[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-ai-should-govern-sdlc-before-code-zh":3,"article-related-ai-should-govern-sdlc-before-code-zh":30,"series-industry-99138a41-b577-49b1-8354-9d00d0a6cc1c":77},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":22,"views":26,"created_at":27,"published_at":28,"topic_cluster_id":29},"99138a41-b577-49b1-8354-9d00d0a6cc1c","ai-should-govern-sdlc-before-code-zh","AI 應先治理 SDLC，再去寫程式碼","\u003Cp data-speakable=\"summary\">AI 最有價值的地方不是先產生程式碼，而是先審核需求與評審流程，提早攔下錯誤決策。\u003C\u002Fp>\u003Cp>我站在這一邊：AI 應先治理 SDLC，再去寫程式碼。因為軟體團隊最大的損失，往往不是少寫了幾行碼，而是把錯誤需求、模糊假設與低品質評審一路傳到實作階段，最後用更多人力補洞。Uber 把 AI 放在 PRD 驗證，DoorDash 把它放進高訊號 \u003Ca href=\"\u002Fnews\u002Fqodo-2-8-multi-repo-ai-code-review-beta-zh\">code\u003C\u002Fa> review，\u003Ca href=\"\u002Ftag\u002Fcloudflare\">Cloudflare\u003C\u002Fa> 則用專門代理做分工\u003Ca href=\"\u002Fnews\u002Fcursor-ai-code-review-fading-zh\">審查\u003C\u002Fa>，這些案例指向同一件事：AI 最適合先當治理層，不是先當產碼機。\u003C\u002Fp>\u003Ch2>第一個論點\u003C\u002Fh2>\u003Cp>AI 在需求階段的價值，遠高於在產碼階段的炫技。Uber 的 PRD 驗證流程就是典型例子，\u003Ca href=\"\u002Fnews\u002Ffund-tokenization-services-onchain-rails-zh\">系統\u003C\u002Fa>先檢查清晰度、完整性與執行風險，再讓工程師動手。這不是形式主義，而是在最便宜的時點抓出最昂貴的錯誤，因為一份有缺陷的需求文件，後面通常會變成設計改一次、實作改一次、驗收再改一次。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782822770792-row9.png\" alt=\"AI 應先治理 SDLC，再去寫程式碼\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>實務上，模糊需求會把成本層層放大。產品寫得不清楚，設計會補自己的理解，工程會補自己的假設，最後 \u003Ca href=\"\u002Ftag\u002Fcode-review\">code review\u003C\u002Fa> 變成「這到底原本想做什麼」的辯論。AI 若能先標出缺失依賴、衝突條件與未定義邊界，就等於在工作流最前端截斷 rework。這種節省不一定會出現在漂亮的儀表板上，但它直接影響團隊吞吐量。\u003C\u002Fp>\u003Ch2>第二個論點\u003C\u002Fh2>\u003Cp>在 review 階段，AI 的核心不是產生更多評論，而是產生更可信的評論。DoorDash 的內部 reviewer 之所以值得參考，不在於它留言多，而在於它刻意追求高訊號、低噪音。這很重要，因為 code review 最怕的不是漏掉一兩個問題，而是工具丟出一堆泛泛警告，最後工程師把它當背景噪音。\u003C\u002Fp>\u003Cp>好的 AI reviewer 應該像成熟的同事，而不是急著表現的實習生。它要能指出具體問題、嵌入既有流程、並且尊重人工決策權。DoorDash 的做法說明一個很現實的標準：如果工具真的改變了上線前的行為，它就有價值；如果它只是增加 comment 數量，那只是把噪音包裝成生產力。\u003C\u002Fp>\u003Ch2>反方可能怎麼說\u003C\u002Fh2>\u003Cp>反方會說，AI 最該先進入的地方是 code generation，因為那裡有最直接的產出；把它放到 PRD 和 review，反而像是多加一層機器關卡。很多團隊本來就已經被流程拖慢，再多一個 AI 審核，會不會只是把開發節奏變得更官僚、更難啟動？\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782822768930-ywiu.png\" alt=\"AI 應先治理 SDLC，再去寫程式碼\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>這個疑慮不是空穴來風。模型會看漏上下文、誤判內部術語，也可能用很像樣的語氣講錯誤的建議。若團隊把 AI 輸出當成權威，就會出現 approval theater：看起來很嚴謹，實際上只是把薄弱需求和粗糙 review 重新包裝。\u003C\u002Fp>\u003Cp>但這個批評只能限制 AI 的角色，不能推翻它應該前移的結論。AI 不該成為額外的批准者，而該是第一道過濾器，先攔下明顯缺陷，再把人的注意力留給真正需要判斷的地方。人仍然要負責決策，AI 則負責把最浪費時間的錯誤提早暴露出來。\u003C\u002Fp>\u003Ch2>你能做什麼\u003C\u002Fh2>\u003Cp>如果你是工程師、PM 或創辦人，別先問 AI 能不能多寫 code，先問它能在哪裡減少返工。把它放進 PRD、設計說明與 review 流程，觀察它是否真的降低歧義、縮短審查時間、提升決策品質。真正值得押注的不是產碼速度，而是治理能力：先用 AI 把需求和審查變乾淨，再讓工程師把時間花在真正的實作上。\u003C\u002Fp>","AI 最有價值的地方不是先產生程式碼，而是先審核需求與評審流程，提早攔下錯誤決策。","www.infoq.com","https:\u002F\u002Fwww.infoq.com\u002Fnews\u002F2026\u002F06\u002Fai-prd-code-review-governance\u002F",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782822770792-row9.png","industry","zh","bc3a5cde-08c3-45db-89c7-1ca4419c1d4e",[17,18,19,20,21],"AI","SDLC","PRD","code review","software governance",[23,24,25],"AI 最適合先用在需求與審查治理，而不是先做程式碼生成。","前移到 PRD 與 review，可更早攔下模糊需求與高成本返工。","AI 應扮演高訊號的治理層，保留人類對決策的最終責任。",0,"2026-06-30T12:32:21.404276+00:00","2026-06-30T12:32:21.393+00:00","7aa69b8b-ff49-4d68-9e8b-f08e577b1239",{"tags":31,"relatedLang":36,"relatedPosts":40},[32,34],{"name":20,"slug":33},"code-review",{"name":17,"slug":35},"ai",{"id":15,"slug":37,"title":38,"language":39},"ai-should-govern-sdlc-before-code-en","AI should govern the SDLC before it writes code","en",[41,47,53,59,65,71],{"id":42,"slug":43,"title":44,"cover_image":45,"image_url":45,"created_at":46,"category":13},"99d12162-7c57-4d83-a454-374e8c832c01","ai-infrastructure-trillion-dollar-asset-class-zh","AI 基礎設施正變成兆級資產","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782826367359-epo1.png","2026-06-30T13:32:23.274067+00:00",{"id":48,"slug":49,"title":50,"cover_image":51,"image_url":51,"created_at":52,"category":13},"0cb511d2-4fa0-496b-b305-f4095621e183","ai-demand-starts-paying-for-data-centers-zh","AI收入開始買單資料中心","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782825476978-jtjs.png","2026-06-30T13:17:34.381956+00:00",{"id":54,"slug":55,"title":56,"cover_image":57,"image_url":57,"created_at":58,"category":13},"d440d000-7e13-4147-9cbb-72d420adcc97","rwa-tokenization-new-default-ownership-zh","RWA 代幣化正成為所有權的新預設","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782819167113-ebue.png","2026-06-30T11:32:21.420997+00:00",{"id":60,"slug":61,"title":62,"cover_image":63,"image_url":63,"created_at":64,"category":13},"ca7ebc2a-cad0-4ec5-b070-ff03a11a30a4","2026-fangdiancai-daibi-hua-zui-qiang-5-pingtai-zh","2026 房地產代幣化最強 5 平台","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782818267912-53ro.png","2026-06-30T11:17:26.082539+00:00",{"id":66,"slug":67,"title":68,"cover_image":69,"image_url":69,"created_at":70,"category":13},"3ccd4bc9-8220-44ca-8792-8cacd6b3c714","docker-monitoring-tools-real-budgets-zh","12 款 Docker 監控工具，按預算挑最省","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782806595915-i55c.png","2026-06-30T08:02:41.691145+00:00",{"id":72,"slug":73,"title":74,"cover_image":75,"image_url":75,"created_at":76,"category":13},"5fc04851-4033-4690-b0b9-f2c2d2dcfa4d","base-sequencer-outage-single-sequencer-fragile-zh","Base 的 sequencer 當機證明單一排序器太脆弱","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782804768122-faii.png","2026-06-30T07:32:20.341651+00:00",[78,83,88,93,98,103,108,113,118,123],{"id":79,"slug":80,"title":81,"created_at":82},"ee073da7-28b3-4752-a319-5a501459fb87","ai-in-2026-what-actually-matters-now-zh","2026 AI 真正重要的事","2026-03-26T07:09:12.008134+00:00",{"id":84,"slug":85,"title":86,"created_at":87},"83bd1795-8548-44c9-9a7e-de50a0923f71","trump-ai-framework-power-speech-state-preemption-zh","川普 AI 框架瞄準電力、言論與州權","2026-03-26T07:12:18.695466+00:00",{"id":89,"slug":90,"title":91,"created_at":92},"ea6be18b-c903-4e54-97b7-5f7447a612e0","nvidia-gtc-2026-big-ai-announcements-zh","NVIDIA GTC 2026 重點拆解","2026-03-26T07:14:26.62638+00:00",{"id":94,"slug":95,"title":96,"created_at":97},"4bcec76f-4c36-4daa-909f-54cd702f7c93","claude-users-spreading-out-and-getting-better-zh","Claude 用戶更分散，也更會用","2026-03-26T07:22:52.325888+00:00",{"id":99,"slug":100,"title":101,"created_at":102},"bd903b15-2473-4178-9789-b7557816e535","openclaw-raises-hard-question-for-ai-models-zh","OpenClaw 逼問 AI 模型價值","2026-03-26T07:24:54.707486+00:00",{"id":104,"slug":105,"title":106,"created_at":107},"eeac6b9e-ad9d-4831-8eec-8bba3f9bca6a","gap-google-gemini-checkout-fashion-search-zh","Gap 把結帳搬進 Gemini","2026-03-26T07:28:23.937768+00:00",{"id":109,"slug":110,"title":111,"created_at":112},"0740e53f-605d-4d57-8601-c10beb126f3c","google-pushes-gemini-transition-to-march-2026-zh","Google 把 Gemini 轉換延到 2026 年 3…","2026-03-26T07:30:12.825269+00:00",{"id":114,"slug":115,"title":116,"created_at":117},"e660d801-2421-4529-8fa9-86b82b066990","metas-llama-4-benchmark-scandal-gets-worse-zh","Meta Llama 4 分數風波又擴大","2026-03-26T07:34:21.156421+00:00",{"id":119,"slug":120,"title":121,"created_at":122},"183f9e7c-e143-40bb-a6d5-67ba84a3a8bc","accenture-mistral-ai-sovereign-enterprise-deal-zh","Accenture 攜手 Mistral AI 賣主權 AI","2026-03-26T07:38:14.818906+00:00",{"id":124,"slug":125,"title":126,"created_at":127},"191d9b1b-768a-478c-978c-dd7431a38149","mistral-ai-faces-its-hardest-year-yet-zh","Mistral AI 迎來最硬的一年","2026-03-26T07:40:23.716374+00:00"]