Rag-Ready Patterns for Data Platforms: Making Systems Ready for Grounded Generative AI
暫譯: 資料平台的 RAG 就緒模式:打造支援基於實證生成式 AI 的系統

Vedula, Ravi, Martinez, Gerardo Bodegas, Chittajallu, Maruti

  • 出版商: O'Reilly
  • 出版日期: 2026-12-15
  • 售價: $2,820
  • 貴賓價: 9.5 折 $2,679
  • 語言: 英文
  • 頁數: 360
  • 裝訂: Quality Paper - also called trade paper
  • ISBN: 9798341670846
  • ISBN-13: 9798341670846
  • 相關分類: Large language model、大數據 Big-data
  • 海外代購書籍(需單獨結帳)

相關主題

商品描述

Enterprises are under pressure to apply generative AI to their data, but most data platforms weren't designed to support retrieval-augmented generation (RAG)-based conversational systems. Built for dashboards and siloed reporting, they often lack context, structure, and governance that AI applications require. RAG-Ready Patterns for Data Platforms provides a practical blueprint for closing that gap, showing how to evolve fragmented data estates into AI-ready foundations delivering trusted insights.

Drawing on real-world experience building one of Microsoft's largest data platforms, the authors present proven patterns for unifying data across silos, capturing metadata, and embedding governance. Rather than treating RAG as an application-level concern, this book reframes it as a platform capability. With clear explanations, architectural diagrams, and reusable playbooks, you'll learn to design platforms that enable developers to build scalable, retrieval-augmented AI systems grounded in enterprise data.

  • Assess whether your data platform is ready for RAG-based AI applications
  • Apply patterns like unified data models and metadata-first pipelines
  • Use data quality signals and curated grounding to improve AI reliability
  • Embed governance and access control into platform design
  • Enable developers to build trustworthy copilots, chatbots, and agents

企業正承受將生成式 AI 應用於自身資料的壓力,但多數資料平台的設計初衷,並不是支援以檢索增強生成(retrieval-augmented generation,RAG)為基礎的對話系統。這些平台主要用於儀表板和各自孤立的報表,往往缺乏 AI 應用所需的脈絡、結構與治理能力。《RAG-Ready Patterns for Data Platforms》提供了填補這項落差的實用藍圖,說明如何將零散的資料環境逐步轉型為 AI 就緒的基礎,提供值得信賴的洞察。

作者以建置 Microsoft 最大型資料平台之一的實務經驗為基礎,提出經驗驗證的模式,協助讀者整合跨越各個資料孤島的資料、擷取中繼資料,並將治理機制嵌入平台。這本書不把 RAG 視為應用程式層級的議題,而是重新定義為平台能力。透過清楚的說明、架構圖與可重複使用的實作手冊,您將學會設計能讓開發人員建置可擴充、以企業資料為基礎且具檢索增強能力之 AI 系統的平台。

• 評估您的資料平台是否已準備好支援以 RAG 為基礎的 AI 應用程式
• 套用統一資料模型、以中繼資料為先的資料管線等模式
• 運用資料品質訊號與經策展的 grounding,提升 AI 的可靠性
• 將治理與存取控制嵌入平台設計
• 協助開發人員建置值得信賴的 copilot、聊天機器人與 agent

以上中文內容由 AI 翻譯