architecture Flow
- CI/CD & Observability Metrics (Example)
- Business Impact (Example)
- 4-Week Practical Getting Started Plan
延伸閱讀
- Innovating at speed: BMW’s generative AI solution for cloud incident analysis | AWS Machine Learning Blog #BMW
- Extend the Amazon Q Developer CLI with Model Context Protocol (MCP) for Richer Context | AWS DevOps & Developer Productivity Blog #MCP
- Ernest PKM Personal Knowledge System Workflow (2023.25) | Ernest Chiang #PKM
- AWS Community Day 2024: Using AWS to build a launchable knowledge rocket 👉 Organize knowledge, accelerate learning and understand AI in the process - Speaker Deck
- Ernest Chiang (@dwchiang) on Speaker Deck