[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"portal-settings:stajic:fr":3,"public-menus:all":38,"post:what-is-rag-the-simplest-explanation-of-how-it-works:fr":205,"related:post:what-is-rag-the-simplest-explanation-of-how-it-works:fr:1":1795},{"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","fr","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",[50,66,79,93,103,118,133],{"id":51,"title":52,"url":60,"target":61,"icon":62,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":64,"portfolioId":10,"children":65},"item-18",{"de":53,"en":54,"es":55,"fr":56,"it":54,"ru":57,"sr":58,"zh":59},"Startseite","Home","Inicio","Accueil","Главная","Почетна","首页","\u002Ffull-stack-web-developer-munich-performance-seo-and-maintainable-builds","_self","i-lucide-home","page",111,[],{"id":67,"title":68,"url":75,"target":61,"icon":76,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":77,"portfolioId":10,"children":78},"item-22",{"de":69,"en":69,"es":70,"fr":69,"it":71,"ru":72,"sr":73,"zh":74},"Vision","Visión","Visione","Видение","Визија","想象","\u002Fueber-uns-webdesign-muenchen-webaplikation","i-lucide-eye",113,[],{"id":80,"title":81,"url":89,"target":61,"icon":90,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":91,"portfolioId":10,"children":92},"item-19",{"de":82,"en":83,"es":84,"fr":83,"it":85,"ru":86,"sr":87,"zh":88},"Leistungen","Services","Servicios","Servizi","Услуги","Услуге","服务","\u002Fservices-dienstleistungen-muenchen","i-lucide-wrench",116,[],{"id":94,"title":95,"url":99,"target":61,"icon":100,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":101,"portfolioId":10,"children":102},"item-23",{"de":96,"en":96,"es":96,"fr":96,"it":96,"ru":97,"sr":97,"zh":98},"Blog","Блог","博客","\u002Fblog","i-lucide-book-open",112,[],{"id":104,"title":105,"url":114,"target":61,"icon":115,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":116,"portfolioId":10,"children":117},"item-32",{"de":106,"en":107,"es":108,"fr":109,"it":110,"ru":111,"sr":112,"zh":113},"Neue Technologien","New Technologies","Nuevas tecnologías","Nouvelles technologies","Nuove tecnologie","Новые технологии","Нове технологије","新技术！","\u002Fneue-webtechnologien","i-lucide-sparkles",122,[],{"id":119,"title":120,"url":129,"target":61,"icon":130,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":131,"portfolioId":10,"children":132},"item-20",{"de":121,"en":122,"es":123,"fr":124,"it":125,"ru":126,"sr":127,"zh":128},"Kontakt","Contact us!","Contacto","Contact","Contatto","Контакт","Контактирајте нас","联系我们！","\u002Fcontact","i-lucide-mail",115,[],{"id":134,"title":135,"url":144,"target":61,"icon":145,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":146,"portfolioId":10,"children":147},"item-21",{"de":136,"en":137,"es":138,"fr":139,"it":140,"ru":141,"sr":142,"zh":143},"Unsere Arbeit","Our Work","Nuestro trabajo","Nos réalisations","I nostri lavori","Наши работы","Наши радови","文件夹","\u002Fportfolio","i-lucide-briefcase",114,[148,161,175,181,193],{"id":149,"title":150,"url":144,"target":61,"icon":159,"isActive":14,"type":63,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":146,"portfolioId":10,"children":160},"item-24",{"de":151,"en":152,"es":153,"fr":154,"it":155,"ru":156,"sr":157,"zh":158},"Alle Projekte","All Projects","Todos los proyectos","Tous les projets","Tutti i progetti","Все проекты","Сви пројекти","所有项目","i-lucide-grid-3x3",[],{"id":162,"title":163,"url":171,"target":61,"icon":172,"isActive":14,"type":173,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":10,"children":174},"item-29",{"de":164,"en":165,"es":166,"fr":167,"it":168,"ru":169,"sr":170,"zh":143},"Local Roots, Global Reach","Local Roots - Global Reach","Empresa local ","Entreprise locale","Azienda locale","Местная компания","Локално предузеће глобално тржиште","\u002Fportfolio\u002Flocal-roots-global-reach-communication-media-systems-for-modern-business","i-lucide-folder","custom",[],{"id":176,"title":177,"url":179,"target":61,"icon":172,"isActive":14,"type":173,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":10,"children":180},"item-28",{"de":178,"en":178,"es":178,"fr":178,"it":178,"ru":178,"sr":178,"zh":178},"Solr Suggester","\u002Fportfolio\u002Fsolr-fuzzy-suggester-und-solr-infix-suggester-abfrage-ueber-ajax-und-filterung",[],{"id":182,"title":183,"url":191,"target":61,"icon":172,"isActive":14,"type":173,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":10,"children":192},"item-27",{"de":184,"en":185,"es":186,"fr":187,"it":188,"ru":189,"sr":190,"zh":185},"Firmenwebseite SEO","Company Website SEO","Sitio web corporativo SEO","Site web d’entreprise SEO","Sito web aziendale SEO","Корпоративный сайт SEO","Пословна веб-страница SEO","\u002Fportfolio\u002Fseo-sem-branding-mobile-webseite-muenchen",[],{"id":194,"title":195,"url":203,"target":61,"icon":172,"isActive":14,"type":173,"productId":10,"categoryId":10,"shopCategoryId":10,"articleId":10,"pageId":10,"portfolioId":10,"children":204},"item-31",{"de":196,"en":197,"es":198,"fr":199,"it":200,"ru":201,"sr":202,"zh":197},"Digitalisierungsportal","Digitalization Portal","Portal de digitalización","Portail de numérisation","Portale di digitalizzazione","Портал цифровизации","Портал за дигитализацију","\u002Fportfolio\u002Fdigitalisierungsportal-archiv-museum-bibliothek-ead-lido-mets-mods",[],{"statusCode":4,"data":206,"message":1794},{"id":207,"title":208,"slug":209,"content":210,"contentJson":211,"excerpt":874,"featuredImage":875,"featuredImageAlt":876,"featuredImageCaption":10,"featuredImageTitle":10,"featuredImageCopyright":10,"featuredImageAuthor":10,"featuredImageSourceUrl":10,"featuredImageLicense":10,"featuredImageIsAiGenerated":43,"status":877,"publishedAt":878,"createdAt":879,"updatedAt":880,"seoLocalePaths":881,"categories":890,"author":915,"translations":920},"478","Qu'est-ce que le RAG ? L'explication la plus simple de son fonctionnement","what-is-rag-the-simplest-explanation-of-how-it-works","\u003Cp>Le RAG semble compliqué parce que le nom est compliqué. L'idée ne l'est pas. Le RAG signifie simplement : avant que l'IA ne réponde, elle recherche d'abord des informations pertinentes dans une source de connaissances et fournit ces informations au modèle de langage.\u003C\u002Fp>\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\">Le RAG en une phrase\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">\u003Cstrong>Le RAG est l'étape où une IA recherche dans une base de connaissances des informations utiles avant que le LLM ne rédige la réponse.\u003C\u002Fstrong>\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Cp>Imaginez un LLM comme une personne intelligente assise à un bureau. Le RAG est le bibliothécaire qui apporte la bonne page du bon livre. Le LLM lit ensuite cette page et vous répond.\u003C\u002Fp>\n\u003Cnav class=\"editorjs-toc\" data-editorjs-toc=\"true\" aria-label=\"Sommaire\">\u003Cstrong class=\"editorjs-toc__title\">Sommaire\u003C\u002Fstrong>\u003Col class=\"editorjs-toc__list editorjs-toc__list--depth-0\">\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-5\" class=\"editorjs-toc__link\">D'abord : que fait le LLM ?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-10\" class=\"editorjs-toc__link\">Ensuite : qu'est-ce que la base de connaissances ?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-15\" class=\"editorjs-toc__link\">Alors, que fait réellement le RAG ?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-19\" class=\"editorjs-toc__link\">Un exemple très simple\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-25\" class=\"editorjs-toc__link\">Maintenant la partie importante : le RAG n'est pas l'état actuel\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-30\" class=\"editorjs-toc__link\">Qu'est-ce qu'une base de données d'état ?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-36\" class=\"editorjs-toc__link\">Comment les trois éléments fonctionnent ensemble\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-40\" class=\"editorjs-toc__link\">Le RAG utilise-t-il toujours une base de données vectorielle ?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-45\" class=\"editorjs-toc__link\">Qu'est-ce qu'un embedding, en termes simples ?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-50\" class=\"editorjs-toc__link\">Le RAG n'est pas non plus de la mémoire\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-54\" class=\"editorjs-toc__link\">Un exemple de jeu réel : PUBG Ally\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-60\" class=\"editorjs-toc__link\">Un exemple complet\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-64\" class=\"editorjs-toc__link\">Pourquoi utiliser le RAG ?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-68\" class=\"editorjs-toc__link\">Ce que le RAG ne garantit pas\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-72\" class=\"editorjs-toc__link\">Le modèle mental le plus simple à retenir\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-75\" class=\"editorjs-toc__link\">Conclusion\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-79\" class=\"editorjs-toc__link\">FAQ\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-81\" class=\"editorjs-toc__link\">Glossaire\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-83\" class=\"editorjs-toc__link\">Sources primaires\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>\u003C\u002Fnav>\n\u003Ch2 id=\"section-5\">D'abord : que fait le LLM ?\u003C\u002Fh2>\n\u003Cp>Le LLM est la partie qui comprend le langage et produit du langage. Il peut lire votre question, comprendre des instructions, comparer des informations, expliquer quelque chose et rédiger une réponse.\u003C\u002Fp>\n\u003Cp>Mais le LLM ne sait pas automatiquement ce qui se trouve actuellement dans la base de données de votre entreprise, votre session de jeu, vos documents privés ou un fichier que vous avez créé il y a cinq minutes.\u003C\u002Fp>\n\u003Cp>Il ne connaît que ce qui est déjà à l'intérieur du modèle, plus toutes les informations que l'application lui fournit dans la requête en cours.\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\">Règle simple\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">Le LLM \u003Cstrong>pense et écrit\u003C\u002Fstrong>. Il ne possède pas automatiquement toutes vos données actuelles.\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"section-10\">Ensuite : qu'est-ce que la base de connaissances ?\u003C\u002Fh2>\n\u003Cp>Une base de connaissances est simplement une information que l'application peut rechercher.\u003C\u002Fp>\n\u003Cp>Elle peut contenir des PDF, des manuels, de la documentation produit, des articles d'assistance, des contrats, des règles de jeu, des données d'armes, des documents internes d'entreprise, des enregistrements de base de données ou d'autres textes.\u003C\u002Fp>\n\u003Cp>La base de connaissances peut être locale sur votre propre machine. Elle peut être sur un serveur. Elle peut être dans une base de données vectorielle. Elle peut aussi être construite à partir de fichiers normaux. Le RAG ne signifie pas Internet.\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\">Important\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">\u003Cstrong>Le RAG ne nécessite pas Internet.\u003C\u002Fstrong> Les informations peuvent être entièrement locales.\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"section-15\">Alors, que fait réellement le RAG ?\u003C\u002Fh2>\n\u003Csection class=\"editorjs-process my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">L'ensemble du processus RAG\u003C\u002Fh3>\u003Cdiv class=\"flex flex-col sm:flex-row gap-3\">\u003Cdiv class=\"editorjs-process__step min-w-0 flex-1 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\">1. Vous posez une question\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Par exemple : quelle munition cette arme utilise-t-elle ?\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__arrow shrink-0 self-center text-xl text-gray-400 rotate-90 sm:rotate-0\" aria-hidden=\"true\">→\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0 flex-1 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\">2. Le RAG recherche dans la base de connaissances\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le système cherche les petits morceaux d'information les plus pertinents pour votre question.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__arrow shrink-0 self-center text-xl text-gray-400 rotate-90 sm:rotate-0\" aria-hidden=\"true\">→\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0 flex-1 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\">3. Le RAG fournit ces morceaux au LLM\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le LLM reçoit la question plus les informations récupérées.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__arrow shrink-0 self-center text-xl text-gray-400 rotate-90 sm:rotate-0\" aria-hidden=\"true\">→\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0 flex-1 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\">4. Le LLM rédige la réponse\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Il utilise les informations récupérées comme contexte pour la réponse.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Cp>C'est ça, le RAG.\u003C\u002Fp>\n\u003Cp>Le nom complet est Retrieval-Augmented Generation. Retrieval signifie trouver les informations pertinentes. Augmented signifie ajouter ces informations au contexte du modèle. Generation signifie que le LLM rédige la réponse finale.\u003C\u002Fp>\n\u003Ch2 id=\"section-19\">Un exemple très simple\u003C\u002Fh2>\n\u003Cp>Imaginez que vous disposez d'une base de connaissances locale sur un jeu.\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\">La base de connaissances contient\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Exemple\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Armes\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">L'AKM utilise des munitions de 7,62 mm\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Objets de soin\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Le kit médical restaure la santé\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Accessoires\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Cet accessoire fonctionne avec ces armes\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Règles de la carte\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Cette zone se comporte de cette manière\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Cp>Vous demandez : « Quelles munitions l'AKM utilise-t-il ? »\u003C\u002Fp>\n\u003Cp>Le RAG parcourt la base de connaissances et trouve l'entrée concernant l'AKM. Il transmet ce petit morceau d'information au LLM. Le LLM répond alors : « L'AKM utilise des munitions de 7,62 mm. »\u003C\u002Fp>\n\u003Cp>Le LLM n'avait pas besoin de toute la base de données. Le RAG n'a apporté que la partie utile.\u003C\u002Fp>\n\u003Ch2 id=\"section-25\">Maintenant la partie importante : le RAG n'est pas l'état actuel\u003C\u002Fh2>\n\u003Cp>C'est là que beaucoup d'explications deviennent confuses.\u003C\u002Fp>\n\u003Cp>Le RAG donne généralement des connaissances à l'IA. Un système d'état donne à l'IA des faits sur ce qui est vrai à l'instant présent.\u003C\u002Fp>\n\u003Csection class=\"editorjs-comparison my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Connaissances vs état actuel\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 \u002F connaissances\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\">État actuel\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\">Arme\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Munitions\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Santé\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Ennemi\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Caside class=\"editorjs-callout editorjs-callout--warning my-6 rounded-xl border p-5 border-amber-300 bg-amber-50 dark:border-amber-900 dark:bg-amber-950\u002F20\" role=\"note\">\u003Cstrong class=\"block mb-2 text-gray-900 dark:text-gray-100\">Ne mélangez pas ces deux choses\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">Le RAG répond : \u003Cstrong>Qu'est-ce qui est généralement vrai ?\u003C\u002Fstrong>\u003Cbr>L'état répond : \u003Cstrong>Qu'est-ce qui est vrai en ce moment ?\u003C\u002Fstrong>\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"section-30\">Qu'est-ce qu'une base de données d'état ?\u003C\u002Fh2>\n\u003Cp>Une base de données d'état ou un magasin d'état est simplement un endroit où l'application conserve les faits actuels.\u003C\u002Fp>\n\u003Cp>Dans un jeu, le moteur sait déjà des choses comme votre santé, votre position, votre inventaire, vos munitions, votre mission actuelle, les objets à proximité et le statut des ennemis. Un système d'IA peut exposer certaines parties sélectionnées de cet état au modèle.\u003C\u002Fp>\n\u003Cp>Dans une application métier, la même idée pourrait être une base de données de commandes, un dossier client, un statut de projet ou la valeur actuelle d'un capteur.\u003C\u002Fp>\n\u003Cp>L'état est créé par l'application elle-même au fur et à mesure que les choses se produisent. Si vous perdez de la santé, le jeu met à jour la valeur de santé. Si vous ramassez des munitions, l'inventaire change. Si une commande est payée, le système métier modifie le statut de la commande.\u003C\u002Fp>\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\">Règle simple\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">L'application crée et met à jour \u003Cstrong>l'état\u003C\u002Fstrong>. Le RAG recherche des \u003Cstrong>connaissances\u003C\u002Fstrong>. Le LLM utilise les deux pour décider quoi dire ou faire.\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"section-36\">Comment les trois éléments fonctionnent ensemble\u003C\u002Fh2>\n\u003Csection class=\"editorjs-process my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">LLM + état + RAG\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\">1. État actuel\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">L'application indique à l'IA ce qui est vrai maintenant : santé 41 %, AKM équipé, 23 balles.\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\">2. RAG\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le système récupère des connaissances utiles : comment fonctionne l'arme, quel objet de soin est disponible, ou une règle pertinente.\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\">3. LLM\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le modèle reçoit la question, l'état actuel et les connaissances récupérées.\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\">4. Raisonnement\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le LLM combine ces entrées et décide quelle réponse ou action de haut niveau a du sens.\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\">5. Application\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Si une action est requise, l'application ou le moteur de jeu l'exécute et met à jour l'état à nouveau.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Cp>Donc l'architecture de base est :\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\">L'architecture la plus simple\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">\u003Cstrong>État = ce qui est vrai maintenant\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = connaissances utiles\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>LLM = comprend, raisonne et écrit\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Application = effectue l'action réelle\u003C\u002Fstrong>\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"section-40\">Le RAG utilise-t-il toujours une base de données vectorielle ?\u003C\u002Fh2>\n\u003Cp>Non.\u003C\u002Fp>\n\u003Cp>Une base de données vectorielle est un moyen courant de construire une recherche sémantique, mais ce n'est pas la définition du RAG.\u003C\u002Fp>\n\u003Cp>L'important est la récupération : le système trouve des informations externes pertinentes et les ajoute au contexte du LLM avant que la réponse ne soit générée.\u003C\u002Fp>\n\u003Cp>La recherche de fichiers d'OpenAI, par exemple, peut fonctionner avec des fichiers stockés dans des magasins vectoriels. Les fichiers sont découpés en morceaux plus petits afin que le système puisse récupérer les parties pertinentes pour une question. C'est une implémentation de la même idée de base.\u003C\u002Fp>\n\u003Ch2 id=\"section-45\">Qu'est-ce qu'un embedding, en termes simples ?\u003C\u002Fh2>\n\u003Cp>Vous n'avez pas besoin de comprendre les embeddings pour comprendre le RAG.\u003C\u002Fp>\n\u003Cp>Mais la version simple est la suivante : un embedding est une représentation numérique du sens. Il aide un système de recherche à trouver un texte conceptuellement similaire même lorsque les mots ne sont pas exactement les mêmes.\u003C\u002Fp>\n\u003Cp>Par exemple, une recherche par mots-clés normale peut chercher les mots exacts « réparation de voiture ». La recherche sémantique peut aussi comprendre que « réparer mon véhicule » concerne un sujet similaire.\u003C\u002Fp>\n\u003Cp>Cela rend les embeddings utiles pour le RAG, mais le RAG peut aussi utiliser la recherche par mots-clés, des requêtes de base de données ou une combinaison de plusieurs méthodes.\u003C\u002Fp>\n\u003Ch2 id=\"section-50\">Le RAG n'est pas non plus de la mémoire\u003C\u002Fh2>\n\u003Cp>La mémoire est un autre concept souvent mélangé avec le RAG.\u003C\u002Fp>\n\u003Cp>La mémoire est généralement l'information que le système conserve sur des interactions ou des événements précédents. Le RAG est le mécanisme utilisé pour récupérer les connaissances pertinentes lorsqu'elles sont nécessaires.\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\">Partie\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Signification simple\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">LLM\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">La partie qui comprend et génère le langage\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">RAG\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">La partie qui recherche les connaissances pertinentes avant la réponse\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Base de connaissances\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">L'information que le RAG peut rechercher\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">État\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Ce qui est vrai en ce moment dans l'application ou le monde\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Mémoire\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">L'information conservée des interactions ou événements précédents\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Outil \u002F action\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Quelque chose que l'IA est autorisée à appeler ou à demander à l'application de faire\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Contexte\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">L'information actuellement placée devant le LLM pour cette requête\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"section-54\">Un exemple de jeu réel : PUBG Ally\u003C\u002Fh2>\n\u003Cp>PUBG Ally est un exemple utile car il rend la différence visible.\u003C\u002Fp>\n\u003Cp>KRAFTON décrit l'état de match en direct comme une source de vérité distincte. Le jeu expose les faits actuels via des outils d'observation : arme actuelle, munitions, santé, statut de zone sûre, objets à proximité et situation de combat.\u003C\u002Fp>\n\u003Cp>La recherche de connaissances est un travail différent. Le système peut utiliser des connaissances organisées sur les armes, les accessoires, les objets et les règles. Le SDK ACE Game Agent de NVIDIA expose également une API RAG distincte pour récupérer des connaissances à partir de bases de données créées par les développeurs.\u003C\u002Fp>\n\u003Cp>Cela nous donne une séparation claire : le moteur de jeu indique ce qui se passe maintenant, la récupération fournit les connaissances pertinentes, et le modèle de langage décide de la signification des informations.\u003C\u002Fp>\n\u003Caside class=\"editorjs-referral my-6\">\u003Ca href=\"https:\u002F\u002Ffigure.rocks\u002Fblog\u002Fpubg-ally-shows-why-ai-teammates-need-two-brains-fast-reflexes-and-slow-reasoning\" target=\"_blank\" rel=\"noopener noreferrer\" class=\"flex flex-col sm:flex-row gap-4 rounded-xl border border-gray-200 dark:border-gray-700 p-4 transition hover:border-primary-500\">\u003Cdiv class=\"min-w-0 flex-1\">\u003Cstrong class=\"block text-lg text-gray-900 dark:text-gray-100\">PUBG Ally montre pourquoi les coéquipiers IA ont besoin de deux cerveaux : des réflexes rapides et un raisonnement lent\u003C\u002Fstrong>\u003Cp class=\"mt-2 text-sm text-gray-600 dark:text-gray-300\">Un exemple de jeu pratique montrant comment l'état en direct, le raisonnement linguistique et le contrôle déterministe côté jeu peuvent fonctionner ensemble.\u003C\u002Fp>\u003Cspan class=\"mt-3 inline-flex text-sm font-medium text-primary-600 dark:text-primary-400\">Lire l'article sur l'architecture de PUBG Ally →\u003C\u002Fspan>\u003C\u002Fdiv>\u003C\u002Fa>\u003C\u002Faside>\n\u003Ch2 id=\"section-60\">Un exemple complet\u003C\u002Fh2>\n\u003Cp>Imaginez que vous dites à un coéquipier IA : « Je suis à court de santé. Devrions-nous attaquer ? »\u003C\u002Fp>\n\u003Csection class=\"editorjs-process my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Ce qui se passe ensuite\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\">État\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le jeu rapporte : santé 24 %, un ennemi à proximité, deux objets de soin disponibles.\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\">RAG\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le système de connaissances récupère les règles pertinentes pour l'objet de soin et peut-être des informations sur l'arme actuelle ou une mécanique tactique.\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\">LLM\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le modèle combine votre demande, l'état actuel et les connaissances récupérées.\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\">Décision\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Il conclut que se soigner d'abord est plus sûr que d'attaquer immédiatement.\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\">Outil \u002F moteur de jeu\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">L'agent demande une action de jeu légale comme se déplacer à couvert ou utiliser l'objet de soin.\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\">Nouvel état\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Le jeu exécute l'action et rapporte la situation mise à jour à l'agent.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Cp>Le RAG ne contrôlait pas le personnage. La base de données d'état ne raisonnait pas. Le LLM ne modifiait pas directement le jeu. Chaque partie avait un seul rôle.\u003C\u002Fp>\n\u003Ch2 id=\"section-64\">Pourquoi utiliser le RAG ?\u003C\u002Fh2>\n\u003Cp>Parce que mettre chaque document, règle et enregistrement de base de données dans chaque prompt serait lent, coûteux et souvent source de confusion.\u003C\u002Fp>\n\u003Cp>Le RAG permet au système de sélectionner uniquement les informations utiles à la question actuelle.\u003C\u002Fp>\n\u003Cp>Il vous permet aussi de mettre à jour la base de connaissances sans réentraîner tout le modèle de langage. Modifiez le document ou la base de données, reconstruisez ou actualisez l'index si nécessaire, et la prochaine récupération pourra utiliser les informations plus récentes.\u003C\u002Fp>\n\u003Ch2 id=\"section-68\">Ce que le RAG ne garantit pas\u003C\u002Fh2>\n\u003Cp>Le RAG peut améliorer l'ancrage, mais il ne rend pas une réponse automatiquement correcte.\u003C\u002Fp>\n\u003Cp>L'étape de récupération peut trouver le mauvais document. Le bon document peut être obsolète. Le LLM peut mal interpréter de bonnes preuves. Ou l'état actuel peut avoir changé.\u003C\u002Fp>\n\u003Cp>Un système fiable doit donc valider séparément la récupération, la fraîcheur de l'état et le raisonnement final du modèle.\u003C\u002Fp>\n\u003Ch2 id=\"section-72\">Le modèle mental le plus simple à retenir\u003C\u002Fh2>\n\u003Csection class=\"editorjs-comparison my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Pensez à un système d'IA comme à une personne à un bureau\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\">Analogie\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\">Système d'IA\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\">Personne qui réfléchit\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Trouver un livre de référence\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Livres sur l'étagère\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Tableau de bord ou panneau d'instruments actuel\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Notes des réunions précédentes\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Faire quelque chose dans le monde réel\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\u003C\u002Fsection>\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\">Si vous ne retenez que ceci\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">\u003Cstrong>LLM = cerveau.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = bibliothécaire.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Base de connaissances = bibliothèque.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>État = ce que le tableau de bord indique en ce moment.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Outils = les mains qui peuvent réellement faire quelque chose.\u003C\u002Fstrong>\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch2 id=\"section-75\">Conclusion\u003C\u002Fh2>\n\u003Cp>Le RAG est beaucoup moins mystérieux une fois que les parties sont séparées.\u003C\u002Fp>\n\u003Cp>Le LLM comprend et génère du langage. L'application maintient l'état actuel. La base de connaissances stocke les informations. Le RAG trouve la partie utile de ces informations et la place dans le contexte du LLM. Les outils ou l'application effectuent des actions réelles.\u003C\u002Fp>\n\u003Cp>C'est l'architecture de base derrière de nombreux assistants et agents d'IA modernes.\u003C\u002Fp>\n\u003Ch2 id=\"section-79\">FAQ\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\">Le RAG en termes simples\u003C\u002Fh3>\u003Cdiv id=\"faq1\" 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\">Qu'est-ce que le RAG en termes simples ?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Le RAG est une étape où une IA recherche des informations pertinentes dans une source de connaissances avant que le modèle de langage ne rédige sa réponse.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq2\" 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\">Le RAG a-t-il besoin d'Internet ?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Non. La base de connaissances peut être entièrement locale sur votre ordinateur ou votre serveur.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq3\" 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\">Le RAG est-il la même chose qu'une base de données ?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Non. La base de données ou les fichiers contiennent les informations. Le RAG est le processus de récupération qui trouve la partie utile et la donne au LLM.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq4\" 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\">Le RAG est-il la même chose que la mémoire ?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Non. La mémoire stocke généralement les interactions ou événements précédents. Le RAG récupère les connaissances pertinentes lorsqu'elles sont nécessaires.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq5\" 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\">L'état actuel de l'application fait-il partie du RAG ?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Pas nécessairement. L'état actuel est généralement obtenu directement depuis l'application ou un magasin d'état. Le RAG se comprend mieux comme une récupération depuis une source de connaissances.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq6\" 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\">Le RAG rend-il les réponses de l'IA correctes ?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Non. Il peut fournir de meilleures preuves, mais la récupération peut encore être erronée ou obsolète et le LLM peut encore raisonner de manière incorrecte.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Ch2 id=\"section-81\">Glossaire\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\">Les termes de base\u003C\u002Fh3>\u003Cdl>\u003Cdiv id=\"llm\" 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\">LLM\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Un modèle de langage qui comprend et génère du texte et peut raisonner sur les informations placées dans son contexte.\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\">Retrieval-Augmented Generation : récupérer des informations externes pertinentes et les ajouter au contexte du modèle avant de générer une réponse.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"knowledge-base\" 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\">Base de connaissances\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Les fichiers, documents, enregistrements ou autres informations que la récupération peut parcourir.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"state\" 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\">État\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Les faits actuels d'une application, d'un système ou du monde à un moment donné.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"context\" 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\">Contexte\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Les informations actuellement fournies au modèle de langage pour une requête ou une étape de raisonnement.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"embedding\" 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\">Embedding\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Une représentation numérique du sens qui peut aider la recherche sémantique à trouver des informations conceptuellement similaires.\u003C\u002Fdd>\u003C\u002Fdiv>\u003C\u002Fdl>\u003C\u002Fsection>\n\u003Ch2 id=\"section-83\">Sources primaires\u003C\u002Fh2>\n\u003Ca href=\"https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fapi-reference\u002Fvector-stores-files\" target=\"_blank\" rel=\"noopener noreferrer\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">OpenAI — Fichiers de Vector Store\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Documentation officielle montrant comment les fichiers peuvent être attachés à des vector stores, découpés en segments et rendus disponibles pour la récupération par recherche de fichiers.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fquickstart\" target=\"_blank\" rel=\"noopener noreferrer\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">OpenAI — Guide de démarrage rapide pour développeurs\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Documentation officielle d'OpenAI décrivant des outils tels que la recherche de fichiers pour donner aux modèles accès à des informations externes.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fdeveloper.nvidia.com\u002Face-for-games\" target=\"_blank\" rel=\"noopener noreferrer\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">NVIDIA Developer — ACE for Games\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Documentation officielle de NVIDIA décrivant des API distinctes Agent, Chat et RAG pour connecter des personnages de jeu à l'état du jeu, aux connaissances contextuelles et aux actions pilotées par le modèle.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fdeveloper.nvidia.com\u002Fblog\u002Fhow-krafton-built-pubg-ally-a-co-playable-character-powered-by-nvidia-ace\u002F\" target=\"_blank\" rel=\"noopener noreferrer\" class=\"editorjs-link-tool block border border-gray-200 dark:border-gray-700 rounded-lg p-4 transition text-gray-900 dark:text-gray-100 hover:border-primary-500 hover:bg-primary-50 dark:hover:bg-gray-900 hover:text-gray-900 dark:hover:text-gray-100\">\u003Cstrong class=\"block font-semibold\">NVIDIA Developer — Comment KRAFTON a construit PUBG Ally\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Explication technique officielle séparant l'état du match en direct de la recherche de connaissances et du raisonnement du modèle de langage.\u003C\u002Fp>\u003C\u002Fa>",{"time":212,"blocks":213,"version":873},1790377591663,[214,220,228,233,241,246,251,256,261,268,273,278,283,288,295,300,320,325,330,335,340,361,366,371,376,381,386,391,422,429,434,439,444,449,454,459,464,485,490,496,501,506,511,516,521,526,531,536,541,546,551,556,561,590,595,600,605,610,615,624,629,634,655,660,665,670,675,680,685,690,695,700,705,741,747,752,757,762,767,772,802,807,831,836,846,855,864],{"id":215,"data":216,"type":218,"tunes":219},"intro",{"text":217},"Le RAG semble compliqué parce que le nom est compliqué. L'idée ne l'est pas. Le RAG signifie simplement : avant que l'IA ne réponde, elle recherche d'abord des informations pertinentes dans une source de connaissances et fournit ces informations au modèle de langage.","paragraph",{},{"id":221,"data":222,"type":226,"tunes":227},"one-sentence",{"body":223,"title":224,"variant":225},"\u003Cstrong>Le RAG est l'étape où une IA recherche dans une base de connaissances des informations utiles avant que le LLM ne rédige la réponse.\u003C\u002Fstrong>","Le RAG en une phrase","info","callout",{},{"id":229,"data":230,"type":218,"tunes":232},"analogy",{"text":231},"Imaginez un LLM comme une personne intelligente assise à un bureau. Le RAG est le bibliothécaire qui apporte la bonne page du bon livre. Le LLM lit ensuite cette page et vous répond.",{},{"id":234,"data":235,"type":239,"tunes":240},"toc",{"title":236,"maxLevel":237,"minLevel":238},"Sommaire",3,2,"tableOfContents",{},{"id":242,"data":243,"type":42,"tunes":245},"h-llm",{"text":244,"level":238},"D'abord : que fait le LLM ?",{},{"id":247,"data":248,"type":218,"tunes":250},"p-llm-1",{"text":249},"Le LLM est la partie qui comprend le langage et produit du langage. Il peut lire votre question, comprendre des instructions, comparer des informations, expliquer quelque chose et rédiger une réponse.",{},{"id":252,"data":253,"type":218,"tunes":255},"p-llm-2",{"text":254},"Mais le LLM ne sait pas automatiquement ce qui se trouve actuellement dans la base de données de votre entreprise, votre session de jeu, vos documents privés ou un fichier que vous avez créé il y a cinq minutes.",{},{"id":257,"data":258,"type":218,"tunes":260},"p-llm-3",{"text":259},"Il ne connaît que ce qui est déjà à l'intérieur du modèle, plus toutes les informations que l'application lui fournit dans la requête en cours.",{},{"id":262,"data":263,"type":226,"tunes":267},"llm-rule",{"body":264,"title":265,"variant":266},"Le LLM \u003Cstrong>pense et écrit\u003C\u002Fstrong>. Il ne possède pas automatiquement toutes vos données actuelles.","Règle simple","note",{},{"id":269,"data":270,"type":42,"tunes":272},"h-kb",{"text":271,"level":238},"Ensuite : qu'est-ce que la base de connaissances ?",{},{"id":274,"data":275,"type":218,"tunes":277},"p-kb-1",{"text":276},"Une base de connaissances est simplement une information que l'application peut rechercher.",{},{"id":279,"data":280,"type":218,"tunes":282},"p-kb-2",{"text":281},"Elle peut contenir des PDF, des manuels, de la documentation produit, des articles d'assistance, des contrats, des règles de jeu, des données d'armes, des documents internes d'entreprise, des enregistrements de base de données ou d'autres textes.",{},{"id":284,"data":285,"type":218,"tunes":287},"p-kb-3",{"text":286},"La base de connaissances peut être locale sur votre propre machine. Elle peut être sur un serveur. Elle peut être dans une base de données vectorielle. Elle peut aussi être construite à partir de fichiers normaux. Le RAG ne signifie pas Internet.",{},{"id":289,"data":290,"type":226,"tunes":294},"no-internet",{"body":291,"title":292,"variant":293},"\u003Cstrong>Le RAG ne nécessite pas Internet.\u003C\u002Fstrong> Les informations peuvent être entièrement locales.","Important","success",{},{"id":296,"data":297,"type":42,"tunes":299},"h-rag",{"text":298,"level":238},"Alors, que fait réellement le RAG ?",{},{"id":301,"data":302,"type":318,"tunes":319},"rag-flow",{"steps":303,"title":316,"orientation":317},[304,307,310,313],{"label":305,"description":306},"1. Vous posez une question","Par exemple : quelle munition cette arme utilise-t-elle ?",{"label":308,"description":309},"2. Le RAG recherche dans la base de connaissances","Le système cherche les petits morceaux d'information les plus pertinents pour votre question.",{"label":311,"description":312},"3. Le RAG fournit ces morceaux au LLM","Le LLM reçoit la question plus les informations récupérées.",{"label":314,"description":315},"4. Le LLM rédige la réponse","Il utilise les informations récupérées comme contexte pour la réponse.","L'ensemble du processus RAG","auto","processFlow",{},{"id":321,"data":322,"type":218,"tunes":324},"rag-that-is-it",{"text":323},"C'est ça, le RAG.",{},{"id":326,"data":327,"type":218,"tunes":329},"rag-name",{"text":328},"Le nom complet est Retrieval-Augmented Generation. Retrieval signifie trouver les informations pertinentes. Augmented signifie ajouter ces informations au contexte du modèle. Generation signifie que le LLM rédige la réponse finale.",{},{"id":331,"data":332,"type":42,"tunes":334},"h-example",{"text":333,"level":238},"Un exemple très simple",{},{"id":336,"data":337,"type":218,"tunes":339},"p-ex-1",{"text":338},"Imaginez que vous disposez d'une base de connaissances locale sur un jeu.",{},{"id":341,"data":342,"type":359,"tunes":360},"kb-table",{"content":343,"stretched":43,"withHeadings":14},[344,347,350,353,356],[345,346],"La base de connaissances contient","Exemple",[348,349],"Armes","L'AKM utilise des munitions de 7,62 mm",[351,352],"Objets de soin","Le kit médical restaure la santé",[354,355],"Accessoires","Cet accessoire fonctionne avec ces armes",[357,358],"Règles de la carte","Cette zone se comporte de cette manière","table",{},{"id":362,"data":363,"type":218,"tunes":365},"p-ex-2",{"text":364},"Vous demandez : « Quelles munitions l'AKM utilise-t-il ? »",{},{"id":367,"data":368,"type":218,"tunes":370},"p-ex-3",{"text":369},"Le RAG parcourt la base de connaissances et trouve l'entrée concernant l'AKM. Il transmet ce petit morceau d'information au LLM. Le LLM répond alors : « L'AKM utilise des munitions de 7,62 mm. »",{},{"id":372,"data":373,"type":218,"tunes":375},"p-ex-4",{"text":374},"Le LLM n'avait pas besoin de toute la base de données. Le RAG n'a apporté que la partie utile.",{},{"id":377,"data":378,"type":42,"tunes":380},"h-state",{"text":379,"level":238},"Maintenant la partie importante : le RAG n'est pas l'état actuel",{},{"id":382,"data":383,"type":218,"tunes":385},"p-state-1",{"text":384},"C'est là que beaucoup d'explications deviennent confuses.",{},{"id":387,"data":388,"type":218,"tunes":390},"p-state-2",{"text":389},"Le RAG donne généralement des connaissances à l'IA. Un système d'état donne à l'IA des faits sur ce qui est vrai à l'instant présent.",{},{"id":392,"data":393,"type":420,"tunes":421},"knowledge-state",{"rows":394,"title":412,"layout":359,"columns":413},[395,400,404,408],{"id":396,"label":397,"values":398},"weapon","Arme",[399,399],"",{"id":401,"label":402,"values":403},"ammo","Munitions",[399,399],{"id":405,"label":406,"values":407},"health","Santé",[399,399],{"id":409,"label":410,"values":411},"enemy","Ennemi",[399,399],"Connaissances vs état actuel",[414,417],{"id":415,"label":416},"knowledge","RAG \u002F connaissances",{"id":418,"label":419},"state","État actuel","comparison",{},{"id":423,"data":424,"type":226,"tunes":428},"dont-mix",{"body":425,"title":426,"variant":427},"Le RAG répond : \u003Cstrong>Qu'est-ce qui est généralement vrai ?\u003C\u002Fstrong>\u003Cbr>L'état répond : \u003Cstrong>Qu'est-ce qui est vrai en ce moment ?\u003C\u002Fstrong>","Ne mélangez pas ces deux choses","warning",{},{"id":430,"data":431,"type":42,"tunes":433},"h-state-db",{"text":432,"level":238},"Qu'est-ce qu'une base de données d'état ?",{},{"id":435,"data":436,"type":218,"tunes":438},"p-statedb-1",{"text":437},"Une base de données d'état ou un magasin d'état est simplement un endroit où l'application conserve les faits actuels.",{},{"id":440,"data":441,"type":218,"tunes":443},"p-statedb-2",{"text":442},"Dans un jeu, le moteur sait déjà des choses comme votre santé, votre position, votre inventaire, vos munitions, votre mission actuelle, les objets à proximité et le statut des ennemis. Un système d'IA peut exposer certaines parties sélectionnées de cet état au modèle.",{},{"id":445,"data":446,"type":218,"tunes":448},"p-statedb-3",{"text":447},"Dans une application métier, la même idée pourrait être une base de données de commandes, un dossier client, un statut de projet ou la valeur actuelle d'un capteur.",{},{"id":450,"data":451,"type":218,"tunes":453},"p-statedb-4",{"text":452},"L'état est créé par l'application elle-même au fur et à mesure que les choses se produisent. Si vous perdez de la santé, le jeu met à jour la valeur de santé. Si vous ramassez des munitions, l'inventaire change. Si une commande est payée, le système métier modifie le statut de la commande.",{},{"id":455,"data":456,"type":226,"tunes":458},"state-rule",{"body":457,"title":265,"variant":225},"L'application crée et met à jour \u003Cstrong>l'état\u003C\u002Fstrong>. Le RAG recherche des \u003Cstrong>connaissances\u003C\u002Fstrong>. Le LLM utilise les deux pour décider quoi dire ou faire.",{},{"id":460,"data":461,"type":42,"tunes":463},"h-together",{"text":462,"level":238},"Comment les trois éléments fonctionnent ensemble",{},{"id":465,"data":466,"type":318,"tunes":484},"together-flow",{"steps":467,"title":483,"orientation":317},[468,471,474,477,480],{"label":469,"description":470},"1. État actuel","L'application indique à l'IA ce qui est vrai maintenant : santé 41 %, AKM équipé, 23 balles.",{"label":472,"description":473},"2. RAG","Le système récupère des connaissances utiles : comment fonctionne l'arme, quel objet de soin est disponible, ou une règle pertinente.",{"label":475,"description":476},"3. LLM","Le modèle reçoit la question, l'état actuel et les connaissances récupérées.",{"label":478,"description":479},"4. Raisonnement","Le LLM combine ces entrées et décide quelle réponse ou action de haut niveau a du sens.",{"label":481,"description":482},"5. Application","Si une action est requise, l'application ou le moteur de jeu l'exécute et met à jour l'état à nouveau.","LLM + état + RAG",{},{"id":486,"data":487,"type":218,"tunes":489},"p-arch-intro",{"text":488},"Donc l'architecture de base est :",{},{"id":491,"data":492,"type":226,"tunes":495},"simple-architecture",{"body":493,"title":494,"variant":266},"\u003Cstrong>État = ce qui est vrai maintenant\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = connaissances utiles\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>LLM = comprend, raisonne et écrit\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Application = effectue l'action réelle\u003C\u002Fstrong>","L'architecture la plus simple",{},{"id":497,"data":498,"type":42,"tunes":500},"h-vector",{"text":499,"level":238},"Le RAG utilise-t-il toujours une base de données vectorielle ?",{},{"id":502,"data":503,"type":218,"tunes":505},"p-vector-1",{"text":504},"Non.",{},{"id":507,"data":508,"type":218,"tunes":510},"p-vector-2",{"text":509},"Une base de données vectorielle est un moyen courant de construire une recherche sémantique, mais ce n'est pas la définition du RAG.",{},{"id":512,"data":513,"type":218,"tunes":515},"p-vector-3",{"text":514},"L'important est la récupération : le système trouve des informations externes pertinentes et les ajoute au contexte du LLM avant que la réponse ne soit générée.",{},{"id":517,"data":518,"type":218,"tunes":520},"p-vector-4",{"text":519},"La recherche de fichiers d'OpenAI, par exemple, peut fonctionner avec des fichiers stockés dans des magasins vectoriels. Les fichiers sont découpés en morceaux plus petits afin que le système puisse récupérer les parties pertinentes pour une question. C'est une implémentation de la même idée de base.",{},{"id":522,"data":523,"type":42,"tunes":525},"h-embedding",{"text":524,"level":238},"Qu'est-ce qu'un embedding, en termes simples ?",{},{"id":527,"data":528,"type":218,"tunes":530},"p-emb-1",{"text":529},"Vous n'avez pas besoin de comprendre les embeddings pour comprendre le RAG.",{},{"id":532,"data":533,"type":218,"tunes":535},"p-emb-2",{"text":534},"Mais la version simple est la suivante : un embedding est une représentation numérique du sens. Il aide un système de recherche à trouver un texte conceptuellement similaire même lorsque les mots ne sont pas exactement les mêmes.",{},{"id":537,"data":538,"type":218,"tunes":540},"p-emb-3",{"text":539},"Par exemple, une recherche par mots-clés normale peut chercher les mots exacts « réparation de voiture ». La recherche sémantique peut aussi comprendre que « réparer mon véhicule » concerne un sujet similaire.",{},{"id":542,"data":543,"type":218,"tunes":545},"p-emb-4",{"text":544},"Cela rend les embeddings utiles pour le RAG, mais le RAG peut aussi utiliser la recherche par mots-clés, des requêtes de base de données ou une combinaison de plusieurs méthodes.",{},{"id":547,"data":548,"type":42,"tunes":550},"h-memory",{"text":549,"level":238},"Le RAG n'est pas non plus de la mémoire",{},{"id":552,"data":553,"type":218,"tunes":555},"p-memory-1",{"text":554},"La mémoire est un autre concept souvent mélangé avec le RAG.",{},{"id":557,"data":558,"type":218,"tunes":560},"p-memory-2",{"text":559},"La mémoire est généralement l'information que le système conserve sur des interactions ou des événements précédents. Le RAG est le mécanisme utilisé pour récupérer les connaissances pertinentes lorsqu'elles sont nécessaires.",{},{"id":562,"data":563,"type":359,"tunes":589},"parts-table",{"content":564,"stretched":43,"withHeadings":14},[565,568,571,574,577,580,583,586],[566,567],"Partie","Signification simple",[569,570],"LLM","La partie qui comprend et génère le langage",[572,573],"RAG","La partie qui recherche les connaissances pertinentes avant la réponse",[575,576],"Base de connaissances","L'information que le RAG peut rechercher",[578,579],"État","Ce qui est vrai en ce moment dans l'application ou le monde",[581,582],"Mémoire","L'information conservée des interactions ou événements précédents",[584,585],"Outil \u002F action","Quelque chose que l'IA est autorisée à appeler ou à demander à l'application de faire",[587,588],"Contexte","L'information actuellement placée devant le LLM pour cette requête",{},{"id":591,"data":592,"type":42,"tunes":594},"h-pubg",{"text":593,"level":238},"Un exemple de jeu réel : PUBG Ally",{},{"id":596,"data":597,"type":218,"tunes":599},"p-pubg-1",{"text":598},"PUBG Ally est un exemple utile car il rend la différence visible.",{},{"id":601,"data":602,"type":218,"tunes":604},"p-pubg-2",{"text":603},"KRAFTON décrit l'état de match en direct comme une source de vérité distincte. Le jeu expose les faits actuels via des outils d'observation : arme actuelle, munitions, santé, statut de zone sûre, objets à proximité et situation de combat.",{},{"id":606,"data":607,"type":218,"tunes":609},"p-pubg-3",{"text":608},"La recherche de connaissances est un travail différent. Le système peut utiliser des connaissances organisées sur les armes, les accessoires, les objets et les règles. Le SDK ACE Game Agent de NVIDIA expose également une API RAG distincte pour récupérer des connaissances à partir de bases de données créées par les développeurs.",{},{"id":611,"data":612,"type":218,"tunes":614},"p-pubg-4",{"text":613},"Cela nous donne une séparation claire : le moteur de jeu indique ce qui se passe maintenant, la récupération fournit les connaissances pertinentes, et le modèle de langage décide de la signification des informations.",{},{"id":616,"data":617,"type":622,"tunes":623},"ref-pubg",{"url":618,"title":619,"excerpt":620,"ctaLabel":621},"https:\u002F\u002Ffigure.rocks\u002Fblog\u002Fpubg-ally-shows-why-ai-teammates-need-two-brains-fast-reflexes-and-slow-reasoning","PUBG Ally montre pourquoi les coéquipiers IA ont besoin de deux cerveaux : des réflexes rapides et un raisonnement lent","Un exemple de jeu pratique montrant comment l'état en direct, le raisonnement linguistique et le contrôle déterministe côté jeu peuvent fonctionner ensemble.","Lire l'article sur l'architecture de PUBG Ally","referralArticle",{},{"id":625,"data":626,"type":42,"tunes":628},"h-complete",{"text":627,"level":238},"Un exemple complet",{},{"id":630,"data":631,"type":218,"tunes":633},"p-complete-1",{"text":632},"Imaginez que vous dites à un coéquipier IA : « Je suis à court de santé. Devrions-nous attaquer ? »",{},{"id":635,"data":636,"type":318,"tunes":654},"complete-flow",{"steps":637,"title":653,"orientation":317},[638,640,642,644,647,650],{"label":578,"description":639},"Le jeu rapporte : santé 24 %, un ennemi à proximité, deux objets de soin disponibles.",{"label":572,"description":641},"Le système de connaissances récupère les règles pertinentes pour l'objet de soin et peut-être des informations sur l'arme actuelle ou une mécanique tactique.",{"label":569,"description":643},"Le modèle combine votre demande, l'état actuel et les connaissances récupérées.",{"label":645,"description":646},"Décision","Il conclut que se soigner d'abord est plus sûr que d'attaquer immédiatement.",{"label":648,"description":649},"Outil \u002F moteur de jeu","L'agent demande une action de jeu légale comme se déplacer à couvert ou utiliser l'objet de soin.",{"label":651,"description":652},"Nouvel état","Le jeu exécute l'action et rapporte la situation mise à jour à l'agent.","Ce qui se passe ensuite",{},{"id":656,"data":657,"type":218,"tunes":659},"p-complete-2",{"text":658},"Le RAG ne contrôlait pas le personnage. La base de données d'état ne raisonnait pas. Le LLM ne modifiait pas directement le jeu. Chaque partie avait un seul rôle.",{},{"id":661,"data":662,"type":42,"tunes":664},"h-why",{"text":663,"level":238},"Pourquoi utiliser le RAG ?",{},{"id":666,"data":667,"type":218,"tunes":669},"p-why-1",{"text":668},"Parce que mettre chaque document, règle et enregistrement de base de données dans chaque prompt serait lent, coûteux et souvent source de confusion.",{},{"id":671,"data":672,"type":218,"tunes":674},"p-why-2",{"text":673},"Le RAG permet au système de sélectionner uniquement les informations utiles à la question actuelle.",{},{"id":676,"data":677,"type":218,"tunes":679},"p-why-3",{"text":678},"Il vous permet aussi de mettre à jour la base de connaissances sans réentraîner tout le modèle de langage. Modifiez le document ou la base de données, reconstruisez ou actualisez l'index si nécessaire, et la prochaine récupération pourra utiliser les informations plus récentes.",{},{"id":681,"data":682,"type":42,"tunes":684},"h-not-guarantee",{"text":683,"level":238},"Ce que le RAG ne garantit pas",{},{"id":686,"data":687,"type":218,"tunes":689},"p-not-1",{"text":688},"Le RAG peut améliorer l'ancrage, mais il ne rend pas une réponse automatiquement correcte.",{},{"id":691,"data":692,"type":218,"tunes":694},"p-not-2",{"text":693},"L'étape de récupération peut trouver le mauvais document. Le bon document peut être obsolète. Le LLM peut mal interpréter de bonnes preuves. Ou l'état actuel peut avoir changé.",{},{"id":696,"data":697,"type":218,"tunes":699},"p-not-3",{"text":698},"Un système fiable doit donc valider séparément la récupération, la fraîcheur de l'état et le raisonnement final du modèle.",{},{"id":701,"data":702,"type":42,"tunes":704},"h-mental",{"text":703,"level":238},"Le modèle mental le plus simple à retenir",{},{"id":706,"data":707,"type":420,"tunes":740},"mental-table",{"rows":708,"title":733,"layout":359,"columns":734},[709,713,717,721,725,729],{"id":710,"label":711,"values":712},"brain","Personne qui réfléchit",[399,399],{"id":714,"label":715,"values":716},"library","Trouver un livre de référence",[399,399],{"id":718,"label":719,"values":720},"books","Livres sur l'étagère",[399,399],{"id":722,"label":723,"values":724},"dashboard","Tableau de bord ou panneau d'instruments actuel",[399,399],{"id":726,"label":727,"values":728},"notes","Notes des réunions précédentes",[399,399],{"id":730,"label":731,"values":732},"hands","Faire quelque chose dans le monde réel",[399,399],"Pensez à un système d'IA comme à une personne à un bureau",[735,737],{"id":229,"label":736},"Analogie",{"id":738,"label":739},"system","Système d'IA",{},{"id":742,"data":743,"type":226,"tunes":746},"remember",{"body":744,"title":745,"variant":293},"\u003Cstrong>LLM = cerveau.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = bibliothécaire.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Base de connaissances = bibliothèque.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>État = ce que le tableau de bord indique en ce moment.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Outils = les mains qui peuvent réellement faire quelque chose.\u003C\u002Fstrong>","Si vous ne retenez que ceci",{},{"id":748,"data":749,"type":42,"tunes":751},"h-conclusion",{"text":750,"level":238},"Conclusion",{},{"id":753,"data":754,"type":218,"tunes":756},"p-conc-1",{"text":755},"Le RAG est beaucoup moins mystérieux une fois que les parties sont séparées.",{},{"id":758,"data":759,"type":218,"tunes":761},"p-conc-2",{"text":760},"Le LLM comprend et génère du langage. L'application maintient l'état actuel. La base de connaissances stocke les informations. Le RAG trouve la partie utile de ces informations et la place dans le contexte du LLM. Les outils ou l'application effectuent des actions réelles.",{},{"id":763,"data":764,"type":218,"tunes":766},"p-conc-3",{"text":765},"C'est l'architecture de base derrière de nombreux assistants et agents d'IA modernes.",{},{"id":768,"data":769,"type":42,"tunes":771},"h-faq",{"text":770,"level":238},"FAQ",{},{"id":773,"data":774,"type":773,"tunes":801},"faq",{"items":775,"title":800},[776,780,784,788,792,796],{"id":777,"answer":778,"question":779},"faq1","Le RAG est une étape où une IA recherche des informations pertinentes dans une source de connaissances avant que le modèle de langage ne rédige sa réponse.","Qu'est-ce que le RAG en termes simples ?",{"id":781,"answer":782,"question":783},"faq2","Non. La base de connaissances peut être entièrement locale sur votre ordinateur ou votre serveur.","Le RAG a-t-il besoin d'Internet ?",{"id":785,"answer":786,"question":787},"faq3","Non. La base de données ou les fichiers contiennent les informations. Le RAG est le processus de récupération qui trouve la partie utile et la donne au LLM.","Le RAG est-il la même chose qu'une base de données ?",{"id":789,"answer":790,"question":791},"faq4","Non. La mémoire stocke généralement les interactions ou événements précédents. Le RAG récupère les connaissances pertinentes lorsqu'elles sont nécessaires.","Le RAG est-il la même chose que la mémoire ?",{"id":793,"answer":794,"question":795},"faq5","Pas nécessairement. L'état actuel est généralement obtenu directement depuis l'application ou un magasin d'état. Le RAG se comprend mieux comme une récupération depuis une source de connaissances.","L'état actuel de l'application fait-il partie du RAG ?",{"id":797,"answer":798,"question":799},"faq6","Non. Il peut fournir de meilleures preuves, mais la récupération peut encore être erronée ou obsolète et le LLM peut encore raisonner de manière incorrecte.","Le RAG rend-il les réponses de l'IA correctes ?","Le RAG en termes simples",{},{"id":803,"data":804,"type":42,"tunes":806},"h-glossary",{"text":805,"level":238},"Glossaire",{},{"id":808,"data":809,"type":808,"tunes":830},"glossary",{"title":810,"entries":811},"Les termes de base",[812,815,818,821,823,826],{"term":569,"anchor":813,"definition":814},"llm","Un modèle de langage qui comprend et génère du texte et peut raisonner sur les informations placées dans son contexte.",{"term":572,"anchor":816,"definition":817},"rag","Retrieval-Augmented Generation : récupérer des informations externes pertinentes et les ajouter au contexte du modèle avant de générer une réponse.",{"term":575,"anchor":819,"definition":820},"knowledge-base","Les fichiers, documents, enregistrements ou autres informations que la récupération peut parcourir.",{"term":578,"anchor":418,"definition":822},"Les faits actuels d'une application, d'un système ou du monde à un moment donné.",{"term":587,"anchor":824,"definition":825},"context","Les informations actuellement fournies au modèle de langage pour une requête ou une étape de raisonnement.",{"term":827,"anchor":828,"definition":829},"Embedding","embedding","Une représentation numérique du sens qui peut aider la recherche sémantique à trouver des informations conceptuellement similaires.",{},{"id":832,"data":833,"type":42,"tunes":835},"h-sources",{"text":834,"level":238},"Sources primaires",{},{"id":837,"data":838,"type":844,"tunes":845},"src-openai-vector",{"link":839,"meta":840},"https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fapi-reference\u002Fvector-stores-files",{"image":841,"title":842,"description":843},{"url":399},"OpenAI — Fichiers de Vector Store","Documentation officielle montrant comment les fichiers peuvent être attachés à des vector stores, découpés en segments et rendus disponibles pour la récupération par recherche de fichiers.","linkTool",{},{"id":847,"data":848,"type":844,"tunes":854},"src-openai-quickstart",{"link":849,"meta":850},"https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fquickstart",{"image":851,"title":852,"description":853},{"url":399},"OpenAI — Guide de démarrage rapide pour développeurs","Documentation officielle d'OpenAI décrivant des outils tels que la recherche de fichiers pour donner aux modèles accès à des informations externes.",{},{"id":856,"data":857,"type":844,"tunes":863},"src-nvidia-ace",{"link":858,"meta":859},"https:\u002F\u002Fdeveloper.nvidia.com\u002Face-for-games",{"image":860,"title":861,"description":862},{"url":399},"NVIDIA Developer — ACE for Games","Documentation officielle de NVIDIA décrivant des API distinctes Agent, Chat et RAG pour connecter des personnages de jeu à l'état du jeu, aux connaissances contextuelles et aux actions pilotées par le modèle.",{},{"id":865,"data":866,"type":844,"tunes":872},"src-nvidia-pubg",{"link":867,"meta":868},"https:\u002F\u002Fdeveloper.nvidia.com\u002Fblog\u002Fhow-krafton-built-pubg-ally-a-co-playable-character-powered-by-nvidia-ace\u002F",{"image":869,"title":870,"description":871},{"url":399},"NVIDIA Developer — Comment KRAFTON a construit PUBG Ally","Explication technique officielle séparant l'état du match en direct de la recherche de connaissances et du raisonnement du modèle de langage.",{},"2.31","Le RAG semble compliqué, mais l'idée est simple : avant qu'une IA ne réponde, elle recherche d'abord des informations utiles dans une source de connaissances et transmet ces informations au modèle de langage. Ce guide explique le RAG, les LLM, l'état, la mémoire et les outils à l'aide d'un modèle mental simple.","\u002Fuploads\u002F2026\u002F09\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works-1790377492124-khjagt.webp","what-is-rag-the-simplest-explanation-of-how-it-works-1790377492124-khjagt","PUBLISHED","2026-09-25T19:03:00.000Z","2026-09-25T23:03:13.651Z","2026-09-25T23:41:17.272Z",{"en":882,"de":883,"sr":884,"es":885,"fr":886,"it":887,"ru":888,"zh":889},"\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works","\u002Fde\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works","\u002Fsr\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works","\u002Fes\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works","\u002Ffr\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works","\u002Fit\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works","\u002Fru\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works","\u002Fzh\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works",[891,895,899,903,907,911],{"id":892,"name":893,"slug":894},46,"Aperçu","overview",{"id":896,"name":897,"slug":898},57,"Limites des données","data-boundaries",{"id":900,"name":901,"slug":902},51,"Anti-patterns","anti-patterns",{"id":904,"name":905,"slug":906},58,"Évaluation et garde-fous qualité","evaluation",{"id":908,"name":909,"slug":910},56,"Portefeuille de cas d’usage","use-case-portfolio",{"id":912,"name":913,"slug":914},60,"Contrôles de coût et latence","cost-and-latency",{"id":916,"login":917,"email":918,"displayName":919},"20","rooth8233","aleksandar@stajic.de","Aleksandar Stajić",[921,1447],{"lang":922,"title":923,"content":924,"contentJson":925,"excerpt":1446},"en","What Is RAG? The Simplest Explanation of How It Works","{\"time\":1790377494031,\"blocks\":[{\"id\":\"intro\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG sounds complicated because the name is complicated. The idea is not. RAG simply means: before the AI answers, it first looks up relevant information from a knowledge source and gives that information to the language model.\"},\"tunes\":{}},{\"id\":\"one-sentence\",\"type\":\"callout\",\"data\":{\"variant\":\"info\",\"title\":\"RAG in one sentence\",\"body\":\"\u003Cstrong>RAG is the step where an AI searches a knowledge base for useful information before the LLM writes the answer.\u003C\u002Fstrong>\"},\"tunes\":{}},{\"id\":\"analogy\",\"type\":\"paragraph\",\"data\":{\"text\":\"Think of an LLM as a smart person sitting at a desk. RAG is the librarian who brings the right page from the right book. The LLM then reads that page and answers you.\"},\"tunes\":{}},{\"id\":\"toc\",\"type\":\"tableOfContents\",\"data\":{\"title\":\"Contents\",\"minLevel\":2,\"maxLevel\":3},\"tunes\":{}},{\"id\":\"h-llm\",\"type\":\"header\",\"data\":{\"text\":\"First: what does the LLM do?\",\"level\":2},\"tunes\":{}},{\"id\":\"p-llm-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"The LLM is the part that understands language and produces language. It can read your question, understand instructions, compare information, explain something and write an answer.\"},\"tunes\":{}},{\"id\":\"p-llm-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"But the LLM does not automatically know what is currently inside your company database, your game session, your private documents or a file you created five minutes ago.\"},\"tunes\":{}},{\"id\":\"p-llm-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"It only knows what is already inside the model plus whatever information the application gives it in the current request.\"},\"tunes\":{}},{\"id\":\"llm-rule\",\"type\":\"callout\",\"data\":{\"variant\":\"note\",\"title\":\"Simple rule\",\"body\":\"The LLM \u003Cstrong>thinks and writes\u003C\u002Fstrong>. It does not automatically own all of your current data.\"},\"tunes\":{}},{\"id\":\"h-kb\",\"type\":\"header\",\"data\":{\"text\":\"Then: what is the knowledge base?\",\"level\":2},\"tunes\":{}},{\"id\":\"p-kb-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"A knowledge base is simply information the application can search.\"},\"tunes\":{}},{\"id\":\"p-kb-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"It could contain PDFs, manuals, product documentation, support articles, contracts, game rules, weapon data, internal company documents, database records or other text.\"},\"tunes\":{}},{\"id\":\"p-kb-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"The knowledge base can be local on your own machine. It can be on a server. It can be in a vector database. It can also be built from normal files. RAG does not mean Internet.\"},\"tunes\":{}},{\"id\":\"no-internet\",\"type\":\"callout\",\"data\":{\"variant\":\"success\",\"title\":\"Important\",\"body\":\"\u003Cstrong>RAG does not require the Internet.\u003C\u002Fstrong> The information can be completely local.\"},\"tunes\":{}},{\"id\":\"h-rag\",\"type\":\"header\",\"data\":{\"text\":\"So what does RAG actually do?\",\"level\":2},\"tunes\":{}},{\"id\":\"rag-flow\",\"type\":\"processFlow\",\"data\":{\"title\":\"The whole RAG process\",\"orientation\":\"auto\",\"steps\":[{\"label\":\"1. You ask a question\",\"description\":\"For example: Which ammunition does this weapon use?\"},{\"label\":\"2. RAG searches the knowledge base\",\"description\":\"The system looks for the small pieces of information most relevant to your question.\"},{\"label\":\"3. RAG gives those pieces to the LLM\",\"description\":\"The LLM receives the question plus the retrieved information.\"},{\"label\":\"4. The LLM writes the answer\",\"description\":\"It uses the retrieved information as context for the response.\"}]},\"tunes\":{}},{\"id\":\"rag-that-is-it\",\"type\":\"paragraph\",\"data\":{\"text\":\"That is RAG.\"},\"tunes\":{}},{\"id\":\"rag-name\",\"type\":\"paragraph\",\"data\":{\"text\":\"The full name is Retrieval-Augmented Generation. Retrieval means finding the relevant information. Augmented means adding that information to the model's context. Generation means the LLM writes the final answer.\"},\"tunes\":{}},{\"id\":\"h-example\",\"type\":\"header\",\"data\":{\"text\":\"A very simple example\",\"level\":2},\"tunes\":{}},{\"id\":\"p-ex-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Imagine you have a local knowledge base about a game.\"},\"tunes\":{}},{\"id\":\"kb-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Knowledge base contains\",\"Example\"],[\"Weapons\",\"AKM uses 7.62 mm ammunition\"],[\"Healing items\",\"Med Kit restores health\"],[\"Attachments\",\"This attachment works with these weapons\"],[\"Map rules\",\"This zone behaves in this way\"]]},\"tunes\":{}},{\"id\":\"p-ex-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"You ask: “Which ammunition does the AKM use?”\"},\"tunes\":{}},{\"id\":\"p-ex-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG searches the knowledge base and finds the entry about the AKM. It gives that small piece of information to the LLM. The LLM then answers: “The AKM uses 7.62 mm ammunition.”\"},\"tunes\":{}},{\"id\":\"p-ex-4\",\"type\":\"paragraph\",\"data\":{\"text\":\"The LLM did not need the entire database. RAG only brought the useful part.\"},\"tunes\":{}},{\"id\":\"h-state\",\"type\":\"header\",\"data\":{\"text\":\"Now the important part: RAG is not the current state\",\"level\":2},\"tunes\":{}},{\"id\":\"p-state-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"This is where many explanations become confusing.\"},\"tunes\":{}},{\"id\":\"p-state-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG usually gives the AI knowledge. A state system gives the AI facts about what is true right now.\"},\"tunes\":{}},{\"id\":\"knowledge-state\",\"type\":\"comparison\",\"data\":{\"title\":\"Knowledge vs current state\",\"layout\":\"table\",\"columns\":[{\"id\":\"knowledge\",\"label\":\"RAG \u002F knowledge\"},{\"id\":\"state\",\"label\":\"Current state\"}],\"rows\":[{\"id\":\"weapon\",\"label\":\"Weapon\",\"values\":[\"\",\"\"]},{\"id\":\"ammo\",\"label\":\"Ammunition\",\"values\":[\"\",\"\"]},{\"id\":\"health\",\"label\":\"Health\",\"values\":[\"\",\"\"]},{\"id\":\"enemy\",\"label\":\"Enemy\",\"values\":[\"\",\"\"]}]},\"tunes\":{}},{\"id\":\"dont-mix\",\"type\":\"callout\",\"data\":{\"variant\":\"warning\",\"title\":\"Do not mix these two\",\"body\":\"RAG answers: \u003Cstrong>What is generally true?\u003C\u002Fstrong>\u003Cbr>State answers: \u003Cstrong>What is true right now?\u003C\u002Fstrong>\"},\"tunes\":{}},{\"id\":\"h-state-db\",\"type\":\"header\",\"data\":{\"text\":\"What is a state database?\",\"level\":2},\"tunes\":{}},{\"id\":\"p-statedb-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"A state database or state store is simply a place where the application keeps current facts.\"},\"tunes\":{}},{\"id\":\"p-statedb-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"In a game, the engine already knows things such as your health, position, inventory, ammunition, current mission, nearby objects and enemy status. An AI system can expose selected parts of that state to the model.\"},\"tunes\":{}},{\"id\":\"p-statedb-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"In a business application, the same idea could be an order database, a customer record, a project status or the current value of a sensor.\"},\"tunes\":{}},{\"id\":\"p-statedb-4\",\"type\":\"paragraph\",\"data\":{\"text\":\"The state is created by the application itself as things happen. If you lose health, the game updates the health value. If you pick up ammunition, the inventory changes. If an order is paid, the business system changes the order status.\"},\"tunes\":{}},{\"id\":\"state-rule\",\"type\":\"callout\",\"data\":{\"variant\":\"info\",\"title\":\"Simple rule\",\"body\":\"The application creates and updates \u003Cstrong>state\u003C\u002Fstrong>. RAG searches \u003Cstrong>knowledge\u003C\u002Fstrong>. The LLM uses both to decide what to say or do.\"},\"tunes\":{}},{\"id\":\"h-together\",\"type\":\"header\",\"data\":{\"text\":\"How the three pieces work together\",\"level\":2},\"tunes\":{}},{\"id\":\"together-flow\",\"type\":\"processFlow\",\"data\":{\"title\":\"LLM + state + RAG\",\"orientation\":\"auto\",\"steps\":[{\"label\":\"1. Current state\",\"description\":\"The application tells the AI what is true now: health 41%, AKM equipped, 23 rounds.\"},{\"label\":\"2. RAG\",\"description\":\"The system retrieves useful knowledge: how the weapon works, which healing item is available, or a relevant rule.\"},{\"label\":\"3. LLM\",\"description\":\"The model receives the question, current state and retrieved knowledge.\"},{\"label\":\"4. Reasoning\",\"description\":\"The LLM combines those inputs and decides what answer or high-level action makes sense.\"},{\"label\":\"5. Application\",\"description\":\"If an action is required, the application or game engine executes it and updates the state again.\"}]},\"tunes\":{}},{\"id\":\"p-arch-intro\",\"type\":\"paragraph\",\"data\":{\"text\":\"So the basic architecture is:\"},\"tunes\":{}},{\"id\":\"simple-architecture\",\"type\":\"callout\",\"data\":{\"variant\":\"note\",\"title\":\"The simplest architecture\",\"body\":\"\u003Cstrong>State = what is true now\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = useful knowledge\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>LLM = understands, reasons and writes\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Application = performs the real action\u003C\u002Fstrong>\"},\"tunes\":{}},{\"id\":\"h-vector\",\"type\":\"header\",\"data\":{\"text\":\"Does RAG always use a vector database?\",\"level\":2},\"tunes\":{}},{\"id\":\"p-vector-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"No.\"},\"tunes\":{}},{\"id\":\"p-vector-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"A vector database is a common way to build semantic search, but it is not the definition of RAG.\"},\"tunes\":{}},{\"id\":\"p-vector-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"The important part is retrieval: the system finds relevant external information and adds it to the LLM's context before the answer is generated.\"},\"tunes\":{}},{\"id\":\"p-vector-4\",\"type\":\"paragraph\",\"data\":{\"text\":\"OpenAI's File Search, for example, can work with files stored in vector stores. Files are chunked into smaller pieces so the system can retrieve the parts that are relevant to a question. That is one implementation of the same basic idea.\"},\"tunes\":{}},{\"id\":\"h-embedding\",\"type\":\"header\",\"data\":{\"text\":\"What is an embedding, in plain English?\",\"level\":2},\"tunes\":{}},{\"id\":\"p-emb-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"You do not need to understand embeddings to understand RAG.\"},\"tunes\":{}},{\"id\":\"p-emb-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"But the simple version is this: an embedding is a numerical representation of meaning. It helps a search system find text that is conceptually similar even when the words are not exactly the same.\"},\"tunes\":{}},{\"id\":\"p-emb-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"For example, a normal keyword search may look for the exact words “car repair.” Semantic search can also understand that “fix my vehicle” is about a similar topic.\"},\"tunes\":{}},{\"id\":\"p-emb-4\",\"type\":\"paragraph\",\"data\":{\"text\":\"That makes embeddings useful for RAG, but RAG can also use keyword search, database queries or a hybrid of several methods.\"},\"tunes\":{}},{\"id\":\"h-memory\",\"type\":\"header\",\"data\":{\"text\":\"RAG is not memory either\",\"level\":2},\"tunes\":{}},{\"id\":\"p-memory-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Memory is another concept that is often mixed together with RAG.\"},\"tunes\":{}},{\"id\":\"p-memory-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Memory is usually information the system keeps about previous interactions or previous events. RAG is the mechanism used to retrieve relevant knowledge when it is needed.\"},\"tunes\":{}},{\"id\":\"parts-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Part\",\"Simple meaning\"],[\"LLM\",\"The part that understands and generates language\"],[\"RAG\",\"The part that looks up relevant knowledge before the answer\"],[\"Knowledge base\",\"The information RAG can search\"],[\"State\",\"What is true right now in the application or world\"],[\"Memory\",\"Information kept from previous interactions or events\"],[\"Tool \u002F action\",\"Something the AI is allowed to call or ask the application to do\"],[\"Context\",\"The information currently placed in front of the LLM for this request\"]]},\"tunes\":{}},{\"id\":\"h-pubg\",\"type\":\"header\",\"data\":{\"text\":\"A real game example: PUBG Ally\",\"level\":2},\"tunes\":{}},{\"id\":\"p-pubg-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"PUBG Ally is a useful example because it makes the difference visible.\"},\"tunes\":{}},{\"id\":\"p-pubg-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"KRAFTON describes live match state as a separate source of truth. The game exposes current facts through observation tools: current weapon, ammunition, health, safe-zone status, nearby items and combat situation.\"},\"tunes\":{}},{\"id\":\"p-pubg-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"Knowledge lookup is a different job. The system can use curated knowledge about weapons, attachments, items and rules. NVIDIA's ACE Game Agent SDK also exposes a separate RAG API for retrieving knowledge from developer-built databases.\"},\"tunes\":{}},{\"id\":\"p-pubg-4\",\"type\":\"paragraph\",\"data\":{\"text\":\"That gives us the clean separation: the game engine says what is happening now, retrieval provides relevant knowledge, and the language model decides what the information means.\"},\"tunes\":{}},{\"id\":\"ref-pubg\",\"type\":\"referralArticle\",\"data\":{\"url\":\"https:\u002F\u002Ffigure.rocks\u002Fblog\u002Fpubg-ally-shows-why-ai-teammates-need-two-brains-fast-reflexes-and-slow-reasoning\",\"title\":\"PUBG Ally Shows Why AI Teammates Need Two Brains: Fast Reflexes and Slow Reasoning\",\"excerpt\":\"A practical game example showing how live state, language reasoning and deterministic game-side control can work together.\",\"ctaLabel\":\"Read the PUBG Ally architecture article\"},\"tunes\":{}},{\"id\":\"h-complete\",\"type\":\"header\",\"data\":{\"text\":\"One complete example\",\"level\":2},\"tunes\":{}},{\"id\":\"p-complete-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Imagine you tell an AI teammate: “I am low on health. Should we attack?”\"},\"tunes\":{}},{\"id\":\"complete-flow\",\"type\":\"processFlow\",\"data\":{\"title\":\"What happens next\",\"orientation\":\"auto\",\"steps\":[{\"label\":\"State\",\"description\":\"The game reports: health 24%, one enemy nearby, two healing items available.\"},{\"label\":\"RAG\",\"description\":\"The knowledge system retrieves the relevant rules for the healing item and perhaps information about the current weapon or tactical mechanic.\"},{\"label\":\"LLM\",\"description\":\"The model combines your request, the current state and the retrieved knowledge.\"},{\"label\":\"Decision\",\"description\":\"It concludes that healing first is safer than attacking immediately.\"},{\"label\":\"Tool \u002F game engine\",\"description\":\"The agent requests a legal game action such as moving to cover or using the healing item.\"},{\"label\":\"New state\",\"description\":\"The game executes the action and reports the updated situation back to the agent.\"}]},\"tunes\":{}},{\"id\":\"p-complete-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG did not control the character. The state database did not reason. The LLM did not directly change the game. Each part had one job.\"},\"tunes\":{}},{\"id\":\"h-why\",\"type\":\"header\",\"data\":{\"text\":\"Why use RAG at all?\",\"level\":2},\"tunes\":{}},{\"id\":\"p-why-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Because putting every document, rule and database record into every prompt would be slow, expensive and often confusing.\"},\"tunes\":{}},{\"id\":\"p-why-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG lets the system select only the information that is useful for the current question.\"},\"tunes\":{}},{\"id\":\"p-why-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"It also lets you update the knowledge base without retraining the entire language model. Change the document or database, rebuild or refresh the index when necessary, and the next retrieval can use the newer information.\"},\"tunes\":{}},{\"id\":\"h-not-guarantee\",\"type\":\"header\",\"data\":{\"text\":\"What RAG does not guarantee\",\"level\":2},\"tunes\":{}},{\"id\":\"p-not-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG can improve grounding, but it does not make an answer automatically correct.\"},\"tunes\":{}},{\"id\":\"p-not-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"The retrieval step can find the wrong document. The correct document can be outdated. The LLM can misunderstand good evidence. Or the current state can have changed.\"},\"tunes\":{}},{\"id\":\"p-not-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"A reliable system therefore has to validate retrieval, state freshness and the model's final reasoning separately.\"},\"tunes\":{}},{\"id\":\"h-mental\",\"type\":\"header\",\"data\":{\"text\":\"The easiest mental model to remember\",\"level\":2},\"tunes\":{}},{\"id\":\"mental-table\",\"type\":\"comparison\",\"data\":{\"title\":\"Think of an AI system like a person at a desk\",\"layout\":\"table\",\"columns\":[{\"id\":\"analogy\",\"label\":\"Analogy\"},{\"id\":\"system\",\"label\":\"AI system\"}],\"rows\":[{\"id\":\"brain\",\"label\":\"Person thinking\",\"values\":[\"\",\"\"]},{\"id\":\"library\",\"label\":\"Finding a reference book\",\"values\":[\"\",\"\"]},{\"id\":\"books\",\"label\":\"Books on the shelf\",\"values\":[\"\",\"\"]},{\"id\":\"dashboard\",\"label\":\"Current dashboard or instrument panel\",\"values\":[\"\",\"\"]},{\"id\":\"notes\",\"label\":\"Notes from earlier meetings\",\"values\":[\"\",\"\"]},{\"id\":\"hands\",\"label\":\"Doing something in the real world\",\"values\":[\"\",\"\"]}]},\"tunes\":{}},{\"id\":\"remember\",\"type\":\"callout\",\"data\":{\"variant\":\"success\",\"title\":\"If you remember only this\",\"body\":\"\u003Cstrong>LLM = brain.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = librarian.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Knowledge base = library.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>State = what the dashboard says right now.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Tools = the hands that can actually do something.\u003C\u002Fstrong>\"},\"tunes\":{}},{\"id\":\"h-conclusion\",\"type\":\"header\",\"data\":{\"text\":\"Conclusion\",\"level\":2},\"tunes\":{}},{\"id\":\"p-conc-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"RAG is much less mysterious once the parts are separated.\"},\"tunes\":{}},{\"id\":\"p-conc-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"The LLM understands and generates language. The application maintains current state. The knowledge base stores information. RAG finds the useful part of that information and puts it into the LLM's context. Tools or the application perform real actions.\"},\"tunes\":{}},{\"id\":\"p-conc-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"That is the basic architecture behind many modern AI assistants and agents.\"},\"tunes\":{}},{\"id\":\"h-faq\",\"type\":\"header\",\"data\":{\"text\":\"FAQ\",\"level\":2},\"tunes\":{}},{\"id\":\"faq\",\"type\":\"faq\",\"data\":{\"title\":\"RAG in plain English\",\"items\":[{\"id\":\"faq1\",\"question\":\"What is RAG in simple terms?\",\"answer\":\"RAG is a step where an AI searches a knowledge source for relevant information before the language model writes its answer.\"},{\"id\":\"faq2\",\"question\":\"Does RAG need the Internet?\",\"answer\":\"No. The knowledge base can be completely local on your computer or server.\"},{\"id\":\"faq3\",\"question\":\"Is RAG the same as a database?\",\"answer\":\"No. The database or files contain the information. RAG is the retrieval process that finds the useful part and gives it to the LLM.\"},{\"id\":\"faq4\",\"question\":\"Is RAG the same as memory?\",\"answer\":\"No. Memory usually stores previous interactions or events. RAG retrieves relevant knowledge when it is needed.\"},{\"id\":\"faq5\",\"question\":\"Is current application state part of RAG?\",\"answer\":\"Not necessarily. Current state is usually obtained directly from the application or a state store. RAG is better understood as retrieval from a knowledge source.\"},{\"id\":\"faq6\",\"question\":\"Does RAG make AI answers correct?\",\"answer\":\"No. It can provide better evidence, but retrieval can still be wrong or outdated and the LLM can still reason incorrectly.\"}]},\"tunes\":{}},{\"id\":\"h-glossary\",\"type\":\"header\",\"data\":{\"text\":\"Glossary\",\"level\":2},\"tunes\":{}},{\"id\":\"glossary\",\"type\":\"glossary\",\"data\":{\"title\":\"The basic terms\",\"entries\":[{\"term\":\"LLM\",\"definition\":\"A language model that understands and generates text and can reason over information placed in its context.\",\"anchor\":\"llm\"},{\"term\":\"RAG\",\"definition\":\"Retrieval-Augmented Generation: retrieving relevant external information and adding it to the model's context before generating an answer.\",\"anchor\":\"rag\"},{\"term\":\"Knowledge base\",\"definition\":\"The files, documents, records or other information that retrieval can search.\",\"anchor\":\"knowledge-base\"},{\"term\":\"State\",\"definition\":\"The current facts of an application, system or world at a particular moment.\",\"anchor\":\"state\"},{\"term\":\"Context\",\"definition\":\"The information currently supplied to the language model for one request or reasoning step.\",\"anchor\":\"context\"},{\"term\":\"Embedding\",\"definition\":\"A numerical representation of meaning that can help semantic search find conceptually similar information.\",\"anchor\":\"embedding\"}]},\"tunes\":{}},{\"id\":\"h-sources\",\"type\":\"header\",\"data\":{\"text\":\"Primary sources\",\"level\":2},\"tunes\":{}},{\"id\":\"src-openai-vector\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fapi-reference\u002Fvector-stores-files\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"OpenAI — Vector Store Files\",\"description\":\"Official documentation showing how files can be attached to vector stores, chunked and made available to file-search retrieval.\"}},\"tunes\":{}},{\"id\":\"src-openai-quickstart\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fquickstart\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"OpenAI — Developer Quickstart\",\"description\":\"Official OpenAI documentation describing tools such as file search for giving models access to external information.\"}},\"tunes\":{}},{\"id\":\"src-nvidia-ace\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fdeveloper.nvidia.com\u002Face-for-games\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"NVIDIA Developer — ACE for Games\",\"description\":\"Official NVIDIA documentation describing separate Agent, Chat and RAG APIs for connecting game characters to game state, contextual knowledge and model-driven actions.\"}},\"tunes\":{}},{\"id\":\"src-nvidia-pubg\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fdeveloper.nvidia.com\u002Fblog\u002Fhow-krafton-built-pubg-ally-a-co-playable-character-powered-by-nvidia-ace\u002F\",\"meta\":{\"image\":{\"url\":\"\"},\"title\":\"NVIDIA Developer — How KRAFTON Built PUBG Ally\",\"description\":\"Official technical explanation separating live match state from knowledge lookup and language-model reasoning.\"}},\"tunes\":{}}],\"version\":\"2.31.6\"}",{"time":926,"blocks":927,"version":1445},1790377494031,[928,932,937,941,945,949,953,957,961,966,970,974,978,982,986,990,1007,1011,1015,1019,1023,1042,1046,1050,1054,1058,1062,1066,1088,1093,1097,1101,1105,1109,1113,1117,1121,1138,1142,1147,1151,1155,1159,1163,1167,1171,1175,1179,1183,1187,1191,1195,1199,1225,1229,1233,1237,1241,1245,1251,1255,1259,1279,1283,1287,1291,1295,1299,1303,1307,1311,1315,1319,1347,1352,1355,1359,1363,1367,1370,1393,1397,1414,1418,1425,1432,1438],{"id":215,"data":929,"type":218,"tunes":931},{"text":930},"RAG sounds complicated because the name is complicated. The idea is not. RAG simply means: before the AI answers, it first looks up relevant information from a knowledge source and gives that information to the language model.",{},{"id":221,"data":933,"type":226,"tunes":936},{"body":934,"title":935,"variant":225},"\u003Cstrong>RAG is the step where an AI searches a knowledge base for useful information before the LLM writes the answer.\u003C\u002Fstrong>","RAG in one sentence",{},{"id":229,"data":938,"type":218,"tunes":940},{"text":939},"Think of an LLM as a smart person sitting at a desk. RAG is the librarian who brings the right page from the right book. The LLM then reads that page and answers you.",{},{"id":234,"data":942,"type":239,"tunes":944},{"title":943,"maxLevel":237,"minLevel":238},"Contents",{},{"id":242,"data":946,"type":42,"tunes":948},{"text":947,"level":238},"First: what does the LLM do?",{},{"id":247,"data":950,"type":218,"tunes":952},{"text":951},"The LLM is the part that understands language and produces language. It can read your question, understand instructions, compare information, explain something and write an answer.",{},{"id":252,"data":954,"type":218,"tunes":956},{"text":955},"But the LLM does not automatically know what is currently inside your company database, your game session, your private documents or a file you created five minutes ago.",{},{"id":257,"data":958,"type":218,"tunes":960},{"text":959},"It only knows what is already inside the model plus whatever information the application gives it in the current request.",{},{"id":262,"data":962,"type":226,"tunes":965},{"body":963,"title":964,"variant":266},"The LLM \u003Cstrong>thinks and writes\u003C\u002Fstrong>. It does not automatically own all of your current data.","Simple rule",{},{"id":269,"data":967,"type":42,"tunes":969},{"text":968,"level":238},"Then: what is the knowledge base?",{},{"id":274,"data":971,"type":218,"tunes":973},{"text":972},"A knowledge base is simply information the application can search.",{},{"id":279,"data":975,"type":218,"tunes":977},{"text":976},"It could contain PDFs, manuals, product documentation, support articles, contracts, game rules, weapon data, internal company documents, database records or other text.",{},{"id":284,"data":979,"type":218,"tunes":981},{"text":980},"The knowledge base can be local on your own machine. It can be on a server. It can be in a vector database. It can also be built from normal files. RAG does not mean Internet.",{},{"id":289,"data":983,"type":226,"tunes":985},{"body":984,"title":292,"variant":293},"\u003Cstrong>RAG does not require the Internet.\u003C\u002Fstrong> The information can be completely local.",{},{"id":296,"data":987,"type":42,"tunes":989},{"text":988,"level":238},"So what does RAG actually do?",{},{"id":301,"data":991,"type":318,"tunes":1006},{"steps":992,"title":1005,"orientation":317},[993,996,999,1002],{"label":994,"description":995},"1. You ask a question","For example: Which ammunition does this weapon use?",{"label":997,"description":998},"2. RAG searches the knowledge base","The system looks for the small pieces of information most relevant to your question.",{"label":1000,"description":1001},"3. RAG gives those pieces to the LLM","The LLM receives the question plus the retrieved information.",{"label":1003,"description":1004},"4. The LLM writes the answer","It uses the retrieved information as context for the response.","The whole RAG process",{},{"id":321,"data":1008,"type":218,"tunes":1010},{"text":1009},"That is RAG.",{},{"id":326,"data":1012,"type":218,"tunes":1014},{"text":1013},"The full name is Retrieval-Augmented Generation. Retrieval means finding the relevant information. Augmented means adding that information to the model's context. Generation means the LLM writes the final answer.",{},{"id":331,"data":1016,"type":42,"tunes":1018},{"text":1017,"level":238},"A very simple example",{},{"id":336,"data":1020,"type":218,"tunes":1022},{"text":1021},"Imagine you have a local knowledge base about a game.",{},{"id":341,"data":1024,"type":359,"tunes":1041},{"content":1025,"stretched":43,"withHeadings":14},[1026,1029,1032,1035,1038],[1027,1028],"Knowledge base contains","Example",[1030,1031],"Weapons","AKM uses 7.62 mm ammunition",[1033,1034],"Healing items","Med Kit restores health",[1036,1037],"Attachments","This attachment works with these weapons",[1039,1040],"Map rules","This zone behaves in this way",{},{"id":362,"data":1043,"type":218,"tunes":1045},{"text":1044},"You ask: “Which ammunition does the AKM use?”",{},{"id":367,"data":1047,"type":218,"tunes":1049},{"text":1048},"RAG searches the knowledge base and finds the entry about the AKM. It gives that small piece of information to the LLM. The LLM then answers: “The AKM uses 7.62 mm ammunition.”",{},{"id":372,"data":1051,"type":218,"tunes":1053},{"text":1052},"The LLM did not need the entire database. RAG only brought the useful part.",{},{"id":377,"data":1055,"type":42,"tunes":1057},{"text":1056,"level":238},"Now the important part: RAG is not the current state",{},{"id":382,"data":1059,"type":218,"tunes":1061},{"text":1060},"This is where many explanations become confusing.",{},{"id":387,"data":1063,"type":218,"tunes":1065},{"text":1064},"RAG usually gives the AI knowledge. A state system gives the AI facts about what is true right now.",{},{"id":392,"data":1067,"type":420,"tunes":1087},{"rows":1068,"title":1081,"layout":359,"columns":1082},[1069,1072,1075,1078],{"id":396,"label":1070,"values":1071},"Weapon",[399,399],{"id":401,"label":1073,"values":1074},"Ammunition",[399,399],{"id":405,"label":1076,"values":1077},"Health",[399,399],{"id":409,"label":1079,"values":1080},"Enemy",[399,399],"Knowledge vs current state",[1083,1085],{"id":415,"label":1084},"RAG \u002F knowledge",{"id":418,"label":1086},"Current state",{},{"id":423,"data":1089,"type":226,"tunes":1092},{"body":1090,"title":1091,"variant":427},"RAG answers: \u003Cstrong>What is generally true?\u003C\u002Fstrong>\u003Cbr>State answers: \u003Cstrong>What is true right now?\u003C\u002Fstrong>","Do not mix these two",{},{"id":430,"data":1094,"type":42,"tunes":1096},{"text":1095,"level":238},"What is a state database?",{},{"id":435,"data":1098,"type":218,"tunes":1100},{"text":1099},"A state database or state store is simply a place where the application keeps current facts.",{},{"id":440,"data":1102,"type":218,"tunes":1104},{"text":1103},"In a game, the engine already knows things such as your health, position, inventory, ammunition, current mission, nearby objects and enemy status. An AI system can expose selected parts of that state to the model.",{},{"id":445,"data":1106,"type":218,"tunes":1108},{"text":1107},"In a business application, the same idea could be an order database, a customer record, a project status or the current value of a sensor.",{},{"id":450,"data":1110,"type":218,"tunes":1112},{"text":1111},"The state is created by the application itself as things happen. If you lose health, the game updates the health value. If you pick up ammunition, the inventory changes. If an order is paid, the business system changes the order status.",{},{"id":455,"data":1114,"type":226,"tunes":1116},{"body":1115,"title":964,"variant":225},"The application creates and updates \u003Cstrong>state\u003C\u002Fstrong>. RAG searches \u003Cstrong>knowledge\u003C\u002Fstrong>. The LLM uses both to decide what to say or do.",{},{"id":460,"data":1118,"type":42,"tunes":1120},{"text":1119,"level":238},"How the three pieces work together",{},{"id":465,"data":1122,"type":318,"tunes":1137},{"steps":1123,"title":1136,"orientation":317},[1124,1127,1129,1131,1134],{"label":1125,"description":1126},"1. Current state","The application tells the AI what is true now: health 41%, AKM equipped, 23 rounds.",{"label":472,"description":1128},"The system retrieves useful knowledge: how the weapon works, which healing item is available, or a relevant rule.",{"label":475,"description":1130},"The model receives the question, current state and retrieved knowledge.",{"label":1132,"description":1133},"4. Reasoning","The LLM combines those inputs and decides what answer or high-level action makes sense.",{"label":481,"description":1135},"If an action is required, the application or game engine executes it and updates the state again.","LLM + state + RAG",{},{"id":486,"data":1139,"type":218,"tunes":1141},{"text":1140},"So the basic architecture is:",{},{"id":491,"data":1143,"type":226,"tunes":1146},{"body":1144,"title":1145,"variant":266},"\u003Cstrong>State = what is true now\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = useful knowledge\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>LLM = understands, reasons and writes\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Application = performs the real action\u003C\u002Fstrong>","The simplest architecture",{},{"id":497,"data":1148,"type":42,"tunes":1150},{"text":1149,"level":238},"Does RAG always use a vector database?",{},{"id":502,"data":1152,"type":218,"tunes":1154},{"text":1153},"No.",{},{"id":507,"data":1156,"type":218,"tunes":1158},{"text":1157},"A vector database is a common way to build semantic search, but it is not the definition of RAG.",{},{"id":512,"data":1160,"type":218,"tunes":1162},{"text":1161},"The important part is retrieval: the system finds relevant external information and adds it to the LLM's context before the answer is generated.",{},{"id":517,"data":1164,"type":218,"tunes":1166},{"text":1165},"OpenAI's File Search, for example, can work with files stored in vector stores. Files are chunked into smaller pieces so the system can retrieve the parts that are relevant to a question. That is one implementation of the same basic idea.",{},{"id":522,"data":1168,"type":42,"tunes":1170},{"text":1169,"level":238},"What is an embedding, in plain English?",{},{"id":527,"data":1172,"type":218,"tunes":1174},{"text":1173},"You do not need to understand embeddings to understand RAG.",{},{"id":532,"data":1176,"type":218,"tunes":1178},{"text":1177},"But the simple version is this: an embedding is a numerical representation of meaning. It helps a search system find text that is conceptually similar even when the words are not exactly the same.",{},{"id":537,"data":1180,"type":218,"tunes":1182},{"text":1181},"For example, a normal keyword search may look for the exact words “car repair.” Semantic search can also understand that “fix my vehicle” is about a similar topic.",{},{"id":542,"data":1184,"type":218,"tunes":1186},{"text":1185},"That makes embeddings useful for RAG, but RAG can also use keyword search, database queries or a hybrid of several methods.",{},{"id":547,"data":1188,"type":42,"tunes":1190},{"text":1189,"level":238},"RAG is not memory either",{},{"id":552,"data":1192,"type":218,"tunes":1194},{"text":1193},"Memory is another concept that is often mixed together with RAG.",{},{"id":557,"data":1196,"type":218,"tunes":1198},{"text":1197},"Memory is usually information the system keeps about previous interactions or previous events. RAG is the mechanism used to retrieve relevant knowledge when it is needed.",{},{"id":562,"data":1200,"type":359,"tunes":1224},{"content":1201,"stretched":43,"withHeadings":14},[1202,1205,1207,1209,1212,1215,1218,1221],[1203,1204],"Part","Simple meaning",[569,1206],"The part that understands and generates language",[572,1208],"The part that looks up relevant knowledge before the answer",[1210,1211],"Knowledge base","The information RAG can search",[1213,1214],"State","What is true right now in the application or world",[1216,1217],"Memory","Information kept from previous interactions or events",[1219,1220],"Tool \u002F action","Something the AI is allowed to call or ask the application to do",[1222,1223],"Context","The information currently placed in front of the LLM for this request",{},{"id":591,"data":1226,"type":42,"tunes":1228},{"text":1227,"level":238},"A real game example: PUBG Ally",{},{"id":596,"data":1230,"type":218,"tunes":1232},{"text":1231},"PUBG Ally is a useful example because it makes the difference visible.",{},{"id":601,"data":1234,"type":218,"tunes":1236},{"text":1235},"KRAFTON describes live match state as a separate source of truth. The game exposes current facts through observation tools: current weapon, ammunition, health, safe-zone status, nearby items and combat situation.",{},{"id":606,"data":1238,"type":218,"tunes":1240},{"text":1239},"Knowledge lookup is a different job. The system can use curated knowledge about weapons, attachments, items and rules. NVIDIA's ACE Game Agent SDK also exposes a separate RAG API for retrieving knowledge from developer-built databases.",{},{"id":611,"data":1242,"type":218,"tunes":1244},{"text":1243},"That gives us the clean separation: the game engine says what is happening now, retrieval provides relevant knowledge, and the language model decides what the information means.",{},{"id":616,"data":1246,"type":622,"tunes":1250},{"url":618,"title":1247,"excerpt":1248,"ctaLabel":1249},"PUBG Ally Shows Why AI Teammates Need Two Brains: Fast Reflexes and Slow Reasoning","A practical game example showing how live state, language reasoning and deterministic game-side control can work together.","Read the PUBG Ally architecture article",{},{"id":625,"data":1252,"type":42,"tunes":1254},{"text":1253,"level":238},"One complete example",{},{"id":630,"data":1256,"type":218,"tunes":1258},{"text":1257},"Imagine you tell an AI teammate: “I am low on health. Should we attack?”",{},{"id":635,"data":1260,"type":318,"tunes":1278},{"steps":1261,"title":1277,"orientation":317},[1262,1264,1266,1268,1271,1274],{"label":1213,"description":1263},"The game reports: health 24%, one enemy nearby, two healing items available.",{"label":572,"description":1265},"The knowledge system retrieves the relevant rules for the healing item and perhaps information about the current weapon or tactical mechanic.",{"label":569,"description":1267},"The model combines your request, the current state and the retrieved knowledge.",{"label":1269,"description":1270},"Decision","It concludes that healing first is safer than attacking immediately.",{"label":1272,"description":1273},"Tool \u002F game engine","The agent requests a legal game action such as moving to cover or using the healing item.",{"label":1275,"description":1276},"New state","The game executes the action and reports the updated situation back to the agent.","What happens next",{},{"id":656,"data":1280,"type":218,"tunes":1282},{"text":1281},"RAG did not control the character. The state database did not reason. The LLM did not directly change the game. Each part had one job.",{},{"id":661,"data":1284,"type":42,"tunes":1286},{"text":1285,"level":238},"Why use RAG at all?",{},{"id":666,"data":1288,"type":218,"tunes":1290},{"text":1289},"Because putting every document, rule and database record into every prompt would be slow, expensive and often confusing.",{},{"id":671,"data":1292,"type":218,"tunes":1294},{"text":1293},"RAG lets the system select only the information that is useful for the current question.",{},{"id":676,"data":1296,"type":218,"tunes":1298},{"text":1297},"It also lets you update the knowledge base without retraining the entire language model. Change the document or database, rebuild or refresh the index when necessary, and the next retrieval can use the newer information.",{},{"id":681,"data":1300,"type":42,"tunes":1302},{"text":1301,"level":238},"What RAG does not guarantee",{},{"id":686,"data":1304,"type":218,"tunes":1306},{"text":1305},"RAG can improve grounding, but it does not make an answer automatically correct.",{},{"id":691,"data":1308,"type":218,"tunes":1310},{"text":1309},"The retrieval step can find the wrong document. The correct document can be outdated. The LLM can misunderstand good evidence. Or the current state can have changed.",{},{"id":696,"data":1312,"type":218,"tunes":1314},{"text":1313},"A reliable system therefore has to validate retrieval, state freshness and the model's final reasoning separately.",{},{"id":701,"data":1316,"type":42,"tunes":1318},{"text":1317,"level":238},"The easiest mental model to remember",{},{"id":706,"data":1320,"type":420,"tunes":1346},{"rows":1321,"title":1340,"layout":359,"columns":1341},[1322,1325,1328,1331,1334,1337],{"id":710,"label":1323,"values":1324},"Person thinking",[399,399],{"id":714,"label":1326,"values":1327},"Finding a reference book",[399,399],{"id":718,"label":1329,"values":1330},"Books on the shelf",[399,399],{"id":722,"label":1332,"values":1333},"Current dashboard or instrument panel",[399,399],{"id":726,"label":1335,"values":1336},"Notes from earlier meetings",[399,399],{"id":730,"label":1338,"values":1339},"Doing something in the real world",[399,399],"Think of an AI system like a person at a desk",[1342,1344],{"id":229,"label":1343},"Analogy",{"id":738,"label":1345},"AI system",{},{"id":742,"data":1348,"type":226,"tunes":1351},{"body":1349,"title":1350,"variant":293},"\u003Cstrong>LLM = brain.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>RAG = librarian.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Knowledge base = library.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>State = what the dashboard says right now.\u003C\u002Fstrong>\u003Cbr>\u003Cstrong>Tools = the hands that can actually do something.\u003C\u002Fstrong>","If you remember only this",{},{"id":748,"data":1353,"type":42,"tunes":1354},{"text":750,"level":238},{},{"id":753,"data":1356,"type":218,"tunes":1358},{"text":1357},"RAG is much less mysterious once the parts are separated.",{},{"id":758,"data":1360,"type":218,"tunes":1362},{"text":1361},"The LLM understands and generates language. The application maintains current state. The knowledge base stores information. RAG finds the useful part of that information and puts it into the LLM's context. Tools or the application perform real actions.",{},{"id":763,"data":1364,"type":218,"tunes":1366},{"text":1365},"That is the basic architecture behind many modern AI assistants and agents.",{},{"id":768,"data":1368,"type":42,"tunes":1369},{"text":770,"level":238},{},{"id":773,"data":1371,"type":773,"tunes":1392},{"items":1372,"title":1391},[1373,1376,1379,1382,1385,1388],{"id":777,"answer":1374,"question":1375},"RAG is a step where an AI searches a knowledge source for relevant information before the language model writes its answer.","What is RAG in simple terms?",{"id":781,"answer":1377,"question":1378},"No. The knowledge base can be completely local on your computer or server.","Does RAG need the Internet?",{"id":785,"answer":1380,"question":1381},"No. The database or files contain the information. RAG is the retrieval process that finds the useful part and gives it to the LLM.","Is RAG the same as a database?",{"id":789,"answer":1383,"question":1384},"No. Memory usually stores previous interactions or events. RAG retrieves relevant knowledge when it is needed.","Is RAG the same as memory?",{"id":793,"answer":1386,"question":1387},"Not necessarily. Current state is usually obtained directly from the application or a state store. RAG is better understood as retrieval from a knowledge source.","Is current application state part of RAG?",{"id":797,"answer":1389,"question":1390},"No. It can provide better evidence, but retrieval can still be wrong or outdated and the LLM can still reason incorrectly.","Does RAG make AI answers correct?","RAG in plain English",{},{"id":803,"data":1394,"type":42,"tunes":1396},{"text":1395,"level":238},"Glossary",{},{"id":808,"data":1398,"type":808,"tunes":1413},{"title":1399,"entries":1400},"The basic terms",[1401,1403,1405,1407,1409,1411],{"term":569,"anchor":813,"definition":1402},"A language model that understands and generates text and can reason over information placed in its context.",{"term":572,"anchor":816,"definition":1404},"Retrieval-Augmented Generation: retrieving relevant external information and adding it to the model's context before generating an answer.",{"term":1210,"anchor":819,"definition":1406},"The files, documents, records or other information that retrieval can search.",{"term":1213,"anchor":418,"definition":1408},"The current facts of an application, system or world at a particular moment.",{"term":1222,"anchor":824,"definition":1410},"The information currently supplied to the language model for one request or reasoning step.",{"term":827,"anchor":828,"definition":1412},"A numerical representation of meaning that can help semantic search find conceptually similar information.",{},{"id":832,"data":1415,"type":42,"tunes":1417},{"text":1416,"level":238},"Primary sources",{},{"id":837,"data":1419,"type":844,"tunes":1424},{"link":839,"meta":1420},{"image":1421,"title":1422,"description":1423},{"url":399},"OpenAI — Vector Store Files","Official documentation showing how files can be attached to vector stores, chunked and made available to file-search retrieval.",{},{"id":847,"data":1426,"type":844,"tunes":1431},{"link":849,"meta":1427},{"image":1428,"title":1429,"description":1430},{"url":399},"OpenAI — Developer Quickstart","Official OpenAI documentation describing tools such as file search for giving models access to external information.",{},{"id":856,"data":1433,"type":844,"tunes":1437},{"link":858,"meta":1434},{"image":1435,"title":861,"description":1436},{"url":399},"Official NVIDIA documentation describing separate Agent, Chat and RAG APIs for connecting game characters to game state, contextual knowledge and model-driven actions.",{},{"id":865,"data":1439,"type":844,"tunes":1444},{"link":867,"meta":1440},{"image":1441,"title":1442,"description":1443},{"url":399},"NVIDIA Developer — How KRAFTON Built PUBG Ally","Official technical explanation separating live match state from knowledge lookup and language-model reasoning.",{},"2.31.6","RAG sounds complicated, but the idea is simple: before an AI answers, it first looks up useful information from a knowledge source and gives that information to the language model. This guide explains RAG, LLMs, state, memory and tools using one simple mental model.",{"lang":7,"title":208,"content":210,"contentJson":1448,"excerpt":874},{"time":212,"blocks":1449,"version":873},[1450,1453,1456,1459,1462,1465,1468,1471,1474,1477,1480,1483,1486,1489,1492,1495,1503,1506,1509,1512,1515,1524,1527,1530,1533,1536,1539,1542,1557,1560,1563,1566,1569,1572,1575,1578,1581,1590,1593,1596,1599,1602,1605,1608,1611,1614,1617,1620,1623,1626,1629,1632,1635,1647,1650,1653,1656,1659,1662,1665,1668,1671,1681,1684,1687,1690,1693,1696,1699,1702,1705,1708,1711,1730,1733,1736,1739,1742,1745,1748,1758,1761,1771,1774,1779,1784,1789],{"id":215,"data":1451,"type":218,"tunes":1452},{"text":217},{},{"id":221,"data":1454,"type":226,"tunes":1455},{"body":223,"title":224,"variant":225},{},{"id":229,"data":1457,"type":218,"tunes":1458},{"text":231},{},{"id":234,"data":1460,"type":239,"tunes":1461},{"title":236,"maxLevel":237,"minLevel":238},{},{"id":242,"data":1463,"type":42,"tunes":1464},{"text":244,"level":238},{},{"id":247,"data":1466,"type":218,"tunes":1467},{"text":249},{},{"id":252,"data":1469,"type":218,"tunes":1470},{"text":254},{},{"id":257,"data":1472,"type":218,"tunes":1473},{"text":259},{},{"id":262,"data":1475,"type":226,"tunes":1476},{"body":264,"title":265,"variant":266},{},{"id":269,"data":1478,"type":42,"tunes":1479},{"text":271,"level":238},{},{"id":274,"data":1481,"type":218,"tunes":1482},{"text":276},{},{"id":279,"data":1484,"type":218,"tunes":1485},{"text":281},{},{"id":284,"data":1487,"type":218,"tunes":1488},{"text":286},{},{"id":289,"data":1490,"type":226,"tunes":1491},{"body":291,"title":292,"variant":293},{},{"id":296,"data":1493,"type":42,"tunes":1494},{"text":298,"level":238},{},{"id":301,"data":1496,"type":318,"tunes":1502},{"steps":1497,"title":316,"orientation":317},[1498,1499,1500,1501],{"label":305,"description":306},{"label":308,"description":309},{"label":311,"description":312},{"label":314,"description":315},{},{"id":321,"data":1504,"type":218,"tunes":1505},{"text":323},{},{"id":326,"data":1507,"type":218,"tunes":1508},{"text":328},{},{"id":331,"data":1510,"type":42,"tunes":1511},{"text":333,"level":238},{},{"id":336,"data":1513,"type":218,"tunes":1514},{"text":338},{},{"id":341,"data":1516,"type":359,"tunes":1523},{"content":1517,"stretched":43,"withHeadings":14},[1518,1519,1520,1521,1522],[345,346],[348,349],[351,352],[354,355],[357,358],{},{"id":362,"data":1525,"type":218,"tunes":1526},{"text":364},{},{"id":367,"data":1528,"type":218,"tunes":1529},{"text":369},{},{"id":372,"data":1531,"type":218,"tunes":1532},{"text":374},{},{"id":377,"data":1534,"type":42,"tunes":1535},{"text":379,"level":238},{},{"id":382,"data":1537,"type":218,"tunes":1538},{"text":384},{},{"id":387,"data":1540,"type":218,"tunes":1541},{"text":389},{},{"id":392,"data":1543,"type":420,"tunes":1556},{"rows":1544,"title":412,"layout":359,"columns":1553},[1545,1547,1549,1551],{"id":396,"label":397,"values":1546},[399,399],{"id":401,"label":402,"values":1548},[399,399],{"id":405,"label":406,"values":1550},[399,399],{"id":409,"label":410,"values":1552},[399,399],[1554,1555],{"id":415,"label":416},{"id":418,"label":419},{},{"id":423,"data":1558,"type":226,"tunes":1559},{"body":425,"title":426,"variant":427},{},{"id":430,"data":1561,"type":42,"tunes":1562},{"text":432,"level":238},{},{"id":435,"data":1564,"type":218,"tunes":1565},{"text":437},{},{"id":440,"data":1567,"type":218,"tunes":1568},{"text":442},{},{"id":445,"data":1570,"type":218,"tunes":1571},{"text":447},{},{"id":450,"data":1573,"type":218,"tunes":1574},{"text":452},{},{"id":455,"data":1576,"type":226,"tunes":1577},{"body":457,"title":265,"variant":225},{},{"id":460,"data":1579,"type":42,"tunes":1580},{"text":462,"level":238},{},{"id":465,"data":1582,"type":318,"tunes":1589},{"steps":1583,"title":483,"orientation":317},[1584,1585,1586,1587,1588],{"label":469,"description":470},{"label":472,"description":473},{"label":475,"description":476},{"label":478,"description":479},{"label":481,"description":482},{},{"id":486,"data":1591,"type":218,"tunes":1592},{"text":488},{},{"id":491,"data":1594,"type":226,"tunes":1595},{"body":493,"title":494,"variant":266},{},{"id":497,"data":1597,"type":42,"tunes":1598},{"text":499,"level":238},{},{"id":502,"data":1600,"type":218,"tunes":1601},{"text":504},{},{"id":507,"data":1603,"type":218,"tunes":1604},{"text":509},{},{"id":512,"data":1606,"type":218,"tunes":1607},{"text":514},{},{"id":517,"data":1609,"type":218,"tunes":1610},{"text":519},{},{"id":522,"data":1612,"type":42,"tunes":1613},{"text":524,"level":238},{},{"id":527,"data":1615,"type":218,"tunes":1616},{"text":529},{},{"id":532,"data":1618,"type":218,"tunes":1619},{"text":534},{},{"id":537,"data":1621,"type":218,"tunes":1622},{"text":539},{},{"id":542,"data":1624,"type":218,"tunes":1625},{"text":544},{},{"id":547,"data":1627,"type":42,"tunes":1628},{"text":549,"level":238},{},{"id":552,"data":1630,"type":218,"tunes":1631},{"text":554},{},{"id":557,"data":1633,"type":218,"tunes":1634},{"text":559},{},{"id":562,"data":1636,"type":359,"tunes":1646},{"content":1637,"stretched":43,"withHeadings":14},[1638,1639,1640,1641,1642,1643,1644,1645],[566,567],[569,570],[572,573],[575,576],[578,579],[581,582],[584,585],[587,588],{},{"id":591,"data":1648,"type":42,"tunes":1649},{"text":593,"level":238},{},{"id":596,"data":1651,"type":218,"tunes":1652},{"text":598},{},{"id":601,"data":1654,"type":218,"tunes":1655},{"text":603},{},{"id":606,"data":1657,"type":218,"tunes":1658},{"text":608},{},{"id":611,"data":1660,"type":218,"tunes":1661},{"text":613},{},{"id":616,"data":1663,"type":622,"tunes":1664},{"url":618,"title":619,"excerpt":620,"ctaLabel":621},{},{"id":625,"data":1666,"type":42,"tunes":1667},{"text":627,"level":238},{},{"id":630,"data":1669,"type":218,"tunes":1670},{"text":632},{},{"id":635,"data":1672,"type":318,"tunes":1680},{"steps":1673,"title":653,"orientation":317},[1674,1675,1676,1677,1678,1679],{"label":578,"description":639},{"label":572,"description":641},{"label":569,"description":643},{"label":645,"description":646},{"label":648,"description":649},{"label":651,"description":652},{},{"id":656,"data":1682,"type":218,"tunes":1683},{"text":658},{},{"id":661,"data":1685,"type":42,"tunes":1686},{"text":663,"level":238},{},{"id":666,"data":1688,"type":218,"tunes":1689},{"text":668},{},{"id":671,"data":1691,"type":218,"tunes":1692},{"text":673},{},{"id":676,"data":1694,"type":218,"tunes":1695},{"text":678},{},{"id":681,"data":1697,"type":42,"tunes":1698},{"text":683,"level":238},{},{"id":686,"data":1700,"type":218,"tunes":1701},{"text":688},{},{"id":691,"data":1703,"type":218,"tunes":1704},{"text":693},{},{"id":696,"data":1706,"type":218,"tunes":1707},{"text":698},{},{"id":701,"data":1709,"type":42,"tunes":1710},{"text":703,"level":238},{},{"id":706,"data":1712,"type":420,"tunes":1729},{"rows":1713,"title":733,"layout":359,"columns":1726},[1714,1716,1718,1720,1722,1724],{"id":710,"label":711,"values":1715},[399,399],{"id":714,"label":715,"values":1717},[399,399],{"id":718,"label":719,"values":1719},[399,399],{"id":722,"label":723,"values":1721},[399,399],{"id":726,"label":727,"values":1723},[399,399],{"id":730,"label":731,"values":1725},[399,399],[1727,1728],{"id":229,"label":736},{"id":738,"label":739},{},{"id":742,"data":1731,"type":226,"tunes":1732},{"body":744,"title":745,"variant":293},{},{"id":748,"data":1734,"type":42,"tunes":1735},{"text":750,"level":238},{},{"id":753,"data":1737,"type":218,"tunes":1738},{"text":755},{},{"id":758,"data":1740,"type":218,"tunes":1741},{"text":760},{},{"id":763,"data":1743,"type":218,"tunes":1744},{"text":765},{},{"id":768,"data":1746,"type":42,"tunes":1747},{"text":770,"level":238},{},{"id":773,"data":1749,"type":773,"tunes":1757},{"items":1750,"title":800},[1751,1752,1753,1754,1755,1756],{"id":777,"answer":778,"question":779},{"id":781,"answer":782,"question":783},{"id":785,"answer":786,"question":787},{"id":789,"answer":790,"question":791},{"id":793,"answer":794,"question":795},{"id":797,"answer":798,"question":799},{},{"id":803,"data":1759,"type":42,"tunes":1760},{"text":805,"level":238},{},{"id":808,"data":1762,"type":808,"tunes":1770},{"title":810,"entries":1763},[1764,1765,1766,1767,1768,1769],{"term":569,"anchor":813,"definition":814},{"term":572,"anchor":816,"definition":817},{"term":575,"anchor":819,"definition":820},{"term":578,"anchor":418,"definition":822},{"term":587,"anchor":824,"definition":825},{"term":827,"anchor":828,"definition":829},{},{"id":832,"data":1772,"type":42,"tunes":1773},{"text":834,"level":238},{},{"id":837,"data":1775,"type":844,"tunes":1778},{"link":839,"meta":1776},{"image":1777,"title":842,"description":843},{"url":399},{},{"id":847,"data":1780,"type":844,"tunes":1783},{"link":849,"meta":1781},{"image":1782,"title":852,"description":853},{"url":399},{},{"id":856,"data":1785,"type":844,"tunes":1788},{"link":858,"meta":1786},{"image":1787,"title":861,"description":862},{"url":399},{},{"id":865,"data":1790,"type":844,"tunes":1793},{"link":867,"meta":1791},{"image":1792,"title":870,"description":871},{"url":399},{},"Post erfolgreich abgerufen",{"items":1796,"source":1879,"manualIds":1880,"manualMatchedIds":1881},[1797,1804,1811,1818,1825,1832,1837,1844,1851,1858,1865,1872],{"id":1798,"slug":1799,"title":1800,"excerpt":1801,"featuredImage":1802,"publishedAt":1803},"469","rag-failed-but-which-layer-actually-failed-a-diagnostic-method","Échec du RAG — mais quelle couche a réellement échoué ? Une méthode de diagnostic","Lorsqu'une réponse RAG est erronée, blâmer la récupération ou le modèle est trop vague. Cette méthode de diagnostic isole la couverture des sources, la construction de la requête, la récupération, le classement, l'assemblage du contexte, la génération, l'attribution des preuves et la fraîcheur—afin que la défaillance réelle puisse être reproduite et corrigée.","\u002Fuploads\u002F2026\u002F09\u002Frag-failed-but-which-layer-actually-failed-a-diagnostic-method-1790350847177-pior4c.webp","2026-09-24T19:39:00.000Z",{"id":1805,"slug":1806,"title":1807,"excerpt":1808,"featuredImage":1809,"publishedAt":1810},"460","ai-agent-reliability-why-the-final-answer-is-not-enough","Fiabilité des agents IA : pourquoi la réponse finale ne suffit pas","Une sortie correcte ne prouve pas un raisonnement correct, une exécution sûre ou un système digne de confiance.","\u002Fuploads\u002F2026\u002F09\u002Fai-agent-reliability-why-the-final-answer-is-not-enough-1788955466306-pl0qhz.webp","2026-09-09T04:01:00.000Z",{"id":1812,"slug":1813,"title":1814,"excerpt":1815,"featuredImage":1816,"publishedAt":1817},"477","computer-use-agents-why-a-successful-demo-can-still-be-an-unreliable-system","Agents d'utilisation de l'ordinateur : pourquoi une démonstration réussie peut tout de même être un système peu fiable","Les agents d'utilisation de l'ordinateur peuvent désormais accomplir d'impressionnants flux de travail sur navigateur et sur bureau, mais une seule exécution réussie prouve la capacité—non la fiabilité. Cet article montre comment tester la répétabilité, la robustesse environnementale, le contrôle à long horizon, la conscience de l'état, la vérification des résultats et la gestion sécurisée des objectifs.","\u002Fuploads\u002F2026\u002F09\u002Fcomputer-use-agents-why-a-successful-demo-can-still-be-an-unreliable-system-1790352854690-75qnrg.webp","2026-09-25T12:13:00.000Z",{"id":1819,"slug":1820,"title":1821,"excerpt":1822,"featuredImage":1823,"publishedAt":1824},"459","ollama-is-not-the-product-building-production-ready-open-llm-applications","Ollama n'est pas le produit : construire des applications Open-LLM prêtes pour la production","Exécuter un modèle local avec Ollama est facile. Construire une application Open-LLM prête pour la production est plus difficile : cela nécessite du RAG, du contrôle d'accès, de l'abstraction de fournisseur, de l'évaluation, de la journalisation, de la discipline de déploiement et une couche applicative contrôlée autour du modèle.","\u002Fuploads\u002F2026\u002F06\u002Follama-is-not-the-product-building-production-ready-open-llm-applications-1782679361640-h0usqf.webp","2026-06-28T16:39:00.000Z",{"id":1826,"slug":1827,"title":1828,"excerpt":1829,"featuredImage":1830,"publishedAt":1831},"473","openai-agents-api-vs-agents-sdk-vs-responses-api-what-should-you-build-on-in-2026","OpenAI Agents API vs Agents SDK vs Responses API : sur quoi devriez-vous construire en 2026 ?","La stack d'agents d'OpenAI a changé en septembre 2026. Ce guide d'architecture sépare l'API Agents, le SDK Agents, l'API Responses et le SDK Codex selon la propriété du runtime — afin que les équipes puissent choisir la bonne limite de contrôle au lieu de comparer des noms de produits.","\u002Fuploads\u002F2026\u002F09\u002Fopenai-agents-api-vs-agents-sdk-vs-responses-api-what-should-you-build-on-in-2026-1790351846714-zi7lus.webp","2026-09-25T11:56:00.000Z",{"id":1833,"slug":1834,"title":1834,"excerpt":10,"featuredImage":1835,"publishedAt":1836},"369","git-with-automatic-upload-and-synchronization-to-a-production-server","\u002Fuploads\u002F2024\u002F05\u002Fstep-by-step-guide-illustration-showing-the-process-of-setting-up-Git-with-auto-upload-and-synchronization-to-a-production-server-large.webp","2024-05-28T22:48:00.000Z",{"id":1838,"slug":1839,"title":1840,"excerpt":1841,"featuredImage":1842,"publishedAt":1843},"364","tipps-fuer-die-verbesserung-der-seo-suchmaschinenoptimierung","Maîtriser le flux de travail SEO : Stratégies d'optimisation essentielles pour la croissance organique","Un flux de travail SEO structuré est crucial pour une croissance organique durable. Découvrez les dix stratégies fondamentales, de la recherche de mots-clés et l'optimisation technique à la qualité du contenu et l'analyse des performances.","\u002Fuploads\u002F2026\u002F03\u002Ftipps-fuer-die-verbesserung-der-seo-suchmaschinenoptimierung-1774866098131-hwkzrg.webp","2024-01-26T06:35:00.000Z",{"id":1845,"slug":1846,"title":1847,"excerpt":1848,"featuredImage":1849,"publishedAt":1850},"470","what-should-an-ai-agent-remember-forget-recompute-or-retrieve-again","Que devrait mémoriser, oublier, recalculer ou récupérer à nouveau un agent IA ?","Les agents à exécution longue ne devraient pas tout retenir. Cet article propose un modèle de cycle de vie pratique pour décider de ce qui a sa place dans la mémoire durable, de ce qui devrait être récupéré à nouveau, de ce qu'il est plus sûr de recalculer et de ce qui devrait expirer ou être remplacé.","\u002Fuploads\u002F2026\u002F09\u002Fwhat-should-an-ai-agent-remember-forget-recompute-or-retrieve-again-1790351131087-iehz28.webp","2026-09-25T09:43:00.000Z",{"id":1852,"slug":1853,"title":1854,"excerpt":1855,"featuredImage":1856,"publishedAt":1857},"466","the-gpu-is-not-the-product-future-proof-private-ai-architecture","Le GPU n'est pas le produit : architecture d'IA privée pérenne","Une infrastructure d'IA privée ne devrait pas être conçue autour d'un seul GPU ou d'un seul modèle. Une approche plus résiliente combine des GPU d'inférence rapides, des systèmes d'IA riches en mémoire, des nœuds d'IA physique et des modèles cloud de pointe optionnels derrière une couche de routage prenant en compte les capacités.","\u002Fuploads\u002F2026\u002F09\u002Fthe-gpu-is-not-the-product-future-proof-private-ai-architecture-1790140878812-8hsl39.webp","2026-09-23T01:19:00.000Z",{"id":1859,"slug":1860,"title":1861,"excerpt":1862,"featuredImage":1863,"publishedAt":1864},"472","why-more-context-can-make-ai-answers-worse","Pourquoi plus de contexte peut rendre les réponses de l'IA pires","Une fenêtre de contexte plus grande ne garantit pas une meilleure réponse. Cet article explique comment la dilution du signal, les preuves contradictoires, l'état obsolète, la sensibilité à la position et la compression avec perte peuvent réduire la fiabilité de l'IA — et présente un test pratique de pression de contexte.","\u002Fuploads\u002F2026\u002F09\u002Fwhy-more-context-can-make-ai-answers-worse-1790351615793-2ntv2v.webp","2026-09-25T11:51:00.000Z",{"id":1866,"slug":1867,"title":1868,"excerpt":1869,"featuredImage":1870,"publishedAt":1871},"476","mcp-vs-a2a-vs-ucp-vs-ap2-vs-a2ui-the-agent-protocol-stack-explained","MCP vs A2A vs UCP vs AP2 vs A2UI : La pile de protocoles d'agent expliquée","MCP, A2A, UCP, AP2 et A2UI sont souvent présentés comme des standards d'agents concurrents. Ils résolvent principalement des problèmes d'interopérabilité différents. Ce guide associe chaque protocole à la frontière qu'il standardise réellement—et montre comment ils peuvent fonctionner ensemble dans un seul système de production.","\u002Fuploads\u002F2026\u002F09\u002Fmcp-vs-a2a-vs-ucp-vs-ap2-vs-a2ui-the-agent-protocol-stack-explained-1790352625869-2ezle0.webp","2026-09-25T12:09:00.000Z",{"id":1873,"slug":1874,"title":1875,"excerpt":1876,"featuredImage":1877,"publishedAt":1878},"467","the-answer-validity-boundary-the-missing-layer-between-relevance-and-reliable-ai-answers","La frontière de validité des réponses : la couche manquante entre la pertinence et les réponses fiables de l'IA","Une source peut être pertinente, faisant autorité et pourtant être erronée pour la question posée. La couche manquante est l'applicabilité : les conditions dans lesquelles une réponse est valable, et les changements qui obligent à la reconsidérer. Cet article présente la Frontière de Validité de la Réponse comme un modèle de conception de source pour les humains, la recherche par IA et les systèmes RAG.","\u002Fuploads\u002F2026\u002F09\u002Fthe-answer-validity-boundary-the-missing-layer-between-relevance-and-reliable-ai-answers-1790272901306-1g5jly.webp","2026-09-24T11:59:00.000Z","fallback",[],[]]