[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-why-rag-in-microsoft-foundry-needs-better-indexes-zh":3,"article-related-why-rag-in-microsoft-foundry-needs-better-indexes-zh":30,"series-industry-27143bae-96b1-4a33-9906-0b546a29df2c":82},{"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":23,"views":27,"created_at":28,"published_at":29,"topic_cluster_id":11},"27143bae-96b1-4a33-9906-0b546a29df2c","why-rag-in-microsoft-foundry-needs-better-indexes-zh","為什麼 Microsoft Foundry 的 RAG 需要更好的索引，不需要…","\u003Cp data-speakable=\"summary\">\u003Ca href=\"\u002Ftag\u002Fmicrosoft\">Microsoft\u003C\u002Fa> Foundry 的 \u003Ca href=\"\u002Ftag\u002Frag\">RAG\u003C\u002Fa> 成敗關鍵在索引與檢索品質，不在把提示詞越寫越長。\u003C\u002Fp>\u003Cp>我站在這一邊：在 Microsoft Foundry 裡做 RAG，真正該投資的是索引設計與檢索品質，不是把 prompt 繼續加長。原因很直接，Foundry 的官方架構本身就把 Azure AI Search、hybrid search、agentic retrieval 和 grounding 放在核心位置，這表示系統的可信度先由資料找不找得到決定，再由模型怎麼回答決定。若檢索拿到的是錯的段落，再強的提示詞也只是把錯誤包裝得更像答案。\u003C\u002Fp>\u003Ch2>第一個論點：索引才是 RAG 的控制平面\u003C\u002Fh2>\u003Cp>Foundry 把 index 定義成讓檢索可靠的結構，這不是語義上的修辭，而是工程上的事實。當系統找不到正確片段時，模型會用看似合理的方式補完，結果就是幻覺。Microsoft 之所以同時提供 keyword、semantic、vector 與 hybrid search，就是因為「相關」不是單一標準，索引策略一變，答案是否被正確 grounding 也會跟著變。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778300458830-etqy.png\" alt=\"為什麼 Microsoft Foundry 的 RAG 需要更好的索引，不需要…\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>更關鍵的是，Foundry 建議把 Azure AI Search 當作 RAG 的 index store。這代表索引不只是存資料，而是把 title、URL、file name 這類 citation metadata 一起帶進流程，讓答案可以被追溯、被審計。換句話說，索引不只是找得到內容，它決定了內容能不能被信任。對生產系統來說，這比 prompt 裡多塞幾句「請謹慎回答」重要得多。\u003C\u002Fp>\u003Ch2>第二個論點：真正有用的是 agentic retrieval，不是單次硬塞上下文\u003C\u002Fh2>\u003Cp>傳統 RAG 常見做法是丟一個 query、抓幾個 chunk、把它們塞進 prompt，然後希望模型自己想清楚。Foundry 的 agentic retrieval 之所以更強，是因為它把檢索變成規劃問題：模型可以把複雜問題拆成子查詢，平行執行，再回傳結構化 grounding 資料。這對多輪對話特別重要，因為使用者的意圖常常不是一次就能問完整。\u003C\u002Fp>\u003Cp>官方功能列表也說明了這點：co\u003Ca href=\"\u002Fnews\u002Fhow-to-choose-third-party-ai-for-apple-intelligence-zh\">nte\u003C\u002Fa>xt-aware planning、parallel execution、semantic ranking、optional answer synth\u003Ca href=\"\u002Fnews\u002Fclaude-design-open-source-clone-github-stars-zh\">esi\u003C\u002Fa>s。這些不是包裝詞，而是直接影響延遲、覆蓋率與可追蹤性的機制。平行子查詢可以降低漏檢的機率，結構化輸出可以讓引用與 tracing 更清楚。對工程團隊來說，這比不停調整 prompt 模板更接近可維護的產品架構。\u003C\u002Fp>\u003Ch2>第三個論點：RAG 本質上是資料管線問題，不是文案問題\u003C\u002Fh2>\u003Cp>Microsoft 的實作順序其實已經把答案寫出來了：先準備資料、再切 chunk、再建 index、再接 Foundry、最後才是測試與評估。這個順序很重要，因為 chunking、embedding 品質與搜尋設定一旦出錯，模型根本沒有機會在正確證據上推理。官方也明講，資料準備不佳會直接傷害回覆品質。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778300444436-bicc.png\" alt=\"為什麼 Microsoft Foundry 的 RAG 需要更好的索引，不需要…\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>所以把心力放在「更長的 prompt」是方向錯置。prompt 無法找回沒被檢索到的段落，也無法修補錯誤的切塊策略。Foundry 的 troubleshooting 其實已經點出常見故障：相關片段不對、明明有 grounding 仍然幻覺、延遲過高、\u003Ca href=\"\u002Ftag\u002Ftoken\">token\u003C\u002Fa> 膨脹。這些都是 pipe\u003Ca href=\"\u002Fnews\u002Fchatgpt-goblin-bug-closed-models-fragile-zh\">lin\u003C\u002Fa>e defect，不是語言修飾問題。要修，得修資料路徑。\u003C\u002Fp>\u003Ch2>反方可能怎麼說\u003C\u002Fh2>\u003Cp>最強的反對意見是：RAG 本來就增加複雜度與成本。檢索會多一次往返與算力，embedding 與索引更新也有代價，檢索回來的內容還會吃掉 token。若需求只是固定風格、穩定行為，fine-tuning 可能更乾淨；若是 \u003Ca href=\"\u002Ftag\u002Fagent\">agent\u003C\u002Fa> 系統，retrieval 也許只是眾多工具之一，不該被神化成唯一答案。\u003C\u002Fp>\u003Cp>這個批評成立，但它打中的不是 index-centered RAG，而是濫用 RAG 的做法。Foundry 的邏輯很清楚：私有資料、快速變動資料、需要來源引用的答案，適合 RAG；要改的是行為模式，fine-tuning 更合適；retrieval 只是工具時，就別硬把它當整個架構。也就是說，限制要承認，但一旦你要的是 freshness、provenance 和 citations，索引仍然是核心資產。\u003C\u002Fp>\u003Cp>真正該反駁的不是「RAG 有成本」，而是「既然有成本，就用更長 prompt 補上」。這條路通常只會讓 token 更貴、延遲更高、上下文更亂，卻不會讓證據更準。當檢索本身錯了，prompt 再漂亮也只是把錯誤回答得更完整。\u003C\u002Fp>\u003Ch2>你能做什麼\u003C\u002Fh2>\u003Cp>如果你是工程師，先設計 retrieval layer，再碰 prompt：選對 index 模式，從第一天就保存 citation metadata，把 access control 放在檢索層，並用真實使用者問題測試，不要只拿合成題目驗證。如果你是 PM 或創辦人，請把 indexing、evaluation 與 security 算進產品成本，而不是當作實作細節。對 Foundry 來說，RAG 的競爭力不在 prompt 長度，而在你是否把索引當成基礎設施，把 grounding 當成產品需求。\u003C\u002Fp>","Microsoft Foundry 的 RAG 成敗關鍵在索引與檢索品質，不在把提示詞越寫越長。","learn.microsoft.com","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Ffoundry\u002Fconcepts\u002Fretrieval-augmented-generation",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1778300458830-etqy.png","industry","zh","7f641864-c532-4bca-908d-fd576ca8772f",[17,18,19,20,21,22],"Microsoft Foundry","RAG","Azure AI Search","索引設計","agentic retrieval","grounding",[24,25,26],"在 Microsoft Foundry 裡，RAG 的核心槓桿是索引與檢索，不是更長的 prompt。","Azure AI Search、hybrid search 與 agentic retrieval 讓 grounding 變成可控的系統能力。","把 RAG 當成資料管線與產品問題來做，才會得到可追溯、可維護的答案。",2,"2026-05-09T04:20:23.667583+00:00","2026-05-09T04:20:23.624+00:00",{"tags":31,"relatedLang":41,"relatedPosts":45},[32,34,36,38,40],{"name":18,"slug":33},"rag",{"name":19,"slug":35},"azure-ai-search",{"name":17,"slug":37},"microsoft-foundry",{"name":21,"slug":39},"agentic-retrieval",{"name":20,"slug":20},{"id":15,"slug":42,"title":43,"language":44},"why-rag-in-microsoft-foundry-needs-better-indexes-en","Why RAG in Microsoft Foundry needs better indexes, not bigger prompts","en",[46,52,58,64,70,76],{"id":47,"slug":48,"title":49,"cover_image":50,"image_url":50,"created_at":51,"category":13},"fd2045a8-4772-4615-981c-eabdfa7f558d","7-ways-to-install-openclaw-zh","7 種安裝 OpenClaw 的方式","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780550292997-e3fd.png","2026-06-04T05:17:41.00891+00:00",{"id":53,"slug":54,"title":55,"cover_image":56,"image_url":56,"created_at":57,"category":13},"06734645-2e2f-4903-9e47-e6ac889e34b7","game-thread-prompt-turns-nba-chatter-into-template-zh","Game-thread prompt 把聊天變模板","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780547608583-tp2j.png","2026-06-04T04:33:05.772212+00:00",{"id":59,"slug":60,"title":61,"cover_image":62,"image_url":62,"created_at":63,"category":13},"c323ffb6-20c8-468a-9d37-68e801588ee5","5-takeaways-from-spurs-vs-trail-blazers-game-5-zh","5 個 Spurs 對 Trail Blazers Game 5 重點","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780546677776-oc0j.png","2026-06-04T04:17:25.558061+00:00",{"id":65,"slug":66,"title":67,"cover_image":68,"image_url":68,"created_at":69,"category":13},"0231f359-f786-4e6c-8104-d3fae443f98b","4-chipotle-promo-details-for-members-zh","4 個 Chipotle 會員活動重點","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780540375071-5xa3.png","2026-06-04T02:32:19.54736+00:00",{"id":71,"slug":72,"title":73,"cover_image":74,"image_url":74,"created_at":75,"category":13},"39e4c1b2-4a8d-4baf-86eb-f65d4f6c3624","why-chipotle-53000-burrito-stunt-smart-brand-marketing-zh","為什麼 Chipotle 的 53,000 捲餅活動是聰明的品牌行銷","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780538579630-nkln.png","2026-06-04T02:02:28.454411+00:00",{"id":77,"slug":78,"title":79,"cover_image":80,"image_url":80,"created_at":81,"category":13},"53955aa8-9120-41c1-b342-6ca40e24b6ee","apples-gemini-deal-turns-cloud-ai-into-local-ai-zh","Apple 把雲端 AI 拆成本機 AI","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780535908899-g9ua.png","2026-06-04T01:18:03.319604+00:00",[83,88,93,98,103,108,113,118,123,128],{"id":84,"slug":85,"title":86,"created_at":87},"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":89,"slug":90,"title":91,"created_at":92},"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":94,"slug":95,"title":96,"created_at":97},"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":99,"slug":100,"title":101,"created_at":102},"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":104,"slug":105,"title":106,"created_at":107},"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":109,"slug":110,"title":111,"created_at":112},"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":114,"slug":115,"title":116,"created_at":117},"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":119,"slug":120,"title":121,"created_at":122},"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":124,"slug":125,"title":126,"created_at":127},"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":129,"slug":130,"title":131,"created_at":132},"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"]