[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"portal-settings:stajic:zh":3,"public-menus:all":38,"$fwk2hhhfhd3a":265,"related:portfolio:source-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value:zh:1":1452},{"statusCode":4,"data":5,"message":37},200,{"tenantId":6,"lang":7,"defaultLang":8,"siteUrl":9,"contactEmail":10,"brandName":11,"logoUrl":12,"siteName":11,"siteDescription":13,"ogImage":10,"robotsIndex":14,"socialLinks":10,"reservedSlugs":10,"seoPolicy":15},"stajic","zh","de","https:\u002F\u002Fstajic.de",null,"Stajic Platform","\u002FLogo_Planet.svg","Stajic Portal",true,{"branding":16,"relatedContent":17,"crossDomainLinks":18},{"logoUrl":12},{"enabled":14},[19,22,25,28,31,34],{"url":20,"label":21,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Ffigure.rocks","figure.rocks",{"url":23,"label":24,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Floving.rocks","loving.rocks",{"url":26,"label":27,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.com","bazify.com",{"url":29,"label":30,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.de","bazify.de",{"url":32,"label":33,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.at","bazify.at",{"url":35,"label":36,"isActive":14,"showInFooter":14,"includeInSameAs":14},"https:\u002F\u002Fbazify.ba","bazify.ba","Portal settings resolved",[39,45],{"id":40,"name":41,"location":42,"isActive":14,"isDefault":43,"items":44},1,"main-navigation","header",false,[],{"id":46,"name":47,"location":48,"isActive":14,"isDefault":14,"items":49},4,"main-menu","sidebar",[50,65,96,112,125,139,149,164,179],{"id":51,"title":52,"url":60,"target":61,"icon":62,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":10,"children":64},"menu-0d07dc94-51c7-48ba-83d8-b994a7cd1e11",{"de":53,"en":54,"es":55,"fr":56,"it":54,"ru":57,"sr":58,"zh":59},"Startseite","Home","Inicio","Accueil","Главная","Početna","首页","\u002F","_self","i-lucide-house","custom",[],{"id":66,"title":67,"url":76,"target":61,"icon":77,"isActive":14,"type":78,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":79,"children":80},"menu-73bf1e4c-e499-41a0-bc9d-c24546b75658",{"de":68,"en":69,"es":70,"fr":71,"it":72,"ru":73,"sr":74,"zh":75},"Source-of-Truth-Plattform","Source of Truth Platform","Plataforma de fuente de verdad","Plateforme source de vérité","Piattaforma Fonte di Verità","Платформа источника истины","Platforma izvora istine","事实来源平台","\u002Fportfolio\u002Fsource-of-truth-enterprise-intelligence-platform","i-lucide-bot","portfolio",23,[81],{"id":82,"title":83,"url":92,"target":61,"icon":93,"isActive":14,"type":78,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":94,"children":95},"menu-8ce035b5-ba9f-4985-a0f7-e168a4e2ea8e",{"de":84,"en":85,"es":86,"fr":87,"it":88,"ru":89,"sr":90,"zh":91},"Source of Truth jenseits von RAG","Source of Truth Beyond RAG","Fuente de verdad más allá de RAG","Source de vérité au-delà du RAG","Fonte di verità oltre RAG","Источник истины за пределами RAG","Izvor istine izvan RAG-a","超越 RAG 的事实来源","\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","i-lucide-route",24,[],{"id":97,"title":98,"url":107,"target":61,"icon":108,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":110,"portfolioId":10,"children":111},"menu-bc3968b2-907c-4851-b8a8-2f273aa53c20",{"de":99,"en":100,"es":101,"fr":102,"it":103,"ru":104,"sr":105,"zh":106},"Produkt- & Plattformarchitektur","Product & Platform Architecture","Arquitectura de producto & plataforma","Architecture produit & plateforme","Architettura di prodotto & piattaforma","Архитектура продукта & платформы","Arhitektura proizvoda & platforme","产品与平台架构","\u002Ftechnical-product-platform-architect-munich-ai-software-delivery","i-lucide-boxes","page",111,[],{"id":113,"title":114,"url":121,"target":61,"icon":122,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":123,"portfolioId":10,"children":124},"item-22",{"de":115,"en":115,"es":116,"fr":115,"it":117,"ru":118,"sr":119,"zh":120},"Vision","Visión","Visione","Видение","Vizija","想象","\u002Fueber-uns-webdesign-muenchen-webaplikation","i-lucide-eye",113,[],{"id":126,"title":127,"url":135,"target":61,"icon":136,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":137,"portfolioId":10,"children":138},"item-19",{"de":128,"en":129,"es":130,"fr":129,"it":131,"ru":132,"sr":133,"zh":134},"Leistungen","Services","Servicios","Servizi","Услуги","Proizvod, platforma i AI usluge","服务","\u002Fservices-dienstleistungen-muenchen","i-lucide-wrench",116,[],{"id":140,"title":141,"url":145,"target":61,"icon":146,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":147,"portfolioId":10,"children":148},"item-23",{"de":142,"en":142,"es":142,"fr":142,"it":142,"ru":143,"sr":142,"zh":144},"Blog","Блог","博客","\u002Fblog","i-lucide-book-open",112,[],{"id":150,"title":151,"url":160,"target":61,"icon":161,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":162,"portfolioId":10,"children":163},"item-32",{"de":152,"en":153,"es":154,"fr":155,"it":156,"ru":157,"sr":158,"zh":159},"Neue Technologien","New Technologies","Nuevas tecnologías","Nouvelles technologies","Nuove tecnologie","Новые технологии","Nove tehnologije!","新技术！","\u002Fneue-webtechnologien","i-lucide-sparkles",122,[],{"id":165,"title":166,"url":175,"target":61,"icon":176,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":177,"portfolioId":10,"children":178},"item-20",{"de":167,"en":168,"es":169,"fr":170,"it":171,"ru":172,"sr":173,"zh":174},"Kontakt","Contact us!","Contacto","Contact","Contatto","Контакт","Kontaktirajte nas","联系我们！","\u002Fcontact","i-lucide-mail",115,[],{"id":180,"title":181,"url":190,"target":61,"icon":191,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":192,"portfolioId":10,"children":193},"item-21",{"de":182,"en":183,"es":184,"fr":185,"it":186,"ru":187,"sr":188,"zh":189},"Unsere Arbeit","Our Work","Nuestro trabajo","Nos réalisations","I nostri lavori","Наши работы","Portfolio","文件夹","\u002Fportfolio","i-lucide-briefcase",114,[194,207,220,235,250],{"id":195,"title":196,"url":190,"target":61,"icon":205,"isActive":14,"type":109,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":192,"portfolioId":10,"children":206},"item-24",{"de":197,"en":198,"es":199,"fr":200,"it":201,"ru":202,"sr":203,"zh":204},"Alle Projekte","All Projects","Todos los proyectos","Tous les projets","Tutti i progetti","Все проекты","Svi projekti","所有项目","i-lucide-grid-3x3",[],{"id":208,"title":209,"url":217,"target":61,"icon":218,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":10,"children":219},"item-29",{"de":210,"en":211,"es":212,"fr":213,"it":214,"ru":215,"sr":216,"zh":189},"Local Roots, Global Reach","Local Roots - Global Reach","Empresa local ","Entreprise locale","Azienda locale","Местная компания","Lokalno preduzeće na globalnom tržištu","\u002Fportfolio\u002Flocal-roots-global-reach-communication-media-systems-for-modern-business","i-lucide-globe-2",[],{"id":221,"title":222,"url":231,"target":61,"icon":232,"isActive":14,"type":78,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":233,"children":234},"menu-2916d060-8681-482a-af9a-c107a8bc30d8",{"de":223,"en":224,"es":225,"fr":226,"it":227,"ru":228,"sr":229,"zh":230},"Von Rohdaten zu verifizierbarem KI-Kontext: Eine praktische Architektur für agentenbereite Informationen","From Raw Data to Verifiable AI Context:","De datos sin procesar a contexto de IA verificable: una arquitectura práctica para información lista para agentes","Des données brutes au contexte d'IA vérifiable : une architecture pratique pour des informations prêtes pour les agents","Dai dati grezzi a un contesto IA verificabile: Un'architettura pratica per informazioni pronte per gli agenti","От необработанных данных к верифицируемому контексту ИИ: практическая архитектура информации, готовой для агентов","Praktična arhitektura za AI agente","从原始数据到可验证的 AI 上下文：面向智能体就绪信息的实用架构","\u002Fportfolio\u002Ffrom-raw-data-to-verifiable-ai-context-a-practical-architecture-for-agent-ready-information","i-lucide-database",22,[],{"id":236,"title":237,"url":246,"target":61,"icon":247,"isActive":14,"type":78,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":248,"children":249},"menu-e2061cd7-ff53-49d5-a0e8-3d49090577c0",{"de":238,"en":239,"es":240,"fr":241,"it":242,"ru":243,"sr":244,"zh":245},"Präsenzplattformen — Unternehmenskommunikation","Presence Platforms — Enterprise Communication","Plataformas de presencia — Comunicación empresarial","Plateformes de présence — Communication d'entreprise","Presence Platforms — Comunicazione aziendale","Платформы присутствия — Корпоративные коммуникации","Presence platforme — Poslovna komunikacija","Presence 平台 — 企业通信","\u002Fportfolio\u002Fpresence-platforms-enterprise-communication-that-wins-without-competing","i-lucide-badge-percent",21,[],{"id":251,"title":252,"url":261,"target":61,"icon":262,"isActive":14,"type":78,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":263,"children":264},"menu-a781582e-e90d-418e-b8c9-b070bc37c37e",{"de":253,"en":254,"es":255,"fr":256,"it":257,"ru":258,"sr":259,"zh":260},"Maximierung des POS-ROI durch Konsistenz","Maximizing POS ROI Through Consistent","Maximizar el ROI del POS a través de constante","Maximiser le ROI du PDV grâce à la cohérence","Massimizzare il ROI del POS attraverso costante","Максимизация ROI POS за счет последовательного","Od vizije do vrednosti","通过持续最大化 POS 投资回报率","\u002Fportfolio\u002Ffrom-vision-to-value-maximizing-pos-roi-through-consistent-ci-creation-and-strategic-branding","i-lucide-briefcase-business",11,[],{"statusCode":4,"data":266,"message":1451},{"id":267,"title":268,"slug":269,"content":270,"contentJson":271,"excerpt":791,"featuredImage":792,"featuredImageAlt":793,"featuredImageCaption":10,"featuredImageTitle":10,"featuredImageCopyright":10,"featuredImageAuthor":10,"featuredImageSourceUrl":10,"featuredImageLicense":10,"featuredImageIsAiGenerated":43,"status":794,"publishedAt":795,"createdAt":796,"updatedAt":797,"seoLocalePaths":798,"author":806,"translations":811},"24","超越RAG的真相来源：证据治理的研究、可追溯性与商业价值","source-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","\u003Cp>\u003Cstrong>Source of Truth（SoT）是一个由证据治理的决策智能平台，旨在将分布式企业知识转化为可追溯、可审查且可复用的结论。\u003C\u002Fstrong>它结合了受控的源数据摄取、检索、研究代理、证据关系、来源追溯、矛盾处理、推理、访问控制和人工审查，使答案不仅被生成，而且由可见的决策路径所支撑。\u003C\u002Fp>\n\u003Cp>核心价值不在于更好的聊天机器人。价值在于一条受治理的路径：\u003Cstrong>问题 → 证据 → 关系 → 结论 → 审查\u003C\u002Fstrong>。相关信息可以被快速找到，但每一个重要陈述也都能与其来源、上下文、竞争性证据和审查状态保持关联。\u003C\u002Fp>\n\u003Cblockquote class=\"border-l-4 border-gray-300 pl-4 italic\">模型可以协助研究和推理。证据结构使组织保持控制。\u003Ccite class=\"block mt-2 text-sm\">— Source of Truth 设计原则\u003C\u002Fcite>\u003C\u002Fblockquote>\n\u003Cnav class=\"editorjs-toc\" data-editorjs-toc=\"true\" aria-label=\"目录\">\u003Cstrong class=\"editorjs-toc__title\">目录\u003C\u002Fstrong>\u003Col class=\"editorjs-toc__list editorjs-toc__list--depth-0\">\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#organizational-value\" class=\"editorjs-toc__link\">Source of Truth 为组织创造的价值\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#dag-vs-rag\" class=\"editorjs-toc__link\">DAG 是结构核心——RAG 是一个检索层\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#evidence\" class=\"editorjs-toc__link\">证据不仅仅是引用\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#research-agents\" class=\"editorjs-toc__link\">研究代理扩展调查范围，而非扩展权限\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#contradictions\" class=\"editorjs-toc__link\">矛盾、替代方案和未决点保持可见\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#validity-boundary\" class=\"editorjs-toc__link\">每个结论都有证据边界\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#governance\" class=\"editorjs-toc__link\">RBAC、命名空间和案例范围将研究限制在组织边界内\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#model-independence\" class=\"editorjs-toc__link\">模型可以更换，而不会成为知识系统\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#human-review\" class=\"editorjs-toc__link\">人工审查仍是最终权限\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#use-cases\" class=\"editorjs-toc__link\">实际价值所在\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#what-it-is\" class=\"editorjs-toc__link\">SoT 是什么——以及不是什么\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#implementation\" class=\"editorjs-toc__link\">当前实现方向\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#faq\" class=\"editorjs-toc__link\">常见问题\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#key-terms\" class=\"editorjs-toc__link\">关键术语\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#core-advantage\" class=\"editorjs-toc__link\">核心优势：决策可追溯至其证据\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>\u003C\u002Fnav>\n\u003Ch2 id=\"organizational-value\">Source of Truth 为组织创造的价值\u003C\u002Fh2>\n\u003Cp>组织很少缺乏信息。更大的挑战在于信息分散在文档、数据库、报告、团队和系统中，而重要决策需要的不仅仅是单一的搜索结果。SoT 创建了一个受控的智能层，在其中相关证据可以被定位、关联、评估和保存。\u003C\u002Fp>\n\u003Caside class=\"editorjs-callout editorjs-callout--success my-6 rounded-xl border p-5 border-emerald-300 bg-emerald-50 dark:border-emerald-900 dark:bg-emerald-950\u002F20\" role=\"note\">\u003Cstrong class=\"block mb-2 text-gray-900 dark:text-gray-100\">商业价值\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">Source of Truth 减少了重复的人工研究，保留了重要结论背后的证据，揭示了不确定性和矛盾，支持受控的 AI 使用，并将经过审查的发现转化为可复用的组织知识。\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Cdiv class=\"overflow-x-auto\">\u003Ctable class=\"w-full border-collapse\">\u003Cthead>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">需求\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">SoT 能力\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">实际价值\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">查找相关知识\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">已批准的索引源和检索\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">减少在孤立系统间搜索的时间\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">理解结论是如何得出的\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">证据关系和来源追溯\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">从主张回溯到来源的可见路径\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">处理冲突信息\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">矛盾和替代证据追踪\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">差异保持可见，而非被静默合并\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">处理不完整的证据\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">未决点和有效性边界\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">不确定性变得明确且可审查\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">跨团队安全使用 AI\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">RBAC、命名空间和有界工作流\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">AI 访问遵循组织权限和上下文\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">复用先前的研究\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">经过审查的知识和持久证据结构\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">重要发现不会消失在一次性对话中\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">更换模型而不丢失知识\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">模型无关的证据层\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">知识结构保持稳定，而 AI 组件持续演进\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"dag-vs-rag\">DAG 是结构核心——RAG 是一个检索层\u003C\u002Fh2>\n\u003Cp>RAG 有助于理解系统的一部分，因为它检索相关源材料并将其作为上下文提供给 AI 模型。然而在 SoT 中，更重要的结构是 \u003Cstrong>DAG——有向无环图\u003C\u002Fstrong>。DAG 表示证据、主张、中间发现和结论如何相互依赖，而不产生循环论证。\u003C\u002Fp>\n\u003Cp>RAG 管道回答：\u003Cem>哪些段落与此查询相关？\u003C\u002Fem>SoT 证据 DAG 回答更广泛的问题：\u003Cem>哪些主张依赖于哪些证据，哪些结论依赖于哪些主张，替代方案或矛盾存在于何处，以及推理路径如何向后追溯？\u003C\u002Fem>\u003C\u002Fp>\n\u003Csection class=\"editorjs-comparison my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">RAG 和 SoT 证据 DAG 解决的是问题的不同部分\u003C\u002Fh3>\u003Cdiv class=\"overflow-x-auto\">\u003Ctable class=\"w-full border-collapse\">\u003Cthead>\u003Ctr>\u003Cth class=\"border border-gray-300 bg-gray-50 px-4 py-3 text-left dark:border-gray-700 dark:bg-gray-900\">\u003C\u002Fth>\u003Cth class=\"border border-gray-300 bg-gray-50 px-4 py-3 text-left font-semibold dark:border-gray-700 dark:bg-gray-900\">RAG\u003C\u002Fth>\u003Cth class=\"border border-gray-300 bg-gray-50 px-4 py-3 text-left font-semibold dark:border-gray-700 dark:bg-gray-900\">SoT 中的证据 DAG\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">主要目的\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">为查询检索相关上下文\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">表示证据、主张、发现和结论之间的依赖关系\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">主要单元\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">检索到的段落或文档\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">节点和有向证据关系\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">可追溯性\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">可以引用检索到的来源\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">可以通过支撑链追溯结论\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">矛盾\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">可能检索到相互冲突的段落\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">可以将冲突保留为显式关系或替代路径\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">推理结构\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">通常在查询时组装\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">作为显式依赖结构持久化\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">复用\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">每次查询都重复检索\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">经过审查的关系可以成为可复用的组织知识\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Caside class=\"editorjs-callout editorjs-callout--info my-6 rounded-xl border p-5 border-blue-300 bg-blue-50 dark:border-blue-900 dark:bg-blue-950\u002F20\" role=\"note\">\u003Cstrong class=\"block mb-2 text-gray-900 dark:text-gray-100\">RAG 仍然重要\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">SoT 并不取代检索。RAG 仍然是寻找候选证据的有价值机制。DAG 增加了仅靠检索无法提供的持久证据结构。\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Csection class=\"editorjs-process my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">从检索到证据治理的决策智能\u003C\u002Fh3>\u003Cdiv class=\"grid grid-cols-1 md:grid-cols-2 xl:grid-cols-3 gap-4\">\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">1\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">问题\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">定义研究目标和案例上下文。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">2\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">检索\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">使用 RAG 和搜索查找相关候选证据。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">3\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">规范化证据\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">保留来源、段落、元数据和来源追溯。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">4\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">构建关系\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">将证据与主张、替代方案、冲突和依赖关系连接起来。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">5\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">在 DAG 上推理\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">评估受支持的路径而不丢失证据链。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">6\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">综合\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">在可用证据的边界内生成结论。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">7\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">审查与发布\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">人工或受治理的工作流决定结果是否成为被接受的知识。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Ch2 id=\"evidence\">证据不仅仅是引用\u003C\u002Fh2>\n\u003Cp>引用显示信息的来源。SoT 更进一步，保留来源、具体段落或观察、由此得出的主张、使用它的推理路径以及依赖于它的结论之间的关系。\u003C\u002Fp>\n\u003Cp>当研究变得重要时，这种区别至关重要。决策者可能需要知道不仅引用了哪个文档，还要知道引用的材料是否真正支持该主张、是否有其他来源持不同意见、该来源对当前案例是否仍然有效，以及结论是否仍在其证据边界内。\u003C\u002Fp>\n\u003Cdiv class=\"overflow-x-auto\">\u003Ctable class=\"w-full border-collapse\">\u003Cthead>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">层级\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">在 SoT 中的角色\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">来源\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">原始文档、数据集、记录或经批准的外部输入\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">证据\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">具体段落、数值、观察或提取的发现\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">主张\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">证据可能支持、削弱或反驳的陈述\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">关系\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">支持、依赖、替代、矛盾或其他显式连接\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">结论\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">从受支持的路径中得出的综合\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">审查状态\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">结果是暂定的、已审查的、已接受的还是已被取代的\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">审计历史\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">知识及其支撑结构如何随时间变化\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"research-agents\">研究代理扩展调查范围，而非扩展权限\u003C\u002Fh2>\n\u003Cp>有些问题无法通过单次检索步骤解决。因此，研究代理可以规划并执行多个有界操作：搜索、检查来源、比较证据、识别差距、制定后续查询并收集更多发现。\u003C\u002Fp>\n\u003Cp>代理加速了调查，但权限仍存在于证据结构和审查流程中。代理输出可以成为候选发现；它不会自动成为组织真相。\u003C\u002Fp>\n\u003Ch2 id=\"contradictions\">矛盾、替代方案和未决点保持可见\u003C\u002Fh2>\n\u003Cp>证据治理系统与传统生成式界面之间最显著的区别之一是，SoT 不需要隐藏分歧。两个来源可能支持不同的主张。所需文档可能缺失。结论可能仅在特定条件下有效。这些状态可以在图中保持明确，而不是被扁平化为一个流畅的答案。\u003C\u002Fp>\n\u003Caside class=\"editorjs-callout editorjs-callout--note my-6 rounded-xl border p-5 border-gray-300 bg-gray-50 dark:border-gray-700 dark:bg-gray-900\u002F40\" role=\"note\">\u003Cstrong class=\"block mb-2 text-gray-900 dark:text-gray-100\">不确定性是有用信息\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">开放、矛盾、支持薄弱或有条件有效都是合法的知识状态。使它们可见可能比强行给出过早的单一答案更有价值。\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"validity-boundary\">每个结论都有证据边界\u003C\u002Fh2>\n\u003Cp>结论的强度取决于支持它的证据路径。因此，SoT 将\u003Cstrong>答案有效性边界\u003C\u002Fstrong>视为架构关注点：哪些证据可用、使用了哪些假设、考虑了哪些替代方案、哪些仍未解决，以及在哪些条件下结论仍应被信任。\u003C\u002Fp>\n\u003Cp>这使得后续审查成为可能。当来源发生变化、新文档到达或先前未决问题得到解决时，可以重新评估受影响的依赖路径，而无需从记忆中重建整个研究过程。\u003C\u002Fp>\n\u003Ch2 id=\"governance\">RBAC、命名空间和案例范围将研究限制在组织边界内\u003C\u002Fh2>\n\u003Cp>企业智能必须尊重组织边界。SoT 结合案例范围、命名空间和基于角色的访问控制，使研究可以限制在正确的知识领域、项目、租户、部门或用户角色内。\u003C\u002Fp>\n\u003Cul>\u003Cli>\u003Cstrong>案例范围：\u003C\u002Fstrong>定义哪些问题和证据空间属于当前研究任务。\u003C\u002Fli>\u003Cli>\u003Cstrong>命名空间：\u003C\u002Fstrong>分隔知识领域，并为访问和生命周期策略提供结构基础。\u003C\u002Fli>\u003Cli>\u003Cstrong>RBAC：\u003C\u002Fstrong>控制哪些用户和角色可以访问功能、来源和结果。\u003C\u002Fli>\u003Cli>\u003Cstrong>数据边界：\u003C\u002Fstrong>使检索和推理与已批准的信息空间保持一致。\u003C\u002Fli>\u003Cli>\u003Cstrong>审查工作流：\u003C\u002Fstrong>将生成或临时输出与已接受的组织知识分开。\u003C\u002Fli>\u003C\u002Ful>\n\u003Ch2 id=\"model-independence\">模型可以更换，而不会成为知识系统\u003C\u002Fh2>\n\u003Cp>不同任务可以使用不同的 AI 能力：本地模型、云模型、提取模型、推理模型或专用代理。SoT 将这些模型选择与证据层分开。模型帮助处理知识；它不拥有知识。\u003C\u002Fp>\n\u003Cp>这种分离创造了实际的灵活性。模型可以被替换、升级或以不同方式路由，而底层来源、主张、证据关系、审查状态和审计历史保持完整。\u003C\u002Fp>\n\u003Ch2 id=\"human-review\">人工审查仍是最终权限\u003C\u002Fh2>\n\u003Cp>SoT 可以自动化研究步骤、证据收集、比较和综合，但承担后果的接受仍是治理决策。负责任的审查者或已定义的组织工作流决定结论是被接受、发布、修订还是拒绝。\u003C\u002Fp>\n\u003Cblockquote class=\"border-l-4 border-gray-300 pl-4 italic\">自动化研究路径，而非组织权限。\u003Ccite class=\"block mt-2 text-sm\">— 真相来源治理原则\u003C\u002Fcite>\u003C\u002Fblockquote>\n\u003Ch2 id=\"use-cases\">实际价值所在\u003C\u002Fh2>\n\u003Cp>只要决策依赖于分布式证据、重复性研究或可解释的结论，该架构就能发挥作用。最有力的用例并非仅由行业定义，而是由对可追溯性、受控访问和可复用知识的需求所定义。\u003C\u002Fp>\n\u003Cdiv class=\"overflow-x-auto\">\u003Ctable class=\"w-full border-collapse\">\u003Cthead>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">用例\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">SoT 的贡献\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">政策与合规研究\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">可追溯的证据、相互竞争的解释及审查历史\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">技术决策支持\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">架构证据、备选方案、依赖关系及决策依据\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">尽职调查与文档审查\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">跨大型来源集的结构化发现及来源信息\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">内部知识研究\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">搜索加上可复用的已审查结论，而非一次性聊天回答\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">事件与根因分析\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">连接观察、假设与已验证原因的证据路径\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">产品与市场情报\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">多来源、主张比较及明确的不确定性\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">受监管或敏感领域\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">RBAC、命名空间、审查状态及可审计的决策路径\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"what-it-is\">SoT 是什么——以及不是什么\u003C\u002Fh2>\n\u003Csection class=\"editorjs-pros-cons my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Source of Truth 的边界\u003C\u002Fh3>\u003Cdiv class=\"grid grid-cols-1 lg:grid-cols-2 gap-4\">\u003Cdiv class=\"rounded-xl border border-emerald-200 bg-emerald-50\u002F50 p-4 dark:border-emerald-900 dark:bg-emerald-950\u002F20\">\u003Ch4 class=\"font-semibold text-emerald-900 dark:text-emerald-200\">SoT 是\u003C\u002Fh4>\u003Cul class=\"mt-3 space-y-2 text-gray-700 dark:text-gray-200\">\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个以证据治理的决策智能层\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个结构化的研究与知识工作流\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个持久化的证据与依赖关系图\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个面向 RAG、智能体和模型的受控集成点\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个用于来源追溯、矛盾处理、审查与审计的系统\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个可复用的组织知识基础\u003C\u002Fspan>\u003C\u002Fli>\u003C\u002Ful>\u003C\u002Fdiv>\u003Cdiv class=\"rounded-xl border border-red-200 bg-red-50\u002F50 p-4 dark:border-red-900 dark:bg-red-950\u002F20\">\u003Ch4 class=\"font-semibold text-red-900 dark:text-red-200\">SoT 不是\u003C\u002Fh4>\u003Cul class=\"mt-3 space-y-2 text-gray-700 dark:text-gray-200\">\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个通过生成流畅回答而变得权威的聊天机器人\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>对来源质量或专家判断的替代品\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个被呈现为完整知识系统的向量数据库\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个单模型架构\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>一个自动解决所有矛盾的机制\u003C\u002Fspan>\u003C\u002Fli>\u003Cli class=\"flex gap-2\">\u003Cspan aria-hidden=\"true\">•\u003C\u002Fspan>\u003Cspan>对组织责任的替代\u003C\u002Fspan>\u003C\u002Fli>\u003C\u002Ful>\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Ch2 id=\"implementation\">当前实现方向\u003C\u002Fh2>\n\u003Caside class=\"editorjs-experience my-6 rounded-xl border border-amber-200 dark:border-amber-900 bg-amber-50\u002F70 dark:bg-amber-950\u002F20 p-5\">\u003Cdiv class=\"flex flex-wrap items-center justify-between gap-2\">\u003Cstrong class=\"text-sm font-semibold text-amber-900 dark:text-amber-200\">实现证据\u003C\u002Fstrong>\u003Ctime class=\"text-xs text-gray-500 dark:text-gray-400\">2026-11-18\u003C\u002Ftime>\u003C\u002Fdiv>\u003Cdiv class=\"mt-3 text-gray-800 dark:text-gray-100\">当前的 Source of Truth 产品工作结合了文档摄取、索引检索、主张与证据关系、矛盾处理、多步研究、证据推理、命名空间、角色与权限边界、模型集成、受监控运行及已审查的知识状态。该架构正被开发为一个模块化系统，其中检索、证据管理和推理也可根据客户用例在不同层级上使用。\u003C\u002Fdiv>\u003Cdiv class=\"mt-3 text-sm text-gray-600 dark:text-gray-300\">产品开发与验证正在进行中。本描述不声称已获得客户采用、收入、产品市场契合度或企业级规模。\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"faq\">常见问题\u003C\u002Fh2>\n\u003Csection class=\"editorjs-faq my-6 rounded-xl border border-gray-200 p-5 dark:border-gray-700\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">面向企业研究与决策支持的 Source of Truth\u003C\u002Fh3>\u003Cdiv id=\"faq-rag\" class=\"border-t border-gray-200 py-4 first:border-t-0 dark:border-gray-700\">\u003Ch4 class=\"font-semibold text-gray-900 dark:text-gray-100\">SoT 是一个 RAG 系统吗？\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">RAG 是 SoT 内部一个重要的检索层。更广泛的架构增加了证据 DAG、来源追溯、主张、矛盾、推理路径、访问控制、审查状态及可审计性。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq-dag\" class=\"border-t border-gray-200 py-4 first:border-t-0 dark:border-gray-700\">\u003Ch4 class=\"font-semibold text-gray-900 dark:text-gray-100\">Source of Truth 中的 DAG 是什么意思？\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">DAG 意为有向无环图。在 SoT 中，它表示证据、主张、发现与结论之间的有向依赖关系，同时防止循环论证。这使得推理路径可追溯且可复用。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq-dag-rag\" class=\"border-t border-gray-200 py-4 first:border-t-0 dark:border-gray-700\">\u003Ch4 class=\"font-semibold text-gray-900 dark:text-gray-100\">DAG 会取代 RAG 吗？\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">不会。RAG 检索候选证据。DAG 构建这些证据与主张和结论之间关系的结构。它们解决同一研究工作流中不同层级的问题。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq-truth\" class=\"border-t border-gray-200 py-4 first:border-t-0 dark:border-gray-700\">\u003Ch4 class=\"font-semibold text-gray-900 dark:text-gray-100\">SoT 会自动判定什么是真实的吗？\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">不会。SoT 构建证据与推理的结构，保留不确定性并支持审查。被接受的组织知识仍然是受治理的人工或工作流决策。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq-model\" class=\"border-t border-gray-200 py-4 first:border-t-0 dark:border-gray-700\">\u003Ch4 class=\"font-semibold text-gray-900 dark:text-gray-100\">SoT 是否绑定于单一 AI 模型或提供商？\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">不是。该架构将模型和智能体与证据层分离，使得本地、云端或混合模型策略可以演进，而不会丢失底层知识结构。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq-business\" class=\"border-t border-gray-200 py-4 first:border-t-0 dark:border-gray-700\">\u003Ch4 class=\"font-semibold text-gray-900 dark:text-gray-100\">主要的业务优势是什么？\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">主要优势在于从企业问题到有支撑结论的可复用且可审计的路径：减少重复的人工研究、更清晰的证据、可见的不确定性、受控的 AI 访问以及更好的组织知识连续性。\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Ch2 id=\"key-terms\">关键术语\u003C\u002Fh2>\n\u003Csection class=\"editorjs-glossary my-6 rounded-xl border border-gray-200 dark:border-gray-700 p-5\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Source of Truth 术语\u003C\u002Fh3>\u003Cdl>\u003Cdiv id=\"source-of-truth\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">Source of Truth (SoT)\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">一个以证据治理的决策智能平台，保留从企业问题到证据、推理、结论和审查的路径。\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"rag\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">RAG\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">检索增强生成：检索相关源材料并将其作为上下文提供给生成式模型。\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"dag\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">DAG\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">有向无环图：一种无环的有向图。在 SoT 中，它表示证据、主张、发现与结论之间的依赖关系。\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"evidence-provenance\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">证据来源追溯\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">显示证据来源及其如何进入研究和推理路径的信息。\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"evidence-relationship\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">证据关系\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">一种明确的连接，描述证据、主张或结论之间如何相互支持、依赖、矛盾或提供替代方案。\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"answer-validity-boundary\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">答案有效性边界\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">结论仍得到支持的条件和证据限制。\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"namespace\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">命名空间\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">按组织、租户、项目、案例或领域分隔知识的逻辑边界。\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"rbac\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">RBAC\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">基于角色的访问控制：基于分配的角色及其允许的操作或数据范围进行授权。\u003C\u002Fdd>\u003C\u002Fdiv>\u003C\u002Fdl>\u003C\u002Fsection>\n\u003Ch2 id=\"core-advantage\">核心优势：决策可追溯至其证据\u003C\u002Fh2>\n\u003Cp>Source of Truth 的最大价值不在于 AI 能更快地产生答案，而在于组织能够保留\u003Cstrong>为什么\u003C\u002Fstrong>得出某个答案、\u003Cstrong>哪些证据\u003C\u002Fstrong>支持它、\u003Cstrong>何处仍存在不确定性\u003C\u002Fstrong>、\u003Cstrong>考虑了哪些备选方案\u003C\u002Fstrong>以及\u003Cstrong>谁批准了结果\u003C\u002Fstrong>。\u003C\u002Fp>\n\u003Cp>RAG 帮助系统查找。DAG 帮助系统构建结构。研究智能体帮助系统调查。证据推理帮助系统论证。治理与人工审查决定组织接受什么。这些层共同将 AI 辅助研究转化为可复用的决策智能。\u003C\u002Fp>",{"time":272,"blocks":273,"version":790},1795027506322,[274,280,285,293,301,307,312,320,358,364,369,374,424,431,460,466,471,476,505,511,516,521,527,532,539,545,550,555,561,566,579,585,590,595,601,606,612,618,623,652,658,681,687,696,702,732,738,774,780,785],{"id":275,"data":276,"type":278,"tunes":279},"intro01",{"text":277},"\u003Cstrong>Source of Truth（SoT）是一个由证据治理的决策智能平台，旨在将分布式企业知识转化为可追溯、可审查且可复用的结论。\u003C\u002Fstrong>它结合了受控的源数据摄取、检索、研究代理、证据关系、来源追溯、矛盾处理、推理、访问控制和人工审查，使答案不仅被生成，而且由可见的决策路径所支撑。","paragraph",{},{"id":281,"data":282,"type":278,"tunes":284},"intro02",{"text":283},"核心价值不在于更好的聊天机器人。价值在于一条受治理的路径：\u003Cstrong>问题 → 证据 → 关系 → 结论 → 审查\u003C\u002Fstrong>。相关信息可以被快速找到，但每一个重要陈述也都能与其来源、上下文、竞争性证据和审查状态保持关联。",{},{"id":286,"data":287,"type":291,"tunes":292},"quote01",{"text":288,"caption":289,"alignment":290},"模型可以协助研究和推理。证据结构使组织保持控制。","Source of Truth 设计原则","left","quote",{},{"id":294,"data":295,"type":299,"tunes":300},"toc01",{"title":296,"maxLevel":297,"minLevel":298},"目录",3,2,"tableOfContents",{},{"id":302,"data":303,"type":42,"tunes":305},"h2value",{"text":304,"level":298},"Source of Truth 为组织创造的价值",{"anchor":306},"organizational-value",{"id":308,"data":309,"type":278,"tunes":311},"pvalue1",{"text":310},"组织很少缺乏信息。更大的挑战在于信息分散在文档、数据库、报告、团队和系统中，而重要决策需要的不仅仅是单一的搜索结果。SoT 创建了一个受控的智能层，在其中相关证据可以被定位、关联、评估和保存。",{},{"id":313,"data":314,"type":318,"tunes":319},"callvalue",{"body":315,"title":316,"variant":317},"Source of Truth 减少了重复的人工研究，保留了重要结论背后的证据，揭示了不确定性和矛盾，支持受控的 AI 使用，并将经过审查的发现转化为可复用的组织知识。","商业价值","success","callout",{},{"id":321,"data":322,"type":356,"tunes":357},"valueTable",{"content":323,"stretched":43,"withHeadings":14},[324,328,332,336,340,344,348,352],[325,326,327],"需求","SoT 能力","实际价值",[329,330,331],"查找相关知识","已批准的索引源和检索","减少在孤立系统间搜索的时间",[333,334,335],"理解结论是如何得出的","证据关系和来源追溯","从主张回溯到来源的可见路径",[337,338,339],"处理冲突信息","矛盾和替代证据追踪","差异保持可见，而非被静默合并",[341,342,343],"处理不完整的证据","未决点和有效性边界","不确定性变得明确且可审查",[345,346,347],"跨团队安全使用 AI","RBAC、命名空间和有界工作流","AI 访问遵循组织权限和上下文",[349,350,351],"复用先前的研究","经过审查的知识和持久证据结构","重要发现不会消失在一次性对话中",[353,354,355],"更换模型而不丢失知识","模型无关的证据层","知识结构保持稳定，而 AI 组件持续演进","table",{},{"id":359,"data":360,"type":42,"tunes":362},"h2dag",{"text":361,"level":298},"DAG 是结构核心——RAG 是一个检索层",{"anchor":363},"dag-vs-rag",{"id":365,"data":366,"type":278,"tunes":368},"pdag1",{"text":367},"RAG 有助于理解系统的一部分，因为它检索相关源材料并将其作为上下文提供给 AI 模型。然而在 SoT 中，更重要的结构是 \u003Cstrong>DAG——有向无环图\u003C\u002Fstrong>。DAG 表示证据、主张、中间发现和结论如何相互依赖，而不产生循环论证。",{},{"id":370,"data":371,"type":278,"tunes":373},"pdag2",{"text":372},"RAG 管道回答：\u003Cem>哪些段落与此查询相关？\u003C\u002Fem>SoT 证据 DAG 回答更广泛的问题：\u003Cem>哪些主张依赖于哪些证据，哪些结论依赖于哪些主张，替代方案或矛盾存在于何处，以及推理路径如何向后追溯？\u003C\u002Fem>",{},{"id":375,"data":376,"type":422,"tunes":423},"cmpDag",{"rows":377,"title":414,"layout":356,"columns":415},[378,384,390,396,402,408],{"id":379,"label":380,"values":381},"purpose","主要目的",{"dag":382,"rag":383},"表示证据、主张、发现和结论之间的依赖关系","为查询检索相关上下文",{"id":385,"label":386,"values":387},"unit","主要单元",{"dag":388,"rag":389},"节点和有向证据关系","检索到的段落或文档",{"id":391,"label":392,"values":393},"trace","可追溯性",{"dag":394,"rag":395},"可以通过支撑链追溯结论","可以引用检索到的来源",{"id":397,"label":398,"values":399},"conflict","矛盾",{"dag":400,"rag":401},"可以将冲突保留为显式关系或替代路径","可能检索到相互冲突的段落",{"id":403,"label":404,"values":405},"reasoning","推理结构",{"dag":406,"rag":407},"作为显式依赖结构持久化","通常在查询时组装",{"id":409,"label":410,"values":411},"reuse","复用",{"dag":412,"rag":413},"经过审查的关系可以成为可复用的组织知识","每次查询都重复检索","RAG 和 SoT 证据 DAG 解决的是问题的不同部分",[416,419],{"id":417,"label":418},"rag","RAG",{"id":420,"label":421},"dag","SoT 中的证据 DAG","comparison",{},{"id":425,"data":426,"type":318,"tunes":430},"callDag",{"body":427,"title":428,"variant":429},"SoT 并不取代检索。RAG 仍然是寻找候选证据的有价值机制。DAG 增加了仅靠检索无法提供的持久证据结构。","RAG 仍然重要","info",{},{"id":432,"data":433,"type":458,"tunes":459},"flowDag",{"steps":434,"title":456,"orientation":457},[435,438,441,444,447,450,453],{"label":436,"description":437},"问题","定义研究目标和案例上下文。",{"label":439,"description":440},"检索","使用 RAG 和搜索查找相关候选证据。",{"label":442,"description":443},"规范化证据","保留来源、段落、元数据和来源追溯。",{"label":445,"description":446},"构建关系","将证据与主张、替代方案、冲突和依赖关系连接起来。",{"label":448,"description":449},"在 DAG 上推理","评估受支持的路径而不丢失证据链。",{"label":451,"description":452},"综合","在可用证据的边界内生成结论。",{"label":454,"description":455},"审查与发布","人工或受治理的工作流决定结果是否成为被接受的知识。","从检索到证据治理的决策智能","auto","processFlow",{},{"id":461,"data":462,"type":42,"tunes":464},"h2evidence",{"text":463,"level":298},"证据不仅仅是引用",{"anchor":465},"evidence",{"id":467,"data":468,"type":278,"tunes":470},"pevidence1",{"text":469},"引用显示信息的来源。SoT 更进一步，保留来源、具体段落或观察、由此得出的主张、使用它的推理路径以及依赖于它的结论之间的关系。",{},{"id":472,"data":473,"type":278,"tunes":475},"pevidence2",{"text":474},"当研究变得重要时，这种区别至关重要。决策者可能需要知道不仅引用了哪个文档，还要知道引用的材料是否真正支持该主张、是否有其他来源持不同意见、该来源对当前案例是否仍然有效，以及结论是否仍在其证据边界内。",{},{"id":477,"data":478,"type":356,"tunes":504},"evidenceTable",{"content":479,"stretched":43,"withHeadings":14},[480,483,486,489,492,495,498,501],[481,482],"层级","在 SoT 中的角色",[484,485],"来源","原始文档、数据集、记录或经批准的外部输入",[487,488],"证据","具体段落、数值、观察或提取的发现",[490,491],"主张","证据可能支持、削弱或反驳的陈述",[493,494],"关系","支持、依赖、替代、矛盾或其他显式连接",[496,497],"结论","从受支持的路径中得出的综合",[499,500],"审查状态","结果是暂定的、已审查的、已接受的还是已被取代的",[502,503],"审计历史","知识及其支撑结构如何随时间变化",{},{"id":506,"data":507,"type":42,"tunes":509},"h2research",{"text":508,"level":298},"研究代理扩展调查范围，而非扩展权限",{"anchor":510},"research-agents",{"id":512,"data":513,"type":278,"tunes":515},"presearch1",{"text":514},"有些问题无法通过单次检索步骤解决。因此，研究代理可以规划并执行多个有界操作：搜索、检查来源、比较证据、识别差距、制定后续查询并收集更多发现。",{},{"id":517,"data":518,"type":278,"tunes":520},"presearch2",{"text":519},"代理加速了调查，但权限仍存在于证据结构和审查流程中。代理输出可以成为候选发现；它不会自动成为组织真相。",{},{"id":522,"data":523,"type":42,"tunes":525},"h2contradictions",{"text":524,"level":298},"矛盾、替代方案和未决点保持可见",{"anchor":526},"contradictions",{"id":528,"data":529,"type":278,"tunes":531},"pcontradictions1",{"text":530},"证据治理系统与传统生成式界面之间最显著的区别之一是，SoT 不需要隐藏分歧。两个来源可能支持不同的主张。所需文档可能缺失。结论可能仅在特定条件下有效。这些状态可以在图中保持明确，而不是被扁平化为一个流畅的答案。",{},{"id":533,"data":534,"type":318,"tunes":538},"callUncertainty",{"body":535,"title":536,"variant":537},"开放、矛盾、支持薄弱或有条件有效都是合法的知识状态。使它们可见可能比强行给出过早的单一答案更有价值。","不确定性是有用信息","note",{},{"id":540,"data":541,"type":42,"tunes":543},"h2validity",{"text":542,"level":298},"每个结论都有证据边界",{"anchor":544},"validity-boundary",{"id":546,"data":547,"type":278,"tunes":549},"pvalidity1",{"text":548},"结论的强度取决于支持它的证据路径。因此，SoT 将\u003Cstrong>答案有效性边界\u003C\u002Fstrong>视为架构关注点：哪些证据可用、使用了哪些假设、考虑了哪些替代方案、哪些仍未解决，以及在哪些条件下结论仍应被信任。",{},{"id":551,"data":552,"type":278,"tunes":554},"pvalidity2",{"text":553},"这使得后续审查成为可能。当来源发生变化、新文档到达或先前未决问题得到解决时，可以重新评估受影响的依赖路径，而无需从记忆中重建整个研究过程。",{},{"id":556,"data":557,"type":42,"tunes":559},"h2governance",{"text":558,"level":298},"RBAC、命名空间和案例范围将研究限制在组织边界内",{"anchor":560},"governance",{"id":562,"data":563,"type":278,"tunes":565},"pgovernance1",{"text":564},"企业智能必须尊重组织边界。SoT 结合案例范围、命名空间和基于角色的访问控制，使研究可以限制在正确的知识领域、项目、租户、部门或用户角色内。",{},{"id":567,"data":568,"type":577,"tunes":578},"listGov",{"meta":569,"items":570,"style":576},{},[571,572,573,574,575],"\u003Cstrong>案例范围：\u003C\u002Fstrong>定义哪些问题和证据空间属于当前研究任务。","\u003Cstrong>命名空间：\u003C\u002Fstrong>分隔知识领域，并为访问和生命周期策略提供结构基础。","\u003Cstrong>RBAC：\u003C\u002Fstrong>控制哪些用户和角色可以访问功能、来源和结果。","\u003Cstrong>数据边界：\u003C\u002Fstrong>使检索和推理与已批准的信息空间保持一致。","\u003Cstrong>审查工作流：\u003C\u002Fstrong>将生成或临时输出与已接受的组织知识分开。","unordered","list",{},{"id":580,"data":581,"type":42,"tunes":583},"h2models",{"text":582,"level":298},"模型可以更换，而不会成为知识系统",{"anchor":584},"model-independence",{"id":586,"data":587,"type":278,"tunes":589},"pmodels1",{"text":588},"不同任务可以使用不同的 AI 能力：本地模型、云模型、提取模型、推理模型或专用代理。SoT 将这些模型选择与证据层分开。模型帮助处理知识；它不拥有知识。",{},{"id":591,"data":592,"type":278,"tunes":594},"pmodels2",{"text":593},"这种分离创造了实际的灵活性。模型可以被替换、升级或以不同方式路由，而底层来源、主张、证据关系、审查状态和审计历史保持完整。",{},{"id":596,"data":597,"type":42,"tunes":599},"h2human",{"text":598,"level":298},"人工审查仍是最终权限",{"anchor":600},"human-review",{"id":602,"data":603,"type":278,"tunes":605},"phuman1",{"text":604},"SoT 可以自动化研究步骤、证据收集、比较和综合，但承担后果的接受仍是治理决策。负责任的审查者或已定义的组织工作流决定结论是被接受、发布、修订还是拒绝。",{},{"id":607,"data":608,"type":291,"tunes":611},"quoteHuman",{"text":609,"caption":610,"alignment":290},"自动化研究路径，而非组织权限。","真相来源治理原则",{},{"id":613,"data":614,"type":42,"tunes":616},"h2usecases",{"text":615,"level":298},"实际价值所在",{"anchor":617},"use-cases",{"id":619,"data":620,"type":278,"tunes":622},"pusecases1",{"text":621},"只要决策依赖于分布式证据、重复性研究或可解释的结论，该架构就能发挥作用。最有力的用例并非仅由行业定义，而是由对可追溯性、受控访问和可复用知识的需求所定义。",{},{"id":624,"data":625,"type":356,"tunes":651},"useTable",{"content":626,"stretched":43,"withHeadings":14},[627,630,633,636,639,642,645,648],[628,629],"用例","SoT 的贡献",[631,632],"政策与合规研究","可追溯的证据、相互竞争的解释及审查历史",[634,635],"技术决策支持","架构证据、备选方案、依赖关系及决策依据",[637,638],"尽职调查与文档审查","跨大型来源集的结构化发现及来源信息",[640,641],"内部知识研究","搜索加上可复用的已审查结论，而非一次性聊天回答",[643,644],"事件与根因分析","连接观察、假设与已验证原因的证据路径",[646,647],"产品与市场情报","多来源、主张比较及明确的不确定性",[649,650],"受监管或敏感领域","RBAC、命名空间、审查状态及可审计的决策路径",{},{"id":653,"data":654,"type":42,"tunes":656},"h2difference",{"text":655,"level":298},"SoT 是什么——以及不是什么",{"anchor":657},"what-it-is",{"id":659,"data":660,"type":679,"tunes":680},"proscons01",{"cons":661,"pros":668,"title":675,"layout":676,"consLabel":677,"prosLabel":678},[662,663,664,665,666,667],"一个通过生成流畅回答而变得权威的聊天机器人","对来源质量或专家判断的替代品","一个被呈现为完整知识系统的向量数据库","一个单模型架构","一个自动解决所有矛盾的机制","对组织责任的替代",[669,670,671,672,673,674],"一个以证据治理的决策智能层","一个结构化的研究与知识工作流","一个持久化的证据与依赖关系图","一个面向 RAG、智能体和模型的受控集成点","一个用于来源追溯、矛盾处理、审查与审计的系统","一个可复用的组织知识基础","Source of Truth 的边界","columns","SoT 不是","SoT 是","prosCons",{},{"id":682,"data":683,"type":42,"tunes":685},"h2impl",{"text":684,"level":298},"当前实现方向",{"anchor":686},"implementation",{"id":688,"data":689,"type":694,"tunes":695},"exp01",{"date":690,"text":691,"label":692,"context":693},"2026-11-18","当前的 Source of Truth 产品工作结合了文档摄取、索引检索、主张与证据关系、矛盾处理、多步研究、证据推理、命名空间、角色与权限边界、模型集成、受监控运行及已审查的知识状态。该架构正被开发为一个模块化系统，其中检索、证据管理和推理也可根据客户用例在不同层级上使用。","实现证据","产品开发与验证正在进行中。本描述不声称已获得客户采用、收入、产品市场契合度或企业级规模。","experience",{},{"id":697,"data":698,"type":42,"tunes":700},"h2faq",{"text":699,"level":298},"常见问题",{"anchor":701},"faq",{"id":703,"data":704,"type":701,"tunes":731},"faq01",{"items":705,"title":730},[706,710,714,718,722,726],{"id":707,"answer":708,"question":709},"faq-rag","RAG 是 SoT 内部一个重要的检索层。更广泛的架构增加了证据 DAG、来源追溯、主张、矛盾、推理路径、访问控制、审查状态及可审计性。","SoT 是一个 RAG 系统吗？",{"id":711,"answer":712,"question":713},"faq-dag","DAG 意为有向无环图。在 SoT 中，它表示证据、主张、发现与结论之间的有向依赖关系，同时防止循环论证。这使得推理路径可追溯且可复用。","Source of Truth 中的 DAG 是什么意思？",{"id":715,"answer":716,"question":717},"faq-dag-rag","不会。RAG 检索候选证据。DAG 构建这些证据与主张和结论之间关系的结构。它们解决同一研究工作流中不同层级的问题。","DAG 会取代 RAG 吗？",{"id":719,"answer":720,"question":721},"faq-truth","不会。SoT 构建证据与推理的结构，保留不确定性并支持审查。被接受的组织知识仍然是受治理的人工或工作流决策。","SoT 会自动判定什么是真实的吗？",{"id":723,"answer":724,"question":725},"faq-model","不是。该架构将模型和智能体与证据层分离，使得本地、云端或混合模型策略可以演进，而不会丢失底层知识结构。","SoT 是否绑定于单一 AI 模型或提供商？",{"id":727,"answer":728,"question":729},"faq-business","主要优势在于从企业问题到有支撑结论的可复用且可审计的路径：减少重复的人工研究、更清晰的证据、可见的不确定性、受控的 AI 访问以及更好的组织知识连续性。","主要的业务优势是什么？","面向企业研究与决策支持的 Source of Truth",{},{"id":733,"data":734,"type":42,"tunes":736},"h2gloss",{"text":735,"level":298},"关键术语",{"anchor":737},"key-terms",{"id":739,"data":740,"type":772,"tunes":773},"gloss01",{"title":741,"entries":742},"Source of Truth 术语",[743,747,749,752,756,760,764,768],{"term":744,"anchor":745,"definition":746},"Source of Truth (SoT)","source-of-truth","一个以证据治理的决策智能平台，保留从企业问题到证据、推理、结论和审查的路径。",{"term":418,"anchor":417,"definition":748},"检索增强生成：检索相关源材料并将其作为上下文提供给生成式模型。",{"term":750,"anchor":420,"definition":751},"DAG","有向无环图：一种无环的有向图。在 SoT 中，它表示证据、主张、发现与结论之间的依赖关系。",{"term":753,"anchor":754,"definition":755},"证据来源追溯","evidence-provenance","显示证据来源及其如何进入研究和推理路径的信息。",{"term":757,"anchor":758,"definition":759},"证据关系","evidence-relationship","一种明确的连接，描述证据、主张或结论之间如何相互支持、依赖、矛盾或提供替代方案。",{"term":761,"anchor":762,"definition":763},"答案有效性边界","answer-validity-boundary","结论仍得到支持的条件和证据限制。",{"term":765,"anchor":766,"definition":767},"命名空间","namespace","按组织、租户、项目、案例或领域分隔知识的逻辑边界。",{"term":769,"anchor":770,"definition":771},"RBAC","rbac","基于角色的访问控制：基于分配的角色及其允许的操作或数据范围进行授权。","glossary",{},{"id":775,"data":776,"type":42,"tunes":778},"h2final",{"text":777,"level":298},"核心优势：决策可追溯至其证据",{"anchor":779},"core-advantage",{"id":781,"data":782,"type":278,"tunes":784},"pfinal1",{"text":783},"Source of Truth 的最大价值不在于 AI 能更快地产生答案，而在于组织能够保留\u003Cstrong>为什么\u003C\u002Fstrong>得出某个答案、\u003Cstrong>哪些证据\u003C\u002Fstrong>支持它、\u003Cstrong>何处仍存在不确定性\u003C\u002Fstrong>、\u003Cstrong>考虑了哪些备选方案\u003C\u002Fstrong>以及\u003Cstrong>谁批准了结果\u003C\u002Fstrong>。",{},{"id":786,"data":787,"type":278,"tunes":789},"pfinal2",{"text":788},"RAG 帮助系统查找。DAG 帮助系统构建结构。研究智能体帮助系统调查。证据推理帮助系统论证。治理与人工审查决定组织接受什么。这些层共同将 AI 辅助研究转化为可复用的决策智能。",{},"2.31","真相源企业智能平台如何超越检索，将受控案例范围、RAG、研究代理、证据推理、矛盾处理、RBAC、命名空间和人工审核相结合，形成从企业问题到有依据决策的可审计路径。","\u002Fuploads\u002F2026\u002F11\u002Fsot-erklaervideo-corporate-blue-v30-1795025960096-p9mxhp.mp4","sot-erklaervideo-corporate-blue-v30-1795025960096-p9mxhp","PUBLISHED","2026-11-18T16:16:00.000Z","2026-11-18T18:16:12.994Z","2026-11-18T18:40:46.627Z",{"en":92,"de":799,"sr":800,"es":801,"fr":802,"it":803,"ru":804,"zh":805},"\u002Fde\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","\u002Fsr\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","\u002Fes\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","\u002Ffr\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","\u002Fit\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","\u002Fru\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value","\u002Fzh\u002Fportfolio\u002Fsource-of-truth-beyond-rag-evidence-governed-research-traceability-and-business-value",{"id":807,"login":808,"email":809,"displayName":810},"20","rooth8233","","Aleksandar Stajić",[812,1227],{"lang":813,"title":814,"content":815,"contentJson":816,"excerpt":1226},"en","Source of Truth Beyond RAG: Evidence-Governed Research, Traceability and Business Value","{\"time\":1795034000000,\"blocks\":[{\"id\":\"intro01\",\"type\":\"paragraph\",\"data\":{\"text\":\"\u003Cstrong>Source of Truth (SoT) is an evidence-governed decision intelligence platform designed to turn distributed enterprise knowledge into traceable, reviewable and reusable conclusions.\u003C\u002Fstrong> It combines controlled source ingestion, retrieval, research agents, evidence relationships, provenance, contradiction handling, reasoning, access control and human review so that an answer is not only generated, but supported by a visible decision path.\"},\"tunes\":{}},{\"id\":\"intro02\",\"type\":\"paragraph\",\"data\":{\"text\":\"The central value is not a better chatbot. The value is a governed path from \u003Cstrong>question → evidence → relationships → conclusion → review\u003C\u002Fstrong>. Relevant information can be found quickly, but every important statement can also remain connected to its source, context, competing evidence and review state.\"},\"tunes\":{}},{\"id\":\"quote01\",\"type\":\"quote\",\"data\":{\"text\":\"The model can assist with research and reasoning. The evidence structure keeps the organization in control.\",\"caption\":\"Source of Truth design principle\",\"alignment\":\"left\"},\"tunes\":{}},{\"id\":\"toc01\",\"type\":\"tableOfContents\",\"data\":{\"title\":\"Contents\",\"minLevel\":2,\"maxLevel\":3},\"tunes\":{}},{\"id\":\"h2value\",\"type\":\"header\",\"data\":{\"text\":\"What Source of Truth Creates for an Organization\",\"level\":2},\"tunes\":{\"anchor\":\"organizational-value\"}},{\"id\":\"pvalue1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Organizations rarely lack information. The greater challenge is that information is distributed across documents, databases, reports, teams and systems, while important decisions require more than a single search result. SoT creates a controlled intelligence layer in which relevant evidence can be located, related, evaluated and preserved.\"},\"tunes\":{}},{\"id\":\"callvalue\",\"type\":\"callout\",\"data\":{\"variant\":\"success\",\"title\":\"Business Mehrwert\",\"body\":\"Source of Truth reduces repeated manual research, preserves the evidence behind important conclusions, exposes uncertainty and contradictions, supports controlled AI use, and turns reviewed findings into reusable organizational knowledge.\"},\"tunes\":{}},{\"id\":\"valueTable\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Need\",\"SoT capability\",\"Practical value\"],[\"Find relevant knowledge\",\"Indexed approved sources and retrieval\",\"Less time spent searching across disconnected systems\"],[\"Understand why a conclusion was reached\",\"Evidence relationships and provenance\",\"A visible path from claim back to source\"],[\"Handle conflicting information\",\"Contradiction and alternative-evidence tracking\",\"Differences remain visible instead of being silently merged\"],[\"Work with incomplete evidence\",\"Open points and validity boundaries\",\"Uncertainty becomes explicit and reviewable\"],[\"Use AI safely across teams\",\"RBAC, namespaces and bounded workflows\",\"AI access follows organizational permissions and context\"],[\"Reuse prior research\",\"Reviewed knowledge and persistent evidence structure\",\"Important findings do not disappear into one-off chats\"],[\"Change models without losing knowledge\",\"Model-independent evidence layer\",\"The knowledge structure remains stable while AI components evolve\"]]},\"tunes\":{}},{\"id\":\"h2dag\",\"type\":\"header\",\"data\":{\"text\":\"DAG Is the Structural Core — RAG Is One Retrieval Layer\",\"level\":2},\"tunes\":{\"anchor\":\"dag-vs-rag\"}},{\"id\":\"pdag1\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG is useful for understanding one part of the system because it retrieves relevant source material and supplies it as context to an AI model. In SoT, however, the more important structure is a \u003Cstrong>DAG — a Directed Acyclic Graph\u003C\u002Fstrong>. The DAG represents how evidence, claims, intermediate findings and conclusions depend on one another without creating circular justification.\"},\"tunes\":{}},{\"id\":\"pdag2\",\"type\":\"paragraph\",\"data\":{\"text\":\"A RAG pipeline answers: \u003Cem>Which passages are relevant to this query?\u003C\u002Fem> A SoT evidence DAG answers a broader question: \u003Cem>Which claims depend on which evidence, which conclusions depend on which claims, where do alternatives or contradictions exist, and how can the reasoning path be traced backward?\u003C\u002Fem>\"},\"tunes\":{}},{\"id\":\"cmpDag\",\"type\":\"comparison\",\"data\":{\"title\":\"RAG and the SoT evidence DAG solve different parts of the problem\",\"layout\":\"table\",\"columns\":[{\"id\":\"rag\",\"label\":\"RAG\"},{\"id\":\"dag\",\"label\":\"Evidence DAG in SoT\"}],\"rows\":[{\"id\":\"purpose\",\"label\":\"Primary purpose\",\"values\":{\"rag\":\"Retrieve relevant context for a query\",\"dag\":\"Represent dependencies between evidence, claims, findings and conclusions\"}},{\"id\":\"unit\",\"label\":\"Main unit\",\"values\":{\"rag\":\"Retrieved passages or documents\",\"dag\":\"Nodes and directed evidence relationships\"}},{\"id\":\"trace\",\"label\":\"Traceability\",\"values\":{\"rag\":\"Can cite retrieved sources\",\"dag\":\"Can trace a conclusion through its supporting chain\"}},{\"id\":\"conflict\",\"label\":\"Contradictions\",\"values\":{\"rag\":\"May retrieve conflicting passages\",\"dag\":\"Can preserve conflict as an explicit relationship or alternative path\"}},{\"id\":\"reasoning\",\"label\":\"Reasoning structure\",\"values\":{\"rag\":\"Usually assembled at query time\",\"dag\":\"Persisted as an explicit dependency structure\"}},{\"id\":\"reuse\",\"label\":\"Reuse\",\"values\":{\"rag\":\"Retrieval is repeated for each query\",\"dag\":\"Reviewed relationships can become reusable organizational knowledge\"}}]},\"tunes\":{}},{\"id\":\"callDag\",\"type\":\"callout\",\"data\":{\"variant\":\"info\",\"title\":\"RAG is still important\",\"body\":\"SoT does not replace retrieval. RAG remains a valuable mechanism for finding candidate evidence. The DAG adds the persistent evidence structure that retrieval alone does not provide.\"},\"tunes\":{}},{\"id\":\"flowDag\",\"type\":\"processFlow\",\"data\":{\"title\":\"From retrieval to evidence-governed decision intelligence\",\"orientation\":\"auto\",\"steps\":[{\"label\":\"Question\",\"description\":\"Define the research objective and case context.\"},{\"label\":\"Retrieve\",\"description\":\"Use RAG and search to find relevant candidate evidence.\"},{\"label\":\"Normalize evidence\",\"description\":\"Preserve source, passage, metadata and provenance.\"},{\"label\":\"Build relationships\",\"description\":\"Connect evidence to claims, alternatives, conflicts and dependencies.\"},{\"label\":\"Reason over the DAG\",\"description\":\"Evaluate supported paths without losing the evidence chain.\"},{\"label\":\"Synthesize\",\"description\":\"Produce a conclusion bounded by the available evidence.\"},{\"label\":\"Review & release\",\"description\":\"Human or governed workflow decides whether the result becomes accepted knowledge.\"}]},\"tunes\":{}},{\"id\":\"h2evidence\",\"type\":\"header\",\"data\":{\"text\":\"Evidence Is More Than a Citation\",\"level\":2},\"tunes\":{\"anchor\":\"evidence\"}},{\"id\":\"pevidence1\",\"type\":\"paragraph\",\"data\":{\"text\":\"A citation shows where information came from. SoT goes further by preserving the relationship between the source, the concrete passage or observation, the claim derived from it, the reasoning path that uses it and the conclusion that depends on it.\"},\"tunes\":{}},{\"id\":\"pevidence2\",\"type\":\"paragraph\",\"data\":{\"text\":\"That distinction matters when research becomes consequential. A decision-maker may need to know not only which document was cited, but whether the cited material actually supports the claim, whether another source disagrees, whether the source is still valid for the current case and whether the conclusion remains inside its evidence boundary.\"},\"tunes\":{}},{\"id\":\"evidenceTable\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Layer\",\"Role in SoT\"],[\"Source\",\"Original document, dataset, record or approved external input\"],[\"Evidence\",\"Concrete passage, value, observation or extracted finding\"],[\"Claim\",\"A statement that the evidence may support, weaken or contradict\"],[\"Relationship\",\"Support, dependency, alternative, contradiction or other explicit connection\"],[\"Conclusion\",\"A synthesis derived from the supported paths\"],[\"Review state\",\"Whether the result is provisional, reviewed, accepted or superseded\"],[\"Audit history\",\"How the knowledge and its supporting structure changed over time\"]]},\"tunes\":{}},{\"id\":\"h2research\",\"type\":\"header\",\"data\":{\"text\":\"Research Agents Extend the Investigation, Not the Authority\",\"level\":2},\"tunes\":{\"anchor\":\"research-agents\"}},{\"id\":\"presearch1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Some questions cannot be resolved through a single retrieval step. Research agents can therefore plan and execute multiple bounded operations: search, inspect sources, compare evidence, identify gaps, formulate follow-up queries and collect additional findings.\"},\"tunes\":{}},{\"id\":\"presearch2\",\"type\":\"paragraph\",\"data\":{\"text\":\"The agent accelerates the investigation, but authority remains in the evidence structure and review process. Agent output can become a candidate finding; it does not automatically become organizational truth.\"},\"tunes\":{}},{\"id\":\"h2contradictions\",\"type\":\"header\",\"data\":{\"text\":\"Contradictions, Alternatives and Open Points Remain Visible\",\"level\":2},\"tunes\":{\"anchor\":\"contradictions\"}},{\"id\":\"pcontradictions1\",\"type\":\"paragraph\",\"data\":{\"text\":\"One of the strongest differences between an evidence-governed system and a conventional generative interface is that SoT does not need to hide disagreement. Two sources may support different claims. A required document may be missing. A conclusion may only be valid under specific conditions. These states can remain explicit in the graph instead of being flattened into one fluent answer.\"},\"tunes\":{}},{\"id\":\"callUncertainty\",\"type\":\"callout\",\"data\":{\"variant\":\"note\",\"title\":\"Uncertainty is useful information\",\"body\":\"Open, contradictory, weakly supported or conditionally valid are legitimate knowledge states. Making them visible can be more valuable than forcing a premature single answer.\"},\"tunes\":{}},{\"id\":\"h2validity\",\"type\":\"header\",\"data\":{\"text\":\"Every Conclusion Has an Evidence Boundary\",\"level\":2},\"tunes\":{\"anchor\":\"validity-boundary\"}},{\"id\":\"pvalidity1\",\"type\":\"paragraph\",\"data\":{\"text\":\"A conclusion is only as strong as the evidence path that supports it. SoT therefore treats the \u003Cstrong>answer validity boundary\u003C\u002Fstrong> as an architectural concern: which evidence was available, which assumptions were used, which alternatives were considered, what remains unresolved and under which conditions the conclusion should still be trusted.\"},\"tunes\":{}},{\"id\":\"pvalidity2\",\"type\":\"paragraph\",\"data\":{\"text\":\"This makes later review possible. When a source changes, a new document arrives or a previously open question is resolved, the affected dependency path can be re-evaluated without reconstructing the entire research process from memory.\"},\"tunes\":{}},{\"id\":\"h2governance\",\"type\":\"header\",\"data\":{\"text\":\"RBAC, Namespaces and Case Scope Keep Research Inside Organizational Boundaries\",\"level\":2},\"tunes\":{\"anchor\":\"governance\"}},{\"id\":\"pgovernance1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Enterprise intelligence must respect organizational boundaries. SoT combines case scope, namespaces and role-based access control so that research can be limited to the correct knowledge domain, project, tenant, department or user role.\"},\"tunes\":{}},{\"id\":\"listGov\",\"type\":\"list\",\"data\":{\"style\":\"unordered\",\"meta\":{},\"items\":[\"\u003Cstrong>Case scope:\u003C\u002Fstrong> defines which question and evidence space belong to the active research task.\",\"\u003Cstrong>Namespaces:\u003C\u002Fstrong> separate knowledge domains and provide a structural basis for access and lifecycle policies.\",\"\u003Cstrong>RBAC:\u003C\u002Fstrong> controls which users and roles may access functions, sources and results.\",\"\u003Cstrong>Data boundaries:\u003C\u002Fstrong> keep retrieval and reasoning aligned with approved information spaces.\",\"\u003Cstrong>Review workflows:\u003C\u002Fstrong> separate generated or provisional output from accepted organizational knowledge.\"]},\"tunes\":{}},{\"id\":\"h2models\",\"type\":\"header\",\"data\":{\"text\":\"Models Can Change Without Becoming the Knowledge System\",\"level\":2},\"tunes\":{\"anchor\":\"model-independence\"}},{\"id\":\"pmodels1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Different tasks can use different AI capabilities: local models, cloud models, extraction models, reasoning models or specialized agents. SoT keeps those model choices separate from the evidence layer. The model helps process knowledge; it does not own the knowledge.\"},\"tunes\":{}},{\"id\":\"pmodels2\",\"type\":\"paragraph\",\"data\":{\"text\":\"This separation creates practical flexibility. A model can be replaced, upgraded or routed differently while the underlying sources, claims, evidence relationships, review states and audit history remain intact.\"},\"tunes\":{}},{\"id\":\"h2human\",\"type\":\"header\",\"data\":{\"text\":\"Human Review Remains the Final Authority\",\"level\":2},\"tunes\":{\"anchor\":\"human-review\"}},{\"id\":\"phuman1\",\"type\":\"paragraph\",\"data\":{\"text\":\"SoT can automate research steps, evidence collection, comparison and synthesis, but consequence-bearing acceptance remains a governance decision. A responsible reviewer or defined organizational workflow decides whether a conclusion is accepted, released, revised or rejected.\"},\"tunes\":{}},{\"id\":\"quoteHuman\",\"type\":\"quote\",\"data\":{\"text\":\"Automate the research path, not the organizational authority.\",\"caption\":\"Source of Truth governance principle\",\"alignment\":\"left\"},\"tunes\":{}},{\"id\":\"h2usecases\",\"type\":\"header\",\"data\":{\"text\":\"Where This Creates Practical Value\",\"level\":2},\"tunes\":{\"anchor\":\"use-cases\"}},{\"id\":\"pusecases1\",\"type\":\"paragraph\",\"data\":{\"text\":\"The architecture is useful wherever decisions depend on distributed evidence, repeated research or explainable conclusions. The strongest use cases are not defined by industry alone, but by the need for traceability, controlled access and reusable knowledge.\"},\"tunes\":{}},{\"id\":\"useTable\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Use case\",\"What SoT contributes\"],[\"Policy and compliance research\",\"Traceable evidence, competing interpretations and review history\"],[\"Technical decision support\",\"Architecture evidence, alternatives, dependencies and decision rationale\"],[\"Due diligence and document review\",\"Structured findings across large source sets with provenance\"],[\"Internal knowledge research\",\"Search plus reusable reviewed conclusions instead of one-off chat answers\"],[\"Incident and root-cause analysis\",\"Evidence paths connecting observations, hypotheses and verified causes\"],[\"Product and market intelligence\",\"Multiple sources, claim comparison and explicit uncertainty\"],[\"Regulated or sensitive domains\",\"RBAC, namespaces, review states and auditable decision paths\"]]},\"tunes\":{}},{\"id\":\"h2difference\",\"type\":\"header\",\"data\":{\"text\":\"What SoT Is — and What It Is Not\",\"level\":2},\"tunes\":{\"anchor\":\"what-it-is\"}},{\"id\":\"proscons01\",\"type\":\"prosCons\",\"data\":{\"title\":\"Source of Truth boundaries\",\"prosLabel\":\"SoT is\",\"consLabel\":\"SoT is not\",\"layout\":\"columns\",\"pros\":[\"An evidence-governed decision intelligence layer\",\"A structured research and knowledge workflow\",\"A persistent evidence and dependency graph\",\"A controlled integration point for RAG, agents and models\",\"A system for provenance, contradictions, review and auditability\",\"A reusable organizational knowledge foundation\"],\"cons\":[\"A chatbot that becomes authoritative by generating fluent answers\",\"A replacement for source quality or expert judgment\",\"A vector database presented as a complete knowledge system\",\"A single-model architecture\",\"A mechanism that automatically resolves every contradiction\",\"A substitute for organizational responsibility\"]},\"tunes\":{}},{\"id\":\"h2impl\",\"type\":\"header\",\"data\":{\"text\":\"Current Implementation Direction\",\"level\":2},\"tunes\":{\"anchor\":\"implementation\"}},{\"id\":\"exp01\",\"type\":\"experience\",\"data\":{\"label\":\"Implementation evidence\",\"text\":\"The current Source of Truth product work combines document ingestion, indexed retrieval, claims and evidence relationships, contradiction handling, multi-step research, evidence reasoning, namespaces, role and permission boundaries, model integration, monitored runs and reviewed knowledge states. The architecture is being developed as a modular system in which retrieval, evidence management and reasoning can also be used at different levels depending on the customer use case.\",\"context\":\"Product development and validation are active. This description does not claim customer adoption, revenue, product-market fit or enterprise scale.\",\"date\":\"2026-11-18\"},\"tunes\":{}},{\"id\":\"h2faq\",\"type\":\"header\",\"data\":{\"text\":\"Frequently Asked Questions\",\"level\":2},\"tunes\":{\"anchor\":\"faq\"}},{\"id\":\"faq01\",\"type\":\"faq\",\"data\":{\"title\":\"Source of Truth for enterprise research and decision support\",\"items\":[{\"id\":\"faq-rag\",\"question\":\"Is SoT a RAG system?\",\"answer\":\"RAG is one important retrieval layer inside SoT. The broader architecture adds an evidence DAG, provenance, claims, contradictions, reasoning paths, access control, review states and auditability.\"},{\"id\":\"faq-dag\",\"question\":\"What does DAG mean in Source of Truth?\",\"answer\":\"DAG means Directed Acyclic Graph. In SoT it represents directed dependencies between evidence, claims, findings and conclusions while preventing circular justification. This makes reasoning paths traceable and reusable.\"},{\"id\":\"faq-dag-rag\",\"question\":\"Does the DAG replace RAG?\",\"answer\":\"No. RAG retrieves candidate evidence. The DAG structures how that evidence relates to claims and conclusions. They solve different layers of the same research workflow.\"},{\"id\":\"faq-truth\",\"question\":\"Does SoT decide what is true automatically?\",\"answer\":\"No. SoT structures evidence and reasoning, preserves uncertainty and supports review. Accepted organizational knowledge remains a governed human or workflow decision.\"},{\"id\":\"faq-model\",\"question\":\"Is SoT tied to one AI model or provider?\",\"answer\":\"No. The architecture separates models and agents from the evidence layer so that local, cloud or hybrid model strategies can evolve without losing the underlying knowledge structure.\"},{\"id\":\"faq-business\",\"question\":\"What is the main business advantage?\",\"answer\":\"The main advantage is a reusable and auditable path from enterprise question to supported conclusion: less repeated manual research, clearer evidence, visible uncertainty, controlled AI access and better continuity of organizational knowledge.\"}]},\"tunes\":{}},{\"id\":\"h2gloss\",\"type\":\"header\",\"data\":{\"text\":\"Key Terms\",\"level\":2},\"tunes\":{\"anchor\":\"key-terms\"}},{\"id\":\"gloss01\",\"type\":\"glossary\",\"data\":{\"title\":\"Source of Truth terminology\",\"entries\":[{\"term\":\"Source of Truth (SoT)\",\"definition\":\"An evidence-governed decision intelligence platform that preserves the path from enterprise question to evidence, reasoning, conclusion and review.\",\"anchor\":\"source-of-truth\"},{\"term\":\"RAG\",\"definition\":\"Retrieval-Augmented Generation: retrieving relevant source material and providing it as context to a generative model.\",\"anchor\":\"rag\"},{\"term\":\"DAG\",\"definition\":\"Directed Acyclic Graph: a directed graph without cycles. In SoT it represents dependencies between evidence, claims, findings and conclusions.\",\"anchor\":\"dag\"},{\"term\":\"Evidence provenance\",\"definition\":\"Information showing where evidence originated and how it entered the research and reasoning path.\",\"anchor\":\"evidence-provenance\"},{\"term\":\"Evidence relationship\",\"definition\":\"An explicit connection describing how evidence, claims or conclusions support, depend on, contradict or provide alternatives to one another.\",\"anchor\":\"evidence-relationship\"},{\"term\":\"Answer validity boundary\",\"definition\":\"The conditions and evidence limits within which a conclusion remains supported.\",\"anchor\":\"answer-validity-boundary\"},{\"term\":\"Namespace\",\"definition\":\"A logical boundary that separates knowledge by organization, tenant, project, case or domain.\",\"anchor\":\"namespace\"},{\"term\":\"RBAC\",\"definition\":\"Role-Based Access Control: authorization based on assigned roles and their permitted actions or data scopes.\",\"anchor\":\"rbac\"}]},\"tunes\":{}},{\"id\":\"h2final\",\"type\":\"header\",\"data\":{\"text\":\"The Core Advantage: A Decision Can Be Traced Back to Its Evidence\",\"level\":2},\"tunes\":{\"anchor\":\"core-advantage\"}},{\"id\":\"pfinal1\",\"type\":\"paragraph\",\"data\":{\"text\":\"The strongest value of Source of Truth is not that AI can produce an answer faster. It is that an organization can preserve \u003Cstrong>why\u003C\u002Fstrong> an answer was reached, \u003Cstrong>which evidence\u003C\u002Fstrong> supports it, \u003Cstrong>where uncertainty remains\u003C\u002Fstrong>, \u003Cstrong>which alternatives were considered\u003C\u002Fstrong> and \u003Cstrong>who approved the result\u003C\u002Fstrong>.\"},\"tunes\":{}},{\"id\":\"pfinal2\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG helps the system find. The DAG helps the system structure. Research agents help the system investigate. Evidence reasoning helps the system justify. Governance and human review determine what the organization accepts. Together, these layers turn AI-assisted research into reusable decision intelligence.\"},\"tunes\":{}}],\"version\":\"2.31.6\"}",{"time":817,"blocks":818,"version":1225},1795034000000,[819,823,827,832,836,840,844,849,885,889,893,897,936,941,967,971,975,979,1007,1011,1015,1019,1023,1027,1032,1036,1040,1044,1048,1052,1062,1066,1070,1074,1078,1082,1087,1091,1095,1123,1127,1147,1151,1157,1161,1184,1188,1213,1217,1221],{"id":275,"data":820,"type":278,"tunes":822},{"text":821},"\u003Cstrong>Source of Truth (SoT) is an evidence-governed decision intelligence platform designed to turn distributed enterprise knowledge into traceable, reviewable and reusable conclusions.\u003C\u002Fstrong> It combines controlled source ingestion, retrieval, research agents, evidence relationships, provenance, contradiction handling, reasoning, access control and human review so that an answer is not only generated, but supported by a visible decision path.",{},{"id":281,"data":824,"type":278,"tunes":826},{"text":825},"The central value is not a better chatbot. The value is a governed path from \u003Cstrong>question → evidence → relationships → conclusion → review\u003C\u002Fstrong>. Relevant information can be found quickly, but every important statement can also remain connected to its source, context, competing evidence and review state.",{},{"id":286,"data":828,"type":291,"tunes":831},{"text":829,"caption":830,"alignment":290},"The model can assist with research and reasoning. The evidence structure keeps the organization in control.","Source of Truth design principle",{},{"id":294,"data":833,"type":299,"tunes":835},{"title":834,"maxLevel":297,"minLevel":298},"Contents",{},{"id":302,"data":837,"type":42,"tunes":839},{"text":838,"level":298},"What Source of Truth Creates for an Organization",{"anchor":306},{"id":308,"data":841,"type":278,"tunes":843},{"text":842},"Organizations rarely lack information. The greater challenge is that information is distributed across documents, databases, reports, teams and systems, while important decisions require more than a single search result. SoT creates a controlled intelligence layer in which relevant evidence can be located, related, evaluated and preserved.",{},{"id":313,"data":845,"type":318,"tunes":848},{"body":846,"title":847,"variant":317},"Source of Truth reduces repeated manual research, preserves the evidence behind important conclusions, exposes uncertainty and contradictions, supports controlled AI use, and turns reviewed findings into reusable organizational knowledge.","Business Mehrwert",{},{"id":321,"data":850,"type":356,"tunes":884},{"content":851,"stretched":43,"withHeadings":14},[852,856,860,864,868,872,876,880],[853,854,855],"Need","SoT capability","Practical value",[857,858,859],"Find relevant knowledge","Indexed approved sources and retrieval","Less time spent searching across disconnected systems",[861,862,863],"Understand why a conclusion was reached","Evidence relationships and provenance","A visible path from claim back to source",[865,866,867],"Handle conflicting information","Contradiction and alternative-evidence tracking","Differences remain visible instead of being silently merged",[869,870,871],"Work with incomplete evidence","Open points and validity boundaries","Uncertainty becomes explicit and reviewable",[873,874,875],"Use AI safely across teams","RBAC, namespaces and bounded workflows","AI access follows organizational permissions and context",[877,878,879],"Reuse prior research","Reviewed knowledge and persistent evidence structure","Important findings do not disappear into one-off chats",[881,882,883],"Change models without losing knowledge","Model-independent evidence layer","The knowledge structure remains stable while AI components evolve",{},{"id":359,"data":886,"type":42,"tunes":888},{"text":887,"level":298},"DAG Is the Structural Core — RAG Is One Retrieval Layer",{"anchor":363},{"id":365,"data":890,"type":278,"tunes":892},{"text":891},"RAG is useful for understanding one part of the system because it retrieves relevant source material and supplies it as context to an AI model. In SoT, however, the more important structure is a \u003Cstrong>DAG — a Directed Acyclic Graph\u003C\u002Fstrong>. The DAG represents how evidence, claims, intermediate findings and conclusions depend on one another without creating circular justification.",{},{"id":370,"data":894,"type":278,"tunes":896},{"text":895},"A RAG pipeline answers: \u003Cem>Which passages are relevant to this query?\u003C\u002Fem> A SoT evidence DAG answers a broader question: \u003Cem>Which claims depend on which evidence, which conclusions depend on which claims, where do alternatives or contradictions exist, and how can the reasoning path be traced backward?\u003C\u002Fem>",{},{"id":375,"data":898,"type":422,"tunes":935},{"rows":899,"title":930,"layout":356,"columns":931},[900,905,910,915,920,925],{"id":379,"label":901,"values":902},"Primary purpose",{"dag":903,"rag":904},"Represent dependencies between evidence, claims, findings and conclusions","Retrieve relevant context for a query",{"id":385,"label":906,"values":907},"Main unit",{"dag":908,"rag":909},"Nodes and directed evidence relationships","Retrieved passages or documents",{"id":391,"label":911,"values":912},"Traceability",{"dag":913,"rag":914},"Can trace a conclusion through its supporting chain","Can cite retrieved sources",{"id":397,"label":916,"values":917},"Contradictions",{"dag":918,"rag":919},"Can preserve conflict as an explicit relationship or alternative path","May retrieve conflicting passages",{"id":403,"label":921,"values":922},"Reasoning structure",{"dag":923,"rag":924},"Persisted as an explicit dependency structure","Usually assembled at query time",{"id":409,"label":926,"values":927},"Reuse",{"dag":928,"rag":929},"Reviewed relationships can become reusable organizational knowledge","Retrieval is repeated for each query","RAG and the SoT evidence DAG solve different parts of the problem",[932,933],{"id":417,"label":418},{"id":420,"label":934},"Evidence DAG in SoT",{},{"id":425,"data":937,"type":318,"tunes":940},{"body":938,"title":939,"variant":429},"SoT does not replace retrieval. RAG remains a valuable mechanism for finding candidate evidence. The DAG adds the persistent evidence structure that retrieval alone does not provide.","RAG is still important",{},{"id":432,"data":942,"type":458,"tunes":966},{"steps":943,"title":965,"orientation":457},[944,947,950,953,956,959,962],{"label":945,"description":946},"Question","Define the research objective and case context.",{"label":948,"description":949},"Retrieve","Use RAG and search to find relevant candidate evidence.",{"label":951,"description":952},"Normalize evidence","Preserve source, passage, metadata and provenance.",{"label":954,"description":955},"Build relationships","Connect evidence to claims, alternatives, conflicts and dependencies.",{"label":957,"description":958},"Reason over the DAG","Evaluate supported paths without losing the evidence chain.",{"label":960,"description":961},"Synthesize","Produce a conclusion bounded by the available evidence.",{"label":963,"description":964},"Review & release","Human or governed workflow decides whether the result becomes accepted knowledge.","From retrieval to evidence-governed decision intelligence",{},{"id":461,"data":968,"type":42,"tunes":970},{"text":969,"level":298},"Evidence Is More Than a Citation",{"anchor":465},{"id":467,"data":972,"type":278,"tunes":974},{"text":973},"A citation shows where information came from. SoT goes further by preserving the relationship between the source, the concrete passage or observation, the claim derived from it, the reasoning path that uses it and the conclusion that depends on it.",{},{"id":472,"data":976,"type":278,"tunes":978},{"text":977},"That distinction matters when research becomes consequential. A decision-maker may need to know not only which document was cited, but whether the cited material actually supports the claim, whether another source disagrees, whether the source is still valid for the current case and whether the conclusion remains inside its evidence boundary.",{},{"id":477,"data":980,"type":356,"tunes":1006},{"content":981,"stretched":43,"withHeadings":14},[982,985,988,991,994,997,1000,1003],[983,984],"Layer","Role in SoT",[986,987],"Source","Original document, dataset, record or approved external input",[989,990],"Evidence","Concrete passage, value, observation or extracted finding",[992,993],"Claim","A statement that the evidence may support, weaken or contradict",[995,996],"Relationship","Support, dependency, alternative, contradiction or other explicit connection",[998,999],"Conclusion","A synthesis derived from the supported paths",[1001,1002],"Review state","Whether the result is provisional, reviewed, accepted or superseded",[1004,1005],"Audit history","How the knowledge and its supporting structure changed over time",{},{"id":506,"data":1008,"type":42,"tunes":1010},{"text":1009,"level":298},"Research Agents Extend the Investigation, Not the Authority",{"anchor":510},{"id":512,"data":1012,"type":278,"tunes":1014},{"text":1013},"Some questions cannot be resolved through a single retrieval step. Research agents can therefore plan and execute multiple bounded operations: search, inspect sources, compare evidence, identify gaps, formulate follow-up queries and collect additional findings.",{},{"id":517,"data":1016,"type":278,"tunes":1018},{"text":1017},"The agent accelerates the investigation, but authority remains in the evidence structure and review process. Agent output can become a candidate finding; it does not automatically become organizational truth.",{},{"id":522,"data":1020,"type":42,"tunes":1022},{"text":1021,"level":298},"Contradictions, Alternatives and Open Points Remain Visible",{"anchor":526},{"id":528,"data":1024,"type":278,"tunes":1026},{"text":1025},"One of the strongest differences between an evidence-governed system and a conventional generative interface is that SoT does not need to hide disagreement. Two sources may support different claims. A required document may be missing. A conclusion may only be valid under specific conditions. These states can remain explicit in the graph instead of being flattened into one fluent answer.",{},{"id":533,"data":1028,"type":318,"tunes":1031},{"body":1029,"title":1030,"variant":537},"Open, contradictory, weakly supported or conditionally valid are legitimate knowledge states. Making them visible can be more valuable than forcing a premature single answer.","Uncertainty is useful information",{},{"id":540,"data":1033,"type":42,"tunes":1035},{"text":1034,"level":298},"Every Conclusion Has an Evidence Boundary",{"anchor":544},{"id":546,"data":1037,"type":278,"tunes":1039},{"text":1038},"A conclusion is only as strong as the evidence path that supports it. SoT therefore treats the \u003Cstrong>answer validity boundary\u003C\u002Fstrong> as an architectural concern: which evidence was available, which assumptions were used, which alternatives were considered, what remains unresolved and under which conditions the conclusion should still be trusted.",{},{"id":551,"data":1041,"type":278,"tunes":1043},{"text":1042},"This makes later review possible. When a source changes, a new document arrives or a previously open question is resolved, the affected dependency path can be re-evaluated without reconstructing the entire research process from memory.",{},{"id":556,"data":1045,"type":42,"tunes":1047},{"text":1046,"level":298},"RBAC, Namespaces and Case Scope Keep Research Inside Organizational Boundaries",{"anchor":560},{"id":562,"data":1049,"type":278,"tunes":1051},{"text":1050},"Enterprise intelligence must respect organizational boundaries. SoT combines case scope, namespaces and role-based access control so that research can be limited to the correct knowledge domain, project, tenant, department or user role.",{},{"id":567,"data":1053,"type":577,"tunes":1061},{"meta":1054,"items":1055,"style":576},{},[1056,1057,1058,1059,1060],"\u003Cstrong>Case scope:\u003C\u002Fstrong> defines which question and evidence space belong to the active research task.","\u003Cstrong>Namespaces:\u003C\u002Fstrong> separate knowledge domains and provide a structural basis for access and lifecycle policies.","\u003Cstrong>RBAC:\u003C\u002Fstrong> controls which users and roles may access functions, sources and results.","\u003Cstrong>Data boundaries:\u003C\u002Fstrong> keep retrieval and reasoning aligned with approved information spaces.","\u003Cstrong>Review workflows:\u003C\u002Fstrong> separate generated or provisional output from accepted organizational knowledge.",{},{"id":580,"data":1063,"type":42,"tunes":1065},{"text":1064,"level":298},"Models Can Change Without Becoming the Knowledge System",{"anchor":584},{"id":586,"data":1067,"type":278,"tunes":1069},{"text":1068},"Different tasks can use different AI capabilities: local models, cloud models, extraction models, reasoning models or specialized agents. SoT keeps those model choices separate from the evidence layer. The model helps process knowledge; it does not own the knowledge.",{},{"id":591,"data":1071,"type":278,"tunes":1073},{"text":1072},"This separation creates practical flexibility. A model can be replaced, upgraded or routed differently while the underlying sources, claims, evidence relationships, review states and audit history remain intact.",{},{"id":596,"data":1075,"type":42,"tunes":1077},{"text":1076,"level":298},"Human Review Remains the Final Authority",{"anchor":600},{"id":602,"data":1079,"type":278,"tunes":1081},{"text":1080},"SoT can automate research steps, evidence collection, comparison and synthesis, but consequence-bearing acceptance remains a governance decision. A responsible reviewer or defined organizational workflow decides whether a conclusion is accepted, released, revised or rejected.",{},{"id":607,"data":1083,"type":291,"tunes":1086},{"text":1084,"caption":1085,"alignment":290},"Automate the research path, not the organizational authority.","Source of Truth governance principle",{},{"id":613,"data":1088,"type":42,"tunes":1090},{"text":1089,"level":298},"Where This Creates Practical Value",{"anchor":617},{"id":619,"data":1092,"type":278,"tunes":1094},{"text":1093},"The architecture is useful wherever decisions depend on distributed evidence, repeated research or explainable conclusions. The strongest use cases are not defined by industry alone, but by the need for traceability, controlled access and reusable knowledge.",{},{"id":624,"data":1096,"type":356,"tunes":1122},{"content":1097,"stretched":43,"withHeadings":14},[1098,1101,1104,1107,1110,1113,1116,1119],[1099,1100],"Use case","What SoT contributes",[1102,1103],"Policy and compliance research","Traceable evidence, competing interpretations and review history",[1105,1106],"Technical decision support","Architecture evidence, alternatives, dependencies and decision rationale",[1108,1109],"Due diligence and document review","Structured findings across large source sets with provenance",[1111,1112],"Internal knowledge research","Search plus reusable reviewed conclusions instead of one-off chat answers",[1114,1115],"Incident and root-cause analysis","Evidence paths connecting observations, hypotheses and verified causes",[1117,1118],"Product and market intelligence","Multiple sources, claim comparison and explicit uncertainty",[1120,1121],"Regulated or sensitive domains","RBAC, namespaces, review states and auditable decision paths",{},{"id":653,"data":1124,"type":42,"tunes":1126},{"text":1125,"level":298},"What SoT Is — and What It Is Not",{"anchor":657},{"id":659,"data":1128,"type":679,"tunes":1146},{"cons":1129,"pros":1136,"title":1143,"layout":676,"consLabel":1144,"prosLabel":1145},[1130,1131,1132,1133,1134,1135],"A chatbot that becomes authoritative by generating fluent answers","A replacement for source quality or expert judgment","A vector database presented as a complete knowledge system","A single-model architecture","A mechanism that automatically resolves every contradiction","A substitute for organizational responsibility",[1137,1138,1139,1140,1141,1142],"An evidence-governed decision intelligence layer","A structured research and knowledge workflow","A persistent evidence and dependency graph","A controlled integration point for RAG, agents and models","A system for provenance, contradictions, review and auditability","A reusable organizational knowledge foundation","Source of Truth boundaries","SoT is not","SoT is",{},{"id":682,"data":1148,"type":42,"tunes":1150},{"text":1149,"level":298},"Current Implementation Direction",{"anchor":686},{"id":688,"data":1152,"type":694,"tunes":1156},{"date":690,"text":1153,"label":1154,"context":1155},"The current Source of Truth product work combines document ingestion, indexed retrieval, claims and evidence relationships, contradiction handling, multi-step research, evidence reasoning, namespaces, role and permission boundaries, model integration, monitored runs and reviewed knowledge states. The architecture is being developed as a modular system in which retrieval, evidence management and reasoning can also be used at different levels depending on the customer use case.","Implementation evidence","Product development and validation are active. This description does not claim customer adoption, revenue, product-market fit or enterprise scale.",{},{"id":697,"data":1158,"type":42,"tunes":1160},{"text":1159,"level":298},"Frequently Asked Questions",{"anchor":701},{"id":703,"data":1162,"type":701,"tunes":1183},{"items":1163,"title":1182},[1164,1167,1170,1173,1176,1179],{"id":707,"answer":1165,"question":1166},"RAG is one important retrieval layer inside SoT. The broader architecture adds an evidence DAG, provenance, claims, contradictions, reasoning paths, access control, review states and auditability.","Is SoT a RAG system?",{"id":711,"answer":1168,"question":1169},"DAG means Directed Acyclic Graph. In SoT it represents directed dependencies between evidence, claims, findings and conclusions while preventing circular justification. This makes reasoning paths traceable and reusable.","What does DAG mean in Source of Truth?",{"id":715,"answer":1171,"question":1172},"No. RAG retrieves candidate evidence. The DAG structures how that evidence relates to claims and conclusions. They solve different layers of the same research workflow.","Does the DAG replace RAG?",{"id":719,"answer":1174,"question":1175},"No. SoT structures evidence and reasoning, preserves uncertainty and supports review. Accepted organizational knowledge remains a governed human or workflow decision.","Does SoT decide what is true automatically?",{"id":723,"answer":1177,"question":1178},"No. The architecture separates models and agents from the evidence layer so that local, cloud or hybrid model strategies can evolve without losing the underlying knowledge structure.","Is SoT tied to one AI model or provider?",{"id":727,"answer":1180,"question":1181},"The main advantage is a reusable and auditable path from enterprise question to supported conclusion: less repeated manual research, clearer evidence, visible uncertainty, controlled AI access and better continuity of organizational knowledge.","What is the main business advantage?","Source of Truth for enterprise research and decision support",{},{"id":733,"data":1185,"type":42,"tunes":1187},{"text":1186,"level":298},"Key Terms",{"anchor":737},{"id":739,"data":1189,"type":772,"tunes":1212},{"title":1190,"entries":1191},"Source of Truth terminology",[1192,1194,1196,1198,1201,1204,1207,1210],{"term":744,"anchor":745,"definition":1193},"An evidence-governed decision intelligence platform that preserves the path from enterprise question to evidence, reasoning, conclusion and review.",{"term":418,"anchor":417,"definition":1195},"Retrieval-Augmented Generation: retrieving relevant source material and providing it as context to a generative model.",{"term":750,"anchor":420,"definition":1197},"Directed Acyclic Graph: a directed graph without cycles. In SoT it represents dependencies between evidence, claims, findings and conclusions.",{"term":1199,"anchor":754,"definition":1200},"Evidence provenance","Information showing where evidence originated and how it entered the research and reasoning path.",{"term":1202,"anchor":758,"definition":1203},"Evidence relationship","An explicit connection describing how evidence, claims or conclusions support, depend on, contradict or provide alternatives to one another.",{"term":1205,"anchor":762,"definition":1206},"Answer validity boundary","The conditions and evidence limits within which a conclusion remains supported.",{"term":1208,"anchor":766,"definition":1209},"Namespace","A logical boundary that separates knowledge by organization, tenant, project, case or domain.",{"term":769,"anchor":770,"definition":1211},"Role-Based Access Control: authorization based on assigned roles and their permitted actions or data scopes.",{},{"id":775,"data":1214,"type":42,"tunes":1216},{"text":1215,"level":298},"The Core Advantage: A Decision Can Be Traced Back to Its Evidence",{"anchor":779},{"id":781,"data":1218,"type":278,"tunes":1220},{"text":1219},"The strongest value of Source of Truth is not that AI can produce an answer faster. It is that an organization can preserve \u003Cstrong>why\u003C\u002Fstrong> an answer was reached, \u003Cstrong>which evidence\u003C\u002Fstrong> supports it, \u003Cstrong>where uncertainty remains\u003C\u002Fstrong>, \u003Cstrong>which alternatives were considered\u003C\u002Fstrong> and \u003Cstrong>who approved the result\u003C\u002Fstrong>.",{},{"id":786,"data":1222,"type":278,"tunes":1224},{"text":1223},"RAG helps the system find. The DAG helps the system structure. Research agents help the system investigate. Evidence reasoning helps the system justify. Governance and human review determine what the organization accepts. Together, these layers turn AI-assisted research into reusable decision intelligence.",{},"2.31.6","How the Source of Truth Enterprise Intelligence Platform moves beyond retrieval by combining controlled case scope, RAG, research agents, evidence reasoning, contradiction handling, RBAC, namespaces and human review into an auditable path from enterprise question to supported decision.",{"lang":7,"title":268,"content":270,"contentJson":1228,"excerpt":791},{"time":272,"blocks":1229,"version":790},[1230,1233,1236,1239,1242,1245,1248,1251,1263,1266,1269,1272,1291,1294,1305,1308,1311,1314,1326,1329,1332,1335,1338,1341,1344,1347,1350,1353,1356,1359,1364,1367,1370,1373,1376,1379,1382,1385,1388,1400,1403,1408,1411,1414,1417,1427,1430,1442,1445,1448],{"id":275,"data":1231,"type":278,"tunes":1232},{"text":277},{},{"id":281,"data":1234,"type":278,"tunes":1235},{"text":283},{},{"id":286,"data":1237,"type":291,"tunes":1238},{"text":288,"caption":289,"alignment":290},{},{"id":294,"data":1240,"type":299,"tunes":1241},{"title":296,"maxLevel":297,"minLevel":298},{},{"id":302,"data":1243,"type":42,"tunes":1244},{"text":304,"level":298},{"anchor":306},{"id":308,"data":1246,"type":278,"tunes":1247},{"text":310},{},{"id":313,"data":1249,"type":318,"tunes":1250},{"body":315,"title":316,"variant":317},{},{"id":321,"data":1252,"type":356,"tunes":1262},{"content":1253,"stretched":43,"withHeadings":14},[1254,1255,1256,1257,1258,1259,1260,1261],[325,326,327],[329,330,331],[333,334,335],[337,338,339],[341,342,343],[345,346,347],[349,350,351],[353,354,355],{},{"id":359,"data":1264,"type":42,"tunes":1265},{"text":361,"level":298},{"anchor":363},{"id":365,"data":1267,"type":278,"tunes":1268},{"text":367},{},{"id":370,"data":1270,"type":278,"tunes":1271},{"text":372},{},{"id":375,"data":1273,"type":422,"tunes":1290},{"rows":1274,"title":414,"layout":356,"columns":1287},[1275,1277,1279,1281,1283,1285],{"id":379,"label":380,"values":1276},{"dag":382,"rag":383},{"id":385,"label":386,"values":1278},{"dag":388,"rag":389},{"id":391,"label":392,"values":1280},{"dag":394,"rag":395},{"id":397,"label":398,"values":1282},{"dag":400,"rag":401},{"id":403,"label":404,"values":1284},{"dag":406,"rag":407},{"id":409,"label":410,"values":1286},{"dag":412,"rag":413},[1288,1289],{"id":417,"label":418},{"id":420,"label":421},{},{"id":425,"data":1292,"type":318,"tunes":1293},{"body":427,"title":428,"variant":429},{},{"id":432,"data":1295,"type":458,"tunes":1304},{"steps":1296,"title":456,"orientation":457},[1297,1298,1299,1300,1301,1302,1303],{"label":436,"description":437},{"label":439,"description":440},{"label":442,"description":443},{"label":445,"description":446},{"label":448,"description":449},{"label":451,"description":452},{"label":454,"description":455},{},{"id":461,"data":1306,"type":42,"tunes":1307},{"text":463,"level":298},{"anchor":465},{"id":467,"data":1309,"type":278,"tunes":1310},{"text":469},{},{"id":472,"data":1312,"type":278,"tunes":1313},{"text":474},{},{"id":477,"data":1315,"type":356,"tunes":1325},{"content":1316,"stretched":43,"withHeadings":14},[1317,1318,1319,1320,1321,1322,1323,1324],[481,482],[484,485],[487,488],[490,491],[493,494],[496,497],[499,500],[502,503],{},{"id":506,"data":1327,"type":42,"tunes":1328},{"text":508,"level":298},{"anchor":510},{"id":512,"data":1330,"type":278,"tunes":1331},{"text":514},{},{"id":517,"data":1333,"type":278,"tunes":1334},{"text":519},{},{"id":522,"data":1336,"type":42,"tunes":1337},{"text":524,"level":298},{"anchor":526},{"id":528,"data":1339,"type":278,"tunes":1340},{"text":530},{},{"id":533,"data":1342,"type":318,"tunes":1343},{"body":535,"title":536,"variant":537},{},{"id":540,"data":1345,"type":42,"tunes":1346},{"text":542,"level":298},{"anchor":544},{"id":546,"data":1348,"type":278,"tunes":1349},{"text":548},{},{"id":551,"data":1351,"type":278,"tunes":1352},{"text":553},{},{"id":556,"data":1354,"type":42,"tunes":1355},{"text":558,"level":298},{"anchor":560},{"id":562,"data":1357,"type":278,"tunes":1358},{"text":564},{},{"id":567,"data":1360,"type":577,"tunes":1363},{"meta":1361,"items":1362,"style":576},{},[571,572,573,574,575],{},{"id":580,"data":1365,"type":42,"tunes":1366},{"text":582,"level":298},{"anchor":584},{"id":586,"data":1368,"type":278,"tunes":1369},{"text":588},{},{"id":591,"data":1371,"type":278,"tunes":1372},{"text":593},{},{"id":596,"data":1374,"type":42,"tunes":1375},{"text":598,"level":298},{"anchor":600},{"id":602,"data":1377,"type":278,"tunes":1378},{"text":604},{},{"id":607,"data":1380,"type":291,"tunes":1381},{"text":609,"caption":610,"alignment":290},{},{"id":613,"data":1383,"type":42,"tunes":1384},{"text":615,"level":298},{"anchor":617},{"id":619,"data":1386,"type":278,"tunes":1387},{"text":621},{},{"id":624,"data":1389,"type":356,"tunes":1399},{"content":1390,"stretched":43,"withHeadings":14},[1391,1392,1393,1394,1395,1396,1397,1398],[628,629],[631,632],[634,635],[637,638],[640,641],[643,644],[646,647],[649,650],{},{"id":653,"data":1401,"type":42,"tunes":1402},{"text":655,"level":298},{"anchor":657},{"id":659,"data":1404,"type":679,"tunes":1407},{"cons":1405,"pros":1406,"title":675,"layout":676,"consLabel":677,"prosLabel":678},[662,663,664,665,666,667],[669,670,671,672,673,674],{},{"id":682,"data":1409,"type":42,"tunes":1410},{"text":684,"level":298},{"anchor":686},{"id":688,"data":1412,"type":694,"tunes":1413},{"date":690,"text":691,"label":692,"context":693},{},{"id":697,"data":1415,"type":42,"tunes":1416},{"text":699,"level":298},{"anchor":701},{"id":703,"data":1418,"type":701,"tunes":1426},{"items":1419,"title":730},[1420,1421,1422,1423,1424,1425],{"id":707,"answer":708,"question":709},{"id":711,"answer":712,"question":713},{"id":715,"answer":716,"question":717},{"id":719,"answer":720,"question":721},{"id":723,"answer":724,"question":725},{"id":727,"answer":728,"question":729},{},{"id":733,"data":1428,"type":42,"tunes":1429},{"text":735,"level":298},{"anchor":737},{"id":739,"data":1431,"type":772,"tunes":1441},{"title":741,"entries":1432},[1433,1434,1435,1436,1437,1438,1439,1440],{"term":744,"anchor":745,"definition":746},{"term":418,"anchor":417,"definition":748},{"term":750,"anchor":420,"definition":751},{"term":753,"anchor":754,"definition":755},{"term":757,"anchor":758,"definition":759},{"term":761,"anchor":762,"definition":763},{"term":765,"anchor":766,"definition":767},{"term":769,"anchor":770,"definition":771},{},{"id":775,"data":1443,"type":42,"tunes":1444},{"text":777,"level":298},{"anchor":779},{"id":781,"data":1446,"type":278,"tunes":1447},{"text":783},{},{"id":786,"data":1449,"type":278,"tunes":1450},{"text":788},{},"Portfolio erfolgreich abgerufen",{"items":1453,"source":1461,"manualIds":1462,"manualMatchedIds":1463},[1454],{"id":1455,"slug":1456,"title":1457,"excerpt":1458,"featuredImage":1459,"publishedAt":1460},"23","source-of-truth-enterprise-intelligence-platform","真相源企业智能平台","一个自托管、以证据为治理的企业智能平台，将公司内部知识、文档和业务数据转化为可追溯的智能，用于研究、决策、监控和受控的AI执行。","\u002Fuploads\u002F2026\u002F10\u002Fsource-of-truth-enterprise-intelligence-platform-1791975594140-vygojf.webp","2026-10-14T06:27:00.000Z","fallback",[],[]]