[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"portal-settings:stajic:sr":3,"public-menus:all":38,"post:what-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs:sr":205,"related:post:what-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs:sr:1":2141},{"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","sr","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":2140},{"id":207,"title":208,"slug":209,"content":210,"contentJson":211,"excerpt":1069,"featuredImage":1070,"featuredImageAlt":1071,"featuredImageCaption":10,"featuredImageTitle":10,"featuredImageCopyright":10,"featuredImageAuthor":10,"featuredImageSourceUrl":10,"featuredImageLicense":10,"featuredImageIsAiGenerated":43,"status":1072,"publishedAt":1073,"createdAt":1074,"updatedAt":1075,"seoLocalePaths":1076,"categories":1085,"author":1102,"translations":1107},"483","Šta je AI rešenje arhitekta? Granice sistema, odgovornosti i kompromisi","what-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u003Cp>\u003Cstrong>AI Solution Architect\u003C\u002Fstrong> prevodi poslovnu ili proizvodnu potrebu u arhitekturu konkretnog AI rešenja. Uloga definiše granice sistema i značajne izbore u okviru aplikacione logike, autoritativnih podataka, pretrage i konteksta, modela i provajdera, alata ili agenata, identiteta i dozvola, bezbednosti, izvršavanja i implementacije, nadzora, evaluacije, troškova i operativnog ponašanja. To nije samo izbor modela ili inženjering upita: arhitektonska odgovornost je da celo rešenje bude implementabilno, upravljivo, testabilno i operabilno.\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\">Direktan odgovor\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">\u003Cstrong>AI Solution Architect projektuje kompletno AI rešenje, ne samo AI model.\u003C\u002Fstrong> Uloga povezuje zahteve i nefunkcionalne zahteve sa arhitektonskim odlukama, komponuje neophodne slojeve aplikacije\u002Fpodataka\u002Fmodela\u002Falata\u002Fizvršavanja, čini granice poverenja i otkaza eksplicitnim i definiše kako će implementirani sistem biti validiran i operisan.\u003C\u002Fdiv>\u003C\u002Faside>\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\">Napomena o terminologiji\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">\u003Cstrong>AI Solution Architect je praktična oznaka uloge, a ne univerzalno standardizovan naziv radnog mesta.\u003C\u002Fstrong> ISO\u002FIEC\u002FIEEE 42010:2022 standardizuje koncepte za opise arhitekture; ne definiše ovu radnu ulogu. Organizacije mogu raspodeliti odgovornosti na više ljudi. U ovom članku, termin označava arhitektonsku odgovornost za jedno konkretno AI rešenje ili radno opterećenje.\u003C\u002Fdiv>\u003C\u002Faside>\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\">Napomena o aktuelnim izvorima — 8. oktobar 2026.\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">Ovde navedeni arhitektonski principi su namerno neutralni prema provajderu, dok se aktuelne smernice provajdera koriste kao dokaz implementacije. NIST AI RMF 1.0 je trenutno u reviziji; NIST AI 600-1 ostaje objavljeni Generative AI Profile. Smernice Microsoft-a i AWS-a citirane u nastavku odražavaju aktuelne proizvodne brige kao što su identitet, granice podataka, apstrakcija modela, bezbednost, nadzor, evaluacija, pouzdanost i troškovi.\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Cnav class=\"editorjs-toc\" data-editorjs-toc=\"true\" aria-label=\"Sadržaj\">\u003Cstrong class=\"editorjs-toc__title\">Sadržaj\u003C\u002Fstrong>\u003Col class=\"editorjs-toc__list editorjs-toc__list--depth-0\">\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-6\" class=\"editorjs-toc__link\">Šta AI Solution Architect zapravo projektuje?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-10\" class=\"editorjs-toc__link\">Najjednostavniji primer\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-14\" class=\"editorjs-toc__link\">Gde se jednostavan primer zaustavlja\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-17\" class=\"editorjs-toc__link\">Mapa arhitektonskih odgovornosti\u003C\u002Fa>\u003Col class=\"editorjs-toc__list editorjs-toc__list--depth-1\">\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-20\" class=\"editorjs-toc__link\">1. Pretvorite potrebu proizvoda u arhitektonske zahteve\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-23\" class=\"editorjs-toc__link\">2. Dizajnirajte autoritativne podatke, pretragu i kontekst\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-26\" class=\"editorjs-toc__link\">3. Tretirajte modele i provajdere kao zavisnosti, ne kao ceo sistem\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-29\" class=\"editorjs-toc__link\">4. Arhitektirajte alate, radnje i granice agenata\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-32\" class=\"editorjs-toc__link\">5. Učinite granice poverenja i dozvole eksplicitnim\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-35\" class=\"editorjs-toc__link\">6. Odlučite gde sistem zaista radi\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-38\" class=\"editorjs-toc__link\">7. Definišite evaluaciju, observabilnost i operativno prihvatanje\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-41\" class=\"editorjs-toc__link\">Šta bi ova uloga trebalo da proizvede?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-44\" class=\"editorjs-toc__link\">Rad se uglavnom sastoji od kompromisa, a ne od izbora „najbolje prakse“\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-47\" class=\"editorjs-toc__link\">Kako se ovo razlikuje od srodnih uloga?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-51\" class=\"editorjs-toc__link\">Dokazi implementacije: kako se ove granice pojavljuju u mom radu\u003C\u002Fa>\u003Col class=\"editorjs-toc__list editorjs-toc__list--depth-1\">\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-53\" class=\"editorjs-toc__link\">SenseFlow: potreba → zahtevi → arhitektura → validacija\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-57\" class=\"editorjs-toc__link\">Aaasaasa AI klijent: razdvojite koncepte pre njihovog integrisanja\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-61\" class=\"editorjs-toc__link\">Kako trenutni arhitektonski okviri podržavaju ovaj širi obim\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-65\" class=\"editorjs-toc__link\">Uobičajene zablude\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-67\" class=\"editorjs-toc__link\">Načini neuspeha koje AI arhitekta rešenja treba da spreči\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-69\" class=\"editorjs-toc__link\">Praktičan redosled odlučivanja\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-71\" class=\"editorjs-toc__link\">Rubni slučajevi i granice uloge\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-75\" class=\"editorjs-toc__link\">Šta bi promenilo ovaj odgovor?\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-78\" class=\"editorjs-toc__link\">AI Solution Architect kontrolna lista\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-80\" class=\"editorjs-toc__link\">Zaključak\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-85\" class=\"editorjs-toc__link\">Povezano kanonsko znanje\u003C\u002Fa>\u003C\u002Fli>\u003Cli class=\"editorjs-toc__item\">\u003Ca href=\"#section-88\" class=\"editorjs-toc__link\">Primarni izvori i trenutne arhitektonske smernice\u003C\u002Fa>\u003C\u002Fli>\u003C\u002Fol>\u003C\u002Fnav>\n\u003Ch2 id=\"section-6\">Šta AI Solution Architect zapravo projektuje?\u003C\u002Fh2>\n\u003Cp>Predmet rada je \u003Cstrong>rešenje\u003C\u002Fstrong>: kompletan socio-tehnički sistem koji potrebu pretvara u korisno, kontrolisano ponašanje. Model može biti centralni za taj sistem, ali je i dalje samo jedna zavisnost. Isti model može učestvovati u bezbednom internom pretraživačkom asistentu, nebezbednom agentu sa prekomernim privilegijama, korisničkoj funkciji sa malim kašnjenjem ili skupom prototipu koji se ne može ekonomski operisati. Arhitektura određuje te razlike.\u003C\u002Fp>\n\u003Cp>Korisna granica je stoga: \u003Cstrong>poslovni ishod → zahtevi → odgovornosti sistema → arhitektonske odluke → implementacija → validacija → operacija\u003C\u002Fstrong>. AI Solution Architect radi duž ovog lanca, sarađujući sa produktom, inženjeringom, podacima, bezbednošću, infrastrukturom, upravljanjem i domenskim specijalistima.\u003C\u002Fp>\n\u003Csection class=\"editorjs-comparison my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Rešenje je šire od modela\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\">Pitanje usmereno na model\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\">Pitanje arhitekture rešenja\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\">Sposobnost\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Which model can generate or reason well enough?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Which combination of model, data, application logic, retrieval, tools and controls produces the required behavior?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Podaci\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">What context can fit in the prompt?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">What is authoritative, who may access it, how is it retrieved, versioned, filtered and cited?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Bezbednost\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Does the provider offer security features?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">What are the trust boundaries, identities, permissions, secrets, data flows and failure containment mechanisms?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Operacije\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">What is the token latency?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">How is the complete workload deployed, observed, evaluated, recovered, versioned and cost-controlled?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Promena\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Can we switch models?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Which dependencies are abstracted, what changes require an ADR, and how do we validate that a replacement still meets requirements?\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Ch2 id=\"section-10\">Najjednostavniji primer\u003C\u002Fh2>\n\u003Cp>Zamislite da kompanija želi internog asistenta koji odgovara na pitanja tehničara na osnovu priručnika za održavanje i operativnih procedura. Vidljiva funkcija zvuči jednostavno: ukucajte pitanje i primite odgovor sa izvorima.\u003C\u002Fp>\n\u003Cp>Arhitektonsko pitanje je mnogo veće. Koji dokumenti su autoritativni? Kako se korisnici autentifikuju? Da li pretraga mora poštovati dozvole odeljenja ili lokacije? Da li odgovor sme koristiti samo pronađene dokaze? Koji model je prihvatljiv za klasifikaciju podataka? Može li cloud provajder primiti sadržaj? Šta se dešava kada pretraga ne pronađe ništa? Kako se proizvode citati? Kako se evaluira kvalitet odgovora? Koje kašnjenje i trošak su prihvatljivi? Ko može videti logove i šta se u njima sme čuvati?\u003C\u002Fp>\n\u003Csection class=\"editorjs-process my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Od potrebe do operabilnog AI rešenja\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. Definišite ishod\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Razjasnite korisnika, poslovnu vrednost, granicu zadatka i šta znači uspešan odgovor ili akcija.\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. Prikupite zahteve\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Učinite eksplicitnim funkcionalne zahteve, nefunkcionalne zahteve, ograničenja, pravila o podacima, toleranciju rizika i kriterijume prihvatanja.\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. Uspostavite granice\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Identifikujte korisnike, identitete, aplikacije, autoritativne podatke, zavisnosti od modela\u002Fprovajdera, alate, eksterne sisteme i zone poverenja.\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. Projektujte arhitekturu\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Izaberite obrasce za podatke\u002Fpretragu, model, orkestraciju, alate, dozvole, izvršavanje, implementaciju, rezervne opcije i nadzor.\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. Zabeležite značajne odluke\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Sačuvajte arhitektonske izbore, alternative, kompromise i posledice kako bi kasnije promene ostale razumljive.\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\">6. Implementirajte i integrišite\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Pretvorite arhitekturu u aplikacioni kod, API-je, politike, infrastrukturu, radne tokove i operativne kontrole.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">7\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">7. Validirajte i operišite\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Testirajte kvalitet, bezbednost, pouzdanost, troškove i korisničke ishode; nadzirite stvarno radno opterećenje i vraćajte dokaze u odluke.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Ch2 id=\"section-14\">Gde se jednostavan primer zaustavlja\u003C\u002Fh2>\n\u003Cp>Proof of concept često može preskočiti arhitekturu koju produkcija ne može. Programer može hardkodirati jednog provajdera, koristiti deljeni API ključ, smestiti sve dokumente u jedan indeks, pokretati pretragu bez filtriranja po korisničkom kontekstu, logovati upite doslovno i procenjivati kvalitet ručno. To može demonstrirati izvodljivost, ali ne uspostavlja produkcijsku arhitekturu.\u003C\u002Fp>\n\u003Cp>Produkcija uvodi ograničenja koja interaguju: izolaciju zakupaca ili korisnika, privatnost, rezidentnost podataka, propusnost, kašnjenje, troškove, kvote provajdera, rezervno ponašanje, revizibilnost, promene verzija modela, kvalitet pretrage, dozvole alata, odgovor na incidente i životni ciklus implementacije. Posao arhitekte nije da maksimizuje svaki kvalitet odjednom; već da kompromise učini eksplicitnim i projektuje rešenje koje zadovoljava stvarni skup prioriteta.\u003C\u002Fp>\n\u003Ch2 id=\"section-17\">Mapa arhitektonskih odgovornosti\u003C\u002Fh2>\n\u003Cp>Tačna podela varira od organizacije do organizacije, ali sledeća mapa obuhvata ponavljajuće odgovornosti arhitekture AI rešenja. Arhitekta možda neće lično implementirati svaki sloj; odgovornost je da slojevi funkcionišu koherentno i da kritične odluke ostanu sledljive.\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\">Arhitektonska oblast\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Pitanja koja AI Solution Architect mora da reši\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Tipični izlazi\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Ishod i obim\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Ko je korisnik? Koji zadatak je u obimu? Šta sistem ne sme da radi? Šta predstavlja uspeh?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Kontekst rešenja, granica sposobnosti, kriterijumi prihvatanja\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Zahtevi i NFR-ovi\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Koja ograničenja kvaliteta, bezbednosti, dostupnosti, latencije, troškova, rezidentnosti i usklađenosti se primenjuju?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Mapa zahteva, NFR-ovi, ograničenja, kriterijumi validacije\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Aplikacija i orkestracija\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Gde se završava deterministička aplikaciona logika, a počinje AI ponašanje? Kako se koordinišu tokovi rada?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Model komponenti, API-ji, granice orkestracije, putevi otkaza\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Autoritativni podaci i pretraga\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Šta je izvor istine? Kako se podaci unose, autorizuju, pronalaze, filtriraju, rangiraju i citiraju?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Tokovi podataka, arhitektura pretrage, metapodaci i pravila autorizacije\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Sloj modela i provajdera\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Koje sposobnosti su potrebne? Koja ograničenja provajdera\u002Fruntime-a su važna? Šta treba apstrahovati?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Odluka o modelu\u002Fprovajderu, politika rutiranja\u002Ffallback-a, granica apstrakcije\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Alati i agenti\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Koje radnje sistem može da preduzme? Koje radnje zahtevaju odobrenje? Kako se sprovode identiteti i dozvole alata?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Ugovori alata, granice agenata, pravila odobravanja i najmanjih privilegija\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Identitet i bezbednost\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Koji ljudski i mašinski identiteti postoje? Gde se čuvaju tajne? Koje granice poverenja se prelaze?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Model pretnji\u002Fgranica poverenja, propagacija identiteta, dizajn tajni i autorizacije\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Runtime i implementacija\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Gde se komponente izvršavaju? Šta je lokalno, cloud, edge ili hibridno? Koje pretpostavke o mreži i dostupnosti postoje?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Prikaz implementacije, runtime topologija, odluke o okruženju i povezivanju\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Evaluacija i observabilnost\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Kako se kvalitet meri pre i posle izdanja? Koji tragovi, metrike, logovi i dokazi su potrebni?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Plan evaluacije, telemetrija, revizorski trag, kapije izdanja\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Operacije i promene\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Kako se verzije modela\u002Fpromptova\u002Fkonfiguracije\u002Fpodataka menjaju, vraćaju i podržavaju?\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Operativni model, kontrole životnog ciklusa, ADR-ovi, runbook-ovi, pravila promena\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch3 id=\"section-20\">1. Pretvorite potrebu proizvoda u arhitektonske zahteve\u003C\u002Fh3>\n\u003Cp>AI arhitektura počinje pre izbora modela. Arhitekta prvo utvrđuje šta se od rešenja očekuje da postigne i pod kojim ograničenjima. To uključuje funkcionalno ponašanje, ali i NFR-ove i politike koje sužavaju prostor dizajna: bezbednost, pouzdanost, latenciju, privatnost, rezidentnost, održivost, troškove i operativnu podršku.\u003C\u002Fp>\n\u003Cp>Ovde je važna razlika iz A02: zahtev kao što je „neovlašćeni korisnici ne smeju da pronađu ograničene dokumente“ nije arhitektonska odluka. To je pokretač. Odluke o propagaciji identiteta, particionisanju indeksa, filtriranju metapodataka, granicama API-ja i sprovođenju autorizacije su arhitektonski odgovori koji kasnije moraju biti validirani.\u003C\u002Fp>\n\u003Ch3 id=\"section-23\">2. Dizajnirajte autoritativne podatke, pretragu i kontekst\u003C\u002Fh3>\n\u003Cp>AI sistemi često padaju na granici između ponašanja modela i istine preduzeća. Arhitekta mora da definiše koji izvori su autoritativni, šta znače svežina i poreklo, kako kontrola pristupa stiže do pretrage i kako pronađeni dokazi postaju kontekst modela. Vektorska baza podataka, model za ugrađivanje ili RAG biblioteka nisu sami po sebi arhitektura.\u003C\u002Fp>\n\u003Cp>Microsoft-ove trenutne smernice za AI radna opterećenja eksplicitno prave istu razliku: aplikacioni kod ne treba da zaobilazi granice pristupa podacima; kontekst korisnika ili zakupca treba da se propagira u pretragu i filtriranje; podaci za utemeljenje moraju biti dizajnirani za pretraživost, a istovremeno da zadovolje bezbednosne i regulatorne zahteve.\u003C\u002Fp>\n\u003Ch3 id=\"section-26\">3. Tretirajte modele i provajdere kao zavisnosti, ne kao ceo sistem\u003C\u002Fh3>\n\u003Cp>Izbor modela je važan, ali treba da bude vođen potrebnim sposobnostima i ograničenjima. Arhitekta razmatra kvalitet rezonovanja ili generisanja, modalitet, ograničenja konteksta, latenciju, rukovanje podacima, lokaciju implementacije, dostupnost provajdera, troškove, observabilnost i rizik zamene.\u003C\u002Fp>\n\u003Cp>Apstrakcija provajdera nije automatski „bolja arhitektura“. Dodaje inženjerske troškove i može da sakrije sposobnosti specifične za provajdera. Opravdana je kada su prenosivost, fallback, razdvajanje politika ili rutiranje ka više provajdera eksplicitni zahtevi. U suprotnom, direktna integracija može biti bolja odluka. Poenta je da kompromis bude nameran.\u003C\u002Fp>\n\u003Ch3 id=\"section-29\">4. Arhitektirajte alate, radnje i granice agenata\u003C\u002Fh3>\n\u003Cp>Kada AI sistem može da poziva alate, menja podatke, šalje poruke, pokreće kod ili upravlja poslovnim sistemima, arhitektonski rizik se menja. Pristup alatima zahteva sopstveni model identiteta i autorizacije. Sposobnost modela da zatraži radnju nije isto što i dozvola da je izvrši.\u003C\u002Fp>\n\u003Cp>Za agentna radna opterećenja, trenutne AWS smernice naglašavaju dodatne dimenzije kao što su identiteti agenata, pristup alatima, orkestracija, ljudski nadzor, praćenje, rukovanje otkazima i troškovi iterativnih petlji rezonovanja. To su pitanja rešenja čak i kada okvir sakriva neke od mehanizama implementacije.\u003C\u002Fp>\n\u003Ch3 id=\"section-32\">5. Učinite granice poverenja i dozvole eksplicitnim\u003C\u002Fh3>\n\u003Cp>Produkcijsko AI rešenje ima više granica poverenja: pregledač ili klijent, aplikacioni backend, AI orkestracija, servisi za pretragu\u002Fpodatke, provajderi modela, API-ji alata, lokalni runtime-ovi i eksterni sistemi. Svaka granica treba da odgovori: ko poziva, u čije ime, sa kojim akreditivom, za koji resurs, sa kojim revizorskim tragom i sa kojim ograničavanjem otkaza?\u003C\u002Fp>\n\u003Cp>Bezbednost se ne može odložiti na „guardrail“ oko modela. Microsoft-ove smernice za AI radna opterećenja eksplicitno postavljaju bezbednost kroz sve arhitektonske slojeve i zahtevaju upravljanje identitetom\u002Fpristupom, zaštitu podataka, kontrolu sadržaja i bezbednost životnog ciklusa. NIST takođe tretira upravljanje i upravljanje rizikom kao kontinuirano kroz AI životni ciklus.\u003C\u002Fp>\n\u003Ch3 id=\"section-35\">6. Odlučite gde sistem zaista radi\u003C\u002Fh3>\n\u003Cp>„Lokalni AI“, „cloud AI“ i „hibridni AI“ su arhitektonske izjave samo kada su putevi izvršavanja i podataka precizni. Lokalni desktop proces i dalje može da poziva cloud model. Aplikacija hostovana u cloudu može da pronalazi podatke iz on-premises izvora podataka. Air-gapped rešenje ima potpuno drugačija ograničenja ažuriranja, distribucije modela i observabilnosti.\u003C\u002Fp>\n\u003Cp>Arhitekta stoga razdvaja \u003Cstrong>lokaciju izvršavanja\u003C\u002Fstrong>, \u003Cstrong>lokaciju zaključivanja\u003C\u002Fstrong>, \u003Cstrong>lokaciju podataka\u003C\u002Fstrong> i \u003Cstrong>kontrolnu ravan\u003C\u002Fstrong>. Njihovo poistovećivanje stvara lažnu sigurnost i pretpostavke o raspoređivanju.\u003C\u002Fp>\n\u003Ch3 id=\"section-38\">7. Definišite evaluaciju, observabilnost i operativno prihvatanje\u003C\u002Fh3>\n\u003Cp>AI ponašanje je delimično nedeterminističko, pa definicija izdanja ne može da se osloni samo na konvencionalne unit testove. Arhitekturi su potrebni merljivi kriterijumi prihvatanja: uspeh zadatka, utemeljenost ili ispravnost citiranja gde je relevantno, ponašanje odbijanja, bezbednost alata, latencija, trošak, pouzdanost i bezbednosni testovi. Tačne metrike zavise od slučaja upotrebe.\u003C\u002Fp>\n\u003Cp>Microsoft-ove trenutne Well-Architected AI smernice tretiraju monitoring kao kontinuiran i primenjuju ga na ponašanje modela, promptove\u002Fkompletiranja, anomalije, bezbednost i kapije kvaliteta u produkciji. AWS na sličan način tretira observabilnost, upravljanje životnim ciklusom i sledljivost modela\u002Fpromptova kao pitanja operativne arhitekture.\u003C\u002Fp>\n\u003Ch2 id=\"section-41\">Šta bi ova uloga trebalo da proizvede?\u003C\u002Fh2>\n\u003Cp>Arhitektura nije slajd prezentacija. Korisni izlazi su artefakti koji omogućavaju inženjeringu, bezbednosti, proizvodu i operacijama da donose konzistentne odluke i kasnije razumeju zašto sistem postoji u svom trenutnom obliku.\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\">Artefakt\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Svrha\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Kontekst i granice rešenja\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Prikazuje korisnike, eksterne sisteme, glavne odgovornosti i šta je van obima\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Mapa zahteva\u002FNFR\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Povezuje potrebe proizvoda i ograničenja sa arhitektonskim radom i validacijom\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Prikazi komponenti i tokova podataka\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Prikazuje interakcije aplikacije, podataka\u002Fpretrage, modela, alata, identiteta i izvršnog okruženja\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Model poverenja i dozvola\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Eksplicitno prikazuje identitete, tajne, autorizaciju, osetljive podatke i radnje visokog rizika\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Zapisi o arhitektonskim odlukama\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Čuva značajne izbore, alternative, kompromise, status i posledice\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Plan evaluacije i prihvatanja\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Definiše dokaze potrebne da bi se tvrdilo da rešenje ispunjava očekivanja kvaliteta i bezbednosti\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Prikaz raspoređivanja i operacija\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Definiše okruženja, lokacije izvršavanja, observabilnost, vraćanje unazad, incidente i odgovornosti životnog ciklusa\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Veze sledljivosti\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Povezuje zahteve, odluke, implementacioni rad, testove i operativne dokaze\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"section-44\">Rad se uglavnom sastoji od kompromisa, a ne od izbora „najbolje prakse“\u003C\u002Fh2>\n\u003Cp>Arhitektura postoji jer su poželjni kvaliteti u sukobu. Model sa nižim troškovima može smanjiti kvalitet. Sposobniji model može povećati latenciju ili ograničenja upravljanja podacima. Agresivno keširanje može poboljšati troškove i brzinu, ali otežati svežinu. Autonomniji agenti mogu smanjiti ljudski napor, ali povećati domet štete i zahteve za revizijom.\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\">Odluka\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Potencijalna korist\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Potencijalni trošak \u002F rizik\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Arhitektonsko pitanje\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Upravljani cloud model\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Brzo usvajanje, jake upravljane mogućnosti\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Spoljna zavisnost, ograničenja podataka i troškova\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li radno opterećenje dozvoljava provajdera\u002Fputanju podataka i ispunjava potrebe otpornosti?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Lokalno\u002Fsamostalno hostovano zaključivanje\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Kontrola, offline\u002Fprivatne opcije\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Hardver, operacije, teret životnog ciklusa modela\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li je korist od kontrole vredna operativne odgovornosti?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Integracija sa jednim provajderom\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Jednostavnija implementacija, pune mogućnosti provajdera\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Veća koncentracija prebacivanja\u002Fotkaza\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su prenosivost ili rezervna opcija zaista potrebni?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Apstrakcija provajdera\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Prenosivost, rutiranje i razdvajanje politika\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Rizik najnižeg zajedničkog imenioca, više koda\u002Ftestova\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Koje razlike moraju ostati vidljive, a ne apstrahovane?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Veliki kontekst\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Više informacija po zahtevu\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Latencija, trošak, razblaživanje pažnje, površina curenja\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li podatke treba preuzimati\u002Ffiltrirati umesto uvek ubacivati?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Moćni alati \u002F autonomija\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Više automatizacije od početka do kraja\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Veći privilegiji i domet štete pri otkazu\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Koje radnje zahtevaju najmanje privilegije, potvrdu ili ljudsko odobrenje?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Stroga validacija i evidentiranje\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Bolji dokazi i operacije\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Latencija, skladištenje, privatnost i trošak složenosti\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Koji dokazi su potrebni za ovaj nivo rizika?\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"section-47\">Kako se ovo razlikuje od srodnih uloga?\u003C\u002Fh2>\n\u003Cp>Titule se u velikoj meri preklapaju među kompanijama. Korisna razlika je \u003Cstrong>obim arhitektonske odgovornosti\u003C\u002Fstrong>, a ne HR oznaka.\u003C\u002Fp>\n\u003Csection class=\"editorjs-comparison my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Srodne uloge odgovaraju na različita primarna pitanja\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\">Uloga\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\">Primarni arhitektonski fokus\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\">AI Solution Architect\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">One concrete AI-enabled solution\u002Fworkload\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">How requirements, data, models, tools, security, runtime and operations fit together to deliver the target outcome\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">AI Platform Architect\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Reusable AI platform capabilities across many solutions\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Shared provider gateways, model access, identity, evaluation, retrieval services, observability, deployment patterns and developer experience\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Enterprise AI Architect\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Organization\u002Fportfolio-level target architecture\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Capability landscape, governance, integration principles, shared platforms, standards, sourcing and strategic constraints across domains\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">AI \u002F ML Engineer\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Implementation of AI\u002FML behavior and pipelines\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Models, data, inference, evaluation, application logic and engineering tasks within the architecture\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Security Architect\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Security architecture across systems\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Threats, identity, authorization, data protection, controls, assurance and compliance boundaries\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Cth class=\"border border-gray-300 px-4 py-3 text-left font-semibold dark:border-gray-700\">Product \u002F Delivery Lead\u003C\u002Fth>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Outcome, scope, prioritization and delivery system\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-3 dark:border-gray-700\">Why\u002Fwhat to build, sequencing, stakeholders, milestones, acceptance and value realization\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Cp>U malom produktnom timu, jedna osoba može pokrivati nekoliko ovih obima. U velikom preduzeću, to mogu biti odvojene uloge sa formalnim odborima za pregled. Arhitektonska odgovornost ne nestaje kada se titula promeni.\u003C\u002Fp>\n\u003Ch2 id=\"section-51\">Dokazi implementacije: kako se ove granice pojavljuju u mom radu\u003C\u002Fh2>\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\">Dokazi implementacije, ne univerzalno pravilo\u003C\u002Fstrong>\u003Cdiv class=\"text-gray-700 dark:text-gray-200\">Primeri ispod su \u003Cstrong>originalni dokazi implementacije\u002Fprojekta\u003C\u002Fstrong>. Oni pokazuju kako sam razdvojio potrebe proizvoda, zahteve, arhitekturu, izvršno okruženje, model\u002Fprovajdera, dozvole i validaciju u stvarnom projektnom radu. Oni nisu tvrdnje da svaka organizacija mora koristiti istu strukturu i ne impliciraju usvajanje od strane kupaca ili raspoređivanje na nivou preduzeća.\u003C\u002Fdiv>\u003C\u002Faside>\n\u003Ch3 id=\"section-53\">SenseFlow: potreba → zahtevi → arhitektura → validacija\u003C\u002Fh3>\n\u003Cp>U projektu SenseFlow Source of Truth, tehnologija je eksplicitno podređena Product Vision-u. Razvojna struktura se kreće od problema i vizije proizvoda kroz korisničke potrebe, vrednost, obim, epove, priče i kriterijume prihvatanja do arhitekture, implementacije, validacije i iteracije.\u003C\u002Fp>\n\u003Cp>Zahtevi su dizajnirani tako da budu sledljivi od Cilja proizvoda → Sposobnosti → Epika → Korisničke priče → Kriterijuma prihvatanja → Tehničkih zadataka. Gde je praktično, uključuju funkcionalne zahteve, nefunkcionalne zahteve, zavisnosti, rizike, pretpostavke, kriterijume prihvatanja i metode validacije. Značajne odluke čuvaju odluku, razlog, alternative, kompromise, status i datum\u002Fverziju.\u003C\u002Fp>\n\u003Cp>To je arhitektonski rad pre nego što se izabere konkretan AI okvir ili model: štiti vezu između namere proizvoda i tehničkih odluka i čini kasnije promene preglednim umesto implicitnim.\u003C\u002Fp>\n\u003Ch3 id=\"section-57\">Aaasaasa AI klijent: razdvojite koncepte pre njihovog integrisanja\u003C\u002Fh3>\n\u003Cp>Aaasaasa AI klijent pruža primer na nivou implementacije. Njegov AI Hub namerno razdvaja \u003Cstrong>agenta\u002Fklijenta\u003C\u002Fstrong>, \u003Cstrong>provajdera\u003C\u002Fstrong>, \u003Cstrong>model\u003C\u002Fstrong>, \u003Cstrong>lokaciju izvršavanja\u002Fkonekcije\u003C\u002Fstrong>, \u003Cstrong>dozvole\u003C\u002Fstrong> i \u003Cstrong>veb klijenta\u003C\u002Fstrong>. Lokalno izvršavanje se ne pretpostavlja da znači lokalno zaključivanje, a dozvole se tretiraju kao politika izvršavanja\u002Falata, a ne kao svojstvo modela.\u003C\u002Fp>\n\u003Cp>Desktop arhitektura takođe definiše granicu poverenja: Nuxt renderer nije pouzdan u odnosu na Electron main. Uzak preload i validirani IPC posreduju u pristupu AI servisima, podešavanjima, enkriptovanim tajnama, servisima radnog prostora\u002Fpodataka i izvršnim okruženjima. Cloud akreditivi ostaju u privilegovanom glavnom procesu; renderer kod prima normalizovano stanje umesto sirovih tajni ili neograničenog pristupa operativnom sistemu.\u003C\u002Fp>\n\u003Cp>Odluke o rutiranju su takođe arhitektonske. Implementacija ne prelazi tiho sa lokalne rute na plaćeno cloud zaključivanje; cloud ruta zahteva eksplicitnu potvrdu. Direktan chat nema podrazumevano alate za fajl sistem ili shell, dok izvršavanje agenta primenjuje izabrani radni prostor i profil dozvola. To su odluke na nivou rešenja o poverenju, troškovima, izvršavanju i očekivanjima korisnika—ne karakteristike modela.\u003C\u002Fp>\n\u003Ch2 id=\"section-61\">Kako trenutni arhitektonski okviri podržavaju ovaj širi obim\u003C\u002Fh2>\n\u003Cp>ISO\u002FIEC\u002FIEEE 42010:2022 pruža opštu disciplinu za opise arhitekture kroz softver, sisteme i preduzeća. Namerno je širi od AI i ne propisuje jednu metodu arhitekture ili naziv radnog mesta. To ga čini korisnim ovde kao granicu: AI arhitektura rešenja je i dalje arhitektura, sa interesima zainteresovanih strana, više pogleda i značajnim odnosima koji moraju biti jasno izraženi.\u003C\u002Fp>\n\u003Cp>NIST AI RMF 1.0 uokviruje upravljanje AI rizikom kroz \u003Cstrong>Govern, Map, Measure i Manage\u003C\u002Fstrong> i naglašava da upravljanje rizikom treba da bude kontinuirano tokom životnog ciklusa AI sistema. Generativni AI profil (NIST AI 600-1) prilagođava taj okvir GAI rizicima i organizacionim prioritetima. To pojačava da arhitektura ne može da se zaustavi na funkcionalnim performansama modela.\u003C\u002Fp>\n\u003Cp>Microsoft-ove trenutne Azure Well-Architected AI smernice razdvajaju dizajn aplikacije, platformu aplikacije, podatke za obuku, podatke za utemeljenje i pitanja platforme podataka i više puta ih povezuju sa pouzdanošću, bezbednošću, operativnom izvrsnošću, performansama i troškovima. AWS-ova Generativna AI i Agentic AI sočiva takođe tretiraju observabilnost, bezbednost, pouzdanost, životni ciklus modela\u002Falata, troškove i ljudski nadzor kao arhitektonske brige.\u003C\u002Fp>\n\u003Ch2 id=\"section-65\">Uobičajene zablude\u003C\u002Fh2>\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\">Zabluda\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Ispravka\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„Arhitekta bira LLM.“\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Izbor modela je jedna odluka unutar veće arhitekture rešenja.\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„Prompt inženjering je arhitektura.“\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Promptovi utiču na ponašanje, ali ne definišu identitet, pristup podacima, granice poverenja, implementaciju, dozvole alata ili operacije.\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„RAG rešava znanje u preduzeću.“\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Pretraga je samo jedan podsistem; autorizacija, poreklo, svežina, dokazi, indeksiranje, evaluacija i upravljanje izvorima i dalje zahtevaju dizajn.\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„Lokalno izvršavanje znači privatni\u002Flokalni AI.“\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Lokacije izvršavanja, zaključivanja, podataka i kontrolne ravni su odvojena arhitektonska svojstva.\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„Ako dobavljač nudi zaštitne mere, bezbednost je pokrivena.“\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Bezbednost obuhvata identitet, autorizaciju, tajne, tokove podataka, alate, evidentiranje, implementaciju, ljudsko odobrenje i granice provajdera.\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„Arhitekta mora da napiše svaku komponentu.“\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Praktična implementacija može poboljšati arhitektonski kvalitet, ali uloga je definisana odgovornošću za integrisane odluke, a ne ličnim kodiranjem svakog sloja.\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„Dijagram arhitekture dokazuje spremnost za produkciju.“\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Spremnost zahteva implementirane kontrole i dokaze validacije kroz kvalitet, bezbednost, operacije i poslovno prihvatanje.\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"section-67\">Načini neuspeha koje AI arhitekta rešenja treba da spreči\u003C\u002Fh2>\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\">Način neuspeha\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Zašto se javlja\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Arhitektonska korekcija\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Dizajn sa modelom na prvom mestu\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Obećavajuća demonstracija modela postaje nacrt sistema\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Počnite od ishoda, ograničenja i validacije; izaberite model unutar tog okvira\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Dozvole prototipa u produkciji\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Deljeni akreditivi i širok pristup prežive PoC\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Definišite propagaciju identiteta, najmanje privilegije, opsege alata i granice odobrenja rano\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Pretraga bez autorizacije\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Kvalitet pretrage se dizajnira pre pravila pristupa podacima\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Prenesite kontekst korisnika\u002Ftenanta u pretragu i sprovedite autorizaciju na granicama pristupa podacima\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Tihe pretpostavke o provajderu\u002Fizvršavanju\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">„Lokalno“, „cloud“ i „offline“ se koriste neprecizno\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Dokumentujte odvojeno lokaciju izvršavanja, zaključivanja, podataka i kontrolne ravni\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Nema ugovora o neuspehu\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Dizajniran je srećan put, ali ne i ponašanje odbijanja\u002Frezerve\u002Fgreške\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Specifikujte ponašanje kada je pretraga prazna, model nedostupan, alat neuspešan i politika odbija\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Evaluacija posle implementacije\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Kvalitet se procenjuje ručno pred lansiranje\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Definišite merljivo prihvatanje i reprezentativne skupove evaluacije pre zamrzavanja arhitekture\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Nesledljiva promena\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Modeli, promptovi, pretraga ili dozvole se menjaju bez arhitektonske istorije\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Verzionirajte kritičnu konfiguraciju i evidentirajte značajne odluke\u002Fdokaze validacije\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Operacije se tretiraju samo kao infrastruktura\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">AI ponašanje nije vidljivo nakon implementacije\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Dizajnirajte tragove, metrike kvaliteta, bezbednosne događaje, telemetriju troškova i vraćanje zajedno\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"section-69\">Praktičan redosled odlučivanja\u003C\u002Fh2>\n\u003Csection class=\"editorjs-process my-6\">\u003Ch3 class=\"mb-3 text-lg font-semibold\">Redosled odlučivanja u arhitekturi AI rešenja\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\">Ishod\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Definišite korisnički\u002Fposlovni rezultat i eksplicitne ne-ciljeve.\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\">Dokazi i ograničenja\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Identifikujte autoritativne podatke, politike, nefunkcionalne zahteve, rizike i uslove prihvatanja.\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\">Granica sistema\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Mapirajte korisnike, identitete, aplikacije, podatke, modele\u002Fprovajdere, alate i eksterne sisteme.\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\">Opcije arhitekture\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Uporedite obrasce za pretragu, pristup modelu, orkestraciju, implementaciju, dozvole, evaluaciju i observabilnost.\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\">Odluke o kompromisima\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Izaberite značajne opcije i sačuvajte obrazloženje, alternative i posledice.\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\">Ugovori implementacije\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Pretvorite odluke u API-je, šeme, pravila dozvola, definicije implementacije i inženjerske zadatke.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv class=\"editorjs-process__step min-w-0  rounded-xl border border-gray-200 dark:border-gray-700 p-4\">\u003Cdiv class=\"text-xs font-semibold text-gray-500 dark:text-gray-400\">7\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">Validacija\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Testirajte implementirani sistem prema originalnim funkcionalnim i nefunkcionalnim zahtevima.\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\">8\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 font-semibold text-gray-900 dark:text-gray-100\">Operativne povratne informacije\u003C\u002Fdiv>\u003Cdiv class=\"mt-1 text-sm text-gray-600 dark:text-gray-300\">Koristite produkcijske dokaze, incidente, metrike kvaliteta i signale troškova\u002Fbezbednosti da pokrenete kontrolisanu promenu.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\n\u003Ch2 id=\"section-71\">Rubni slučajevi i granice uloge\u003C\u002Fh2>\n\u003Cp>Neki AI proizvodi su dominirani obukom modela, naučnim eksperimentisanjem ili specijalizovanim hardverom. U tim slučajevima, model\u002Fdata nauka i arhitektura ML sistema mogu postati mnogo dublji od mape na nivou rešenja prikazane ovde. AI arhitekta rešenja i dalje treba integracione i operativne granice, ali specijalistička arhitektura može posedovati samu platformu za obuku.\u003C\u002Fp>\n\u003Cp>Na drugom kraju spektra, jednostavna SaaS integracija možda ne opravdava posvećenog arhitektu. Senior inženjer ili tehnički vođa proizvoda može nositi istu arhitektonsku odgovornost. Koristan test nije titula, već da li se značajne odluke koje prelaze slojeve donose namerno i validiraju.\u003C\u002Fp>\n\u003Cp>Regulisani, suvereni, vazdušno izolovani, bezbednosno kritični, visoko autonomni ili multi-tenant sistemi takođe pomeraju težište. Identitet, izolacija, rezidentnost, garancija, mehanizmi ažuriranja, ljudski nadzor i mogućnost revizije mogu dominirati kvalitetom modela u arhitekturi.\u003C\u002Fp>\n\u003Ch2 id=\"section-75\">Šta bi promenilo ovaj odgovor?\u003C\u002Fh2>\n\u003Cp>Tačna granica odgovornosti se menja kada arhitektura pređe sa jedne aplikacije na platformu koja se može ponovo koristiti ili na ciljnu arhitekturu na nivou preduzeća. Zato \u003Cstrong>AI Platform Architect\u003C\u002Fstrong> i \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> zaslužuju odvojen kanonski tretman umesto da budu spojeni u ovu ulogu.\u003C\u002Fp>\n\u003Cp>Promene tehnologije su takođe važne. Nove mogućnosti modela, protokoli, lokalna izvršna okruženja i upravljane usluge mogu ukloniti deo implementacionog rada dok istovremeno stvaraju nove granice poverenja ili operativne granice. Stabilna odgovornost je razumeti te promene kao promene sistema—a ne tretirati novi okvir kao zamenu za arhitekturu.\u003C\u002Fp>\n\u003Ch2 id=\"section-78\">AI Solution Architect kontrolna lista\u003C\u002Fh2>\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\">Provera\u003C\u002Fth>\u003Cth class=\"border border-gray-300 px-4 py-2 text-left font-semibold\">Pitanje\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Ishod\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su korisnički\u002Fposlovni rezultat i granica ne-ciljeva eksplicitni?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Zahtevi\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su funkcionalni zahtevi, nefunkcionalni zahtevi, ograničenja i kriterijumi prihvatanja sledljivi?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Podaci\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su autoritativni izvori, poreklo, svežina, zadržavanje i pravila pristupa definisani?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Pretraga\u002Fkontekst\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li autorizacija doseže do pretrage i konstrukcije konteksta?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Model\u002Fprovajder\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li je izbor modela\u002Fprovajdera vezan za mogućnosti i ograničenja, a ne za preferencije?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Alati\u002Fagenti\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su granice akcija, dozvole, odobrenja i ponašanje pri neuspehu eksplicitni?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Identitet\u002Fbezbednost\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su ljudski\u002Fmašinski identiteti, tajne i granice poverenja definisani?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Izvršno okruženje\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su lokacije izvršnog okruženja, inferencije, podataka i kontrolne ravni razlikovane?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Evaluacija\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li postoje merljivi dokazi za kvalitet, bezbednost i prihvatanje?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Opservabilnost\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Mogu li se ponašanje u produkciji, neuspesi, troškovi i bezbednosni događaji istražiti?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Promena\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li su značajne arhitektonske odluke i zamene sledljive?\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Operacije\u003C\u002Ftd>\u003Ctd class=\"border border-gray-300 px-4 py-2\">Da li je vlasništvo nad implementacijom, vraćanjem, incidentima i životnim ciklusom jasno?\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Ch2 id=\"section-80\">Zaključak\u003C\u002Fh2>\n\u003Cp>AI Solution Architect je osoba ili arhitektonska funkcija koja pretvara AI priliku u koherentan tehnički sistem. Ključna veština nije poznavanje najvećeg broja imena modela; to je povezivanje potreba proizvoda, zahteva, podataka, arhitekture aplikacije, AI mogućnosti, bezbednosti, izvršnog okruženja, isporuke i validacije bez gubljenja granica između njih.\u003C\u002Fp>\n\u003Cp>Jaka arhitektura AI rešenja se stoga može sažeti kao: \u003Cstrong>definiši cilj → uspostavi zahteve i ograničenja → projektuj granice sistema → učini značajne kompromise eksplicitnim → implementiraj kroz jasne ugovore → validiraj na osnovu dokaza → upravljaj i razvijaj namerno.\u003C\u002Fstrong> Model je važan. Rešenje je proizvod.\u003C\u002Fp>\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\">AI Solution Architect — Često postavljana pitanja\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\">Šta je AI Solution Architect?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">AI Solution Architect prevodi poslovnu ili proizvodnu potrebu u arhitekturu konkretnog AI rešenja, definišući kako logika aplikacije, podaci\u002Fpretraga, modeli, alati, identitet, bezbednost, izvršno okruženje, evaluacija i operacije funkcionišu zajedno.\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\">Da li je AI Solution Architect isto što i AI inženjer?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Ne. Uloge se mogu preklapati, posebno u malim timovima, ali AI inženjer je prvenstveno implementaciona uloga dok arhitekta rešenja poseduje ili koordinira arhitektonske odluke i kompromise koji prelaze slojeve za kompletan radni opseg.\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\">Da li AI Solution Architect mora da kodira?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Ne po definiciji, ali praktično znanje implementacije je veoma vredno jer AI arhitektura prelazi preko API-ja, podataka, pretrage, bezbednosti, izvršnih okruženja i operativnog ponašanja. Uloga je definisana arhitektonskom odgovornošću, a ne ličnim pisanjem svake komponente.\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\">Da li je izbor LLM-a glavni posao?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Ne. Izbor modela je jedna odluka. Produkciona arhitektura takođe zahteva granice podataka i pretrage, dozvole, alate, izbore provajdera\u002Fizvršnog okruženja, opservabilnost, evaluaciju, pouzdanost, troškove i dizajn životnog ciklusa.\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\">Koja je razlika između AI Solution Architect i AI Platform Architect?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">AI Solution Architect se fokusira na jedno konkretno rešenje ili radni opseg. AI Platform Architect se fokusira na AI mogućnosti i zaštitne mere koje se mogu ponovo koristiti i podržavaju više rešenja.\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\">Koja je razlika između AI Solution Architect i Enterprise AI Architect?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Arhitekta rešenja radi na nivou aplikacije\u002Fradnog opsega. Enterprise AI arhitektura radi preko organizacionog portfolija, ciljne arhitekture, upravljanja, deljenih mogućnosti, principa integracije i strateških ograničenja.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq7\" 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\">Gde se uklapaju RAG i agenti?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Oni su arhitektonski obrasci ili podsistemi unutar rešenja kada to zahtevi opravdavaju. RAG se bavi kontekstom zasnovanim na pretrazi; agenti dodaju planiranje\u002Fizvršavanje alata i stoga dodatne brige o identitetu, dozvolama, orkestraciji i operacijama.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003Cdiv id=\"faq8\" 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\">Šta dokazuje da arhitektura funkcioniše?\u003C\u002Fh4>\u003Cdiv class=\"mt-2 text-gray-600 dark:text-gray-300\">Implementacija plus dokazi validacije: funkcionalni testovi, rezultati evaluacije, bezbednosni testovi\u002F testovi autorizacije, merenja performansi i pouzdanosti, opservabilnost, operativna proba i prihvatanje u odnosu na prvobitne zahteve.\u003C\u002Fdiv>\u003C\u002Fdiv>\u003C\u002Fsection>\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\">Ključni pojmovi\u003C\u002Fh3>\u003Cdl>\u003Cdiv id=\"ai-solution-architect\" 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\">AI Solution Architect\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Arhitektonska odgovornost za jedno konkretno AI rešenje ili radni opseg, integrišući zahteve proizvoda sa aplikacijom, podacima, modelom, alatima, bezbednošću, izvršnim okruženjem i operativnim dizajnom.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"system-boundary\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">Granica sistema\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Eksplicitno razdvajanje između onoga što pripada rešenju i korisnika, sistema, provajdera, izvora podataka i okruženja sa kojima interaguje.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"trust-boundary\" class=\"border-t border-gray-200 dark:border-gray-700 py-3 first:border-t-0\">\u003Cdt class=\"font-semibold text-gray-900 dark:text-gray-100\">Granica poverenja\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Tačka u kojoj podaci, identiteti ili kontrola prelaze između komponenti sa različitim pretpostavkama poverenja i stoga zahtevaju eksplicitne bezbednosne kontrole.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"grounding\" 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\">Uzemljenje\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Snabdevanje AI modela relevantnim spoljnim informacijama ili dokazima tako da njegov odgovor može biti zasnovan na izvorima izvan parametara modela.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"provider-abstraction\" 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\">Apstrakcija provajdera\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Granica aplikacije koja razdvaja delove rešenja od interfejsa jednog modela\u002Fprovajdera. Korisna kada je opravdana potrebama rutiranja, prenosivosti ili politike, ali nije bez kompromisa.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"evaluation\" 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\">Evaluacija\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Strukturirano merenje ponašanja AI radnog opsega u odnosu na definisane kriterijume prihvatanja, uključujući kvalitet zadatka i relevantne bezbednosne, sigurnosne, performansne i operativne osobine.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"ai-platform-architect\" 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\">AI Platform Architect\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Arhitektonska uloga fokusirana na AI mogućnosti platforme koje se mogu ponovo koristiti i koristi ih više rešenja, a ne na arhitekturu jednog radnog opsega.\u003C\u002Fdd>\u003C\u002Fdiv>\u003Cdiv id=\"enterprise-ai-architecture\" 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\">Enterprise AI Architecture\u003C\u002Fdt>\u003Cdd class=\"mt-1 text-gray-600 dark:text-gray-300\">Arhitektura na nivou organizacije koja koordinira AI mogućnosti, platforme, upravljanje, integraciju i strateška ograničenja preko portfolija.\u003C\u002Fdd>\u003C\u002Fdiv>\u003C\u002Fdl>\u003C\u002Fsection>\n\u003Ch2 id=\"section-85\">Povezano kanonsko znanje\u003C\u002Fh2>\n\u003Cp>Ovaj članak pripada klasteru AI Architecture Foundations. Njegovi direktni temelji su \u003Cstrong>Generative AI Explained: Models, Retrieval, Tools and Applications Are Not the Same Thing\u003C\u002Fstrong> i \u003Cstrong>ADR vs NFR: Architecture Decisions and System Quality Are Not the Same Thing\u003C\u002Fstrong>. Susedni kanonski čvorovi uključuju \u003Cstrong>Agentic AI Explained\u003C\u002Fstrong>, \u003Cstrong>Source of Truth in AI Systems\u003C\u002Fstrong>, \u003Cstrong>Vector Databases, Embeddings and Reranking\u003C\u002Fstrong>, \u003Cstrong>What Is Context Engineering?\u003C\u002Fstrong>, \u003Cstrong>RBAC vs Tenant Isolation\u003C\u002Fstrong>, \u003Cstrong>AI Platform Architect\u003C\u002Fstrong>, \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> i \u003Cstrong>AI Governance\u003C\u002Fstrong>. URL-ovi namerno nisu izmišljeni tamo gde ti čvorovi još nisu objavljeni.\u003C\u002Fp>\n\u003Ca href=\"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works\" 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\">Šta je RAG? Najjednostavnije objašnjenje kako funkcioniše\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Postojeće stajic.de kanonsko objašnjenje generacije sa pretragom, korisno za deo pretrage\u002Fuzemljenja u arhitekturi AI rešenja.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ch2 id=\"section-88\">Primarni izvori i trenutne arhitektonske smernice\u003C\u002Fh2>\n\u003Cp>Spoljni izvori ispod podržavaju opšte arhitektonske tvrdnje; odeljci SenseFlow i Aaasaasa AI Client su eksplicitno originalni dokazi projekta\u002Fimplementacije. Reference trenutnog stanja su proverene 8. oktobra 2026. NIST napominje da se AI RMF 1.0 revidira, tako da reference upravljanja osetljive na verziju treba ponovo proveriti kada naslednik bude objavljen.\u003C\u002Fp>\n\u003Ca href=\"https:\u002F\u002Fwww.iso.org\u002Fstandard\u002F74393.html\" 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\">ISO\u002FIEC\u002FIEEE 42010:2022 — Opis arhitekture\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Trenutni međunarodni standard za strukturu i izražavanje opisa arhitekture. Razlikuje arhitekturu od njenog opisa i ne propisuje jednu metodu arhitekture, alat ili format snimanja.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fwww.nist.gov\u002Fitl\u002Fai-risk-management-framework\" 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\">NIST okvir za upravljanje rizicima veštačke inteligencije\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">NIST stranica sa resursima o AI RMF. Od oktobra 2026. navodi da se AI RMF 1.0 revidira i povezuje Generativni AI profil i povezane resurse.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fairc.nist.gov\u002Fairmf-resources\u002Fairmf\u002F5-sec-core\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\">NIST AI RMF jezgro — Upravljaj, Mapiraj, Meri, Upravljaj\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Zvanična NIST AIRC prezentacija jezgra AI RMF 1.0, uključujući četiri funkcije i okvir za upravljanje rizicima orijentisan na životni ciklus.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fwww.nist.gov\u002Fpublications\u002Fartificial-intelligence-risk-management-framework-generative-artificial-intelligence\" 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\">NIST AI 600-1 — Generativni AI profil\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Međusektorski generativni AI profil za AI RMF 1.0, objavljen 26. jula 2024. i ažuriran od strane NIST-a 2026. godine.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fget-started\" 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\">Microsoft Azure Well-Architected — AI radna opterećenja\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Aktuelne smernice za arhitekturu na nivou radnog opterećenja koje pokrivaju dizajn AI aplikacija, platformu aplikacija, podatke za obuku, podatke za utemeljenje, platformu podataka i pitanja spremnosti za produkciju.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fapplication-design\" 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\">Microsoft — Dizajn aplikacija za AI radna opterećenja\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Smernice o apstrakciji modela\u002Falata, granicama pristupa podacima, propagaciji identiteta, autorizaciji i razdvajanju slojeva klijenta, inteligencije, znanja i alata.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fdesign-principles\" 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\">Microsoft — Principi dizajna za AI radna opterećenja\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Aktuelni principi dizajna AI radnih opterećenja u pogledu pouzdanosti, bezbednosti, troškova, operativne izvrsnosti i performansi, uključujući odgovornosti za identitet i zaštitu podataka.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fmlops-genaiops\" 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\">Microsoft — MLOps i GenAIOps za AI radna opterećenja\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Smernice za životni ciklus u produkciji koje pokrivaju nadzor, kapije kvaliteta, ponašanje modela\u002Fupita, bezbednost i operativno merenje.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fdocs.aws.amazon.com\u002Fwellarchitected\u002Flatest\u002Fgenerative-ai-lens\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\">AWS Well-Architected sočivo za generativnu veštačku inteligenciju\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">AWS smernice za arhitekturu generativnih AI radnih opterećenja u pogledu operativne izvrsnosti, bezbednosti, pouzdanosti, efikasnosti performansi, optimizacije troškova i održivosti.\u003C\u002Fp>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fdocs.aws.amazon.com\u002Fwellarchitected\u002Flatest\u002Fagentic-ai-lens\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\">AWS Well-Architected sočivo za agentnu veštačku inteligenciju\u003C\u002Fstrong>\u003Cp class=\"text-sm text-gray-600 dark:text-gray-400\">Objavljeno 2026. godine, pokriva arhitektonska pitanja specifična za agente, uključujući identitete, alate, orkestraciju, ljudski nadzor, pouzdanost, praćenje i troškove petlje zaključivanja.\u003C\u002Fp>\u003C\u002Fa>",{"time":212,"blocks":213,"version":1068},1791476824021,[214,219,226,232,237,244,248,252,256,300,304,308,312,340,344,348,352,356,360,408,412,416,420,424,428,432,436,440,444,448,452,456,460,464,468,472,476,480,484,488,492,496,500,531,535,539,583,587,591,639,643,647,653,657,661,665,669,673,677,681,685,689,693,697,701,705,733,737,777,781,810,814,818,822,826,830,834,838,842,881,885,889,893,930,962,966,970,980,984,988,996,1004,1012,1020,1028,1036,1044,1052,1060],{"id":215,"data":216,"type":218},"intro",{"text":217},"\u003Cstrong>AI Solution Architect\u003C\u002Fstrong> prevodi poslovnu ili proizvodnu potrebu u arhitekturu konkretnog AI rešenja. Uloga definiše granice sistema i značajne izbore u okviru aplikacione logike, autoritativnih podataka, pretrage i konteksta, modela i provajdera, alata ili agenata, identiteta i dozvola, bezbednosti, izvršavanja i implementacije, nadzora, evaluacije, troškova i operativnog ponašanja. To nije samo izbor modela ili inženjering upita: arhitektonska odgovornost je da celo rešenje bude implementabilno, upravljivo, testabilno i operabilno.","paragraph",{"id":220,"data":221,"type":225},"direct",{"body":222,"title":223,"variant":224},"\u003Cstrong>AI Solution Architect projektuje kompletno AI rešenje, ne samo AI model.\u003C\u002Fstrong> Uloga povezuje zahteve i nefunkcionalne zahteve sa arhitektonskim odlukama, komponuje neophodne slojeve aplikacije\u002Fpodataka\u002Fmodela\u002Falata\u002Fizvršavanja, čini granice poverenja i otkaza eksplicitnim i definiše kako će implementirani sistem biti validiran i operisan.","Direktan odgovor","info","callout",{"id":227,"data":228,"type":225},"role-note",{"body":229,"title":230,"variant":231},"\u003Cstrong>AI Solution Architect je praktična oznaka uloge, a ne univerzalno standardizovan naziv radnog mesta.\u003C\u002Fstrong> ISO\u002FIEC\u002FIEEE 42010:2022 standardizuje koncepte za opise arhitekture; ne definiše ovu radnu ulogu. Organizacije mogu raspodeliti odgovornosti na više ljudi. U ovom članku, termin označava arhitektonsku odgovornost za jedno konkretno AI rešenje ili radno opterećenje.","Napomena o terminologiji","note",{"id":233,"data":234,"type":225},"version-note",{"body":235,"title":236,"variant":231},"Ovde navedeni arhitektonski principi su namerno neutralni prema provajderu, dok se aktuelne smernice provajdera koriste kao dokaz implementacije. NIST AI RMF 1.0 je trenutno u reviziji; NIST AI 600-1 ostaje objavljeni Generative AI Profile. Smernice Microsoft-a i AWS-a citirane u nastavku odražavaju aktuelne proizvodne brige kao što su identitet, granice podataka, apstrakcija modela, bezbednost, nadzor, evaluacija, pouzdanost i troškovi.","Napomena o aktuelnim izvorima — 8. oktobar 2026.",{"id":238,"data":239,"type":243},"toc",{"title":240,"maxLevel":241,"minLevel":242},"Sadržaj",3,2,"tableOfContents",{"id":245,"data":246,"type":42},"h-meaning",{"text":247,"level":242},"Šta AI Solution Architect zapravo projektuje?",{"id":249,"data":250,"type":218},"p-meaning-1",{"text":251},"Predmet rada je \u003Cstrong>rešenje\u003C\u002Fstrong>: kompletan socio-tehnički sistem koji potrebu pretvara u korisno, kontrolisano ponašanje. Model može biti centralni za taj sistem, ali je i dalje samo jedna zavisnost. Isti model može učestvovati u bezbednom internom pretraživačkom asistentu, nebezbednom agentu sa prekomernim privilegijama, korisničkoj funkciji sa malim kašnjenjem ili skupom prototipu koji se ne može ekonomski operisati. Arhitektura određuje te razlike.",{"id":253,"data":254,"type":218},"p-meaning-2",{"text":255},"Korisna granica je stoga: \u003Cstrong>poslovni ishod → zahtevi → odgovornosti sistema → arhitektonske odluke → implementacija → validacija → operacija\u003C\u002Fstrong>. AI Solution Architect radi duž ovog lanca, sarađujući sa produktom, inženjeringom, podacima, bezbednošću, infrastrukturom, upravljanjem i domenskim specijalistima.",{"id":257,"data":258,"type":299},"solution-vs-model",{"rows":259,"title":290,"layout":291,"columns":292},[260,266,272,278,284],{"id":261,"label":262,"values":263},"m1","Sposobnost",{"model":264,"solution":265},"Which model can generate or reason well enough?","Which combination of model, data, application logic, retrieval, tools and controls produces the required behavior?",{"id":267,"label":268,"values":269},"m2","Podaci",{"model":270,"solution":271},"What context can fit in the prompt?","What is authoritative, who may access it, how is it retrieved, versioned, filtered and cited?",{"id":273,"label":274,"values":275},"m3","Bezbednost",{"model":276,"solution":277},"Does the provider offer security features?","What are the trust boundaries, identities, permissions, secrets, data flows and failure containment mechanisms?",{"id":279,"label":280,"values":281},"m4","Operacije",{"model":282,"solution":283},"What is the token latency?","How is the complete workload deployed, observed, evaluated, recovered, versioned and cost-controlled?",{"id":285,"label":286,"values":287},"m5","Promena",{"model":288,"solution":289},"Can we switch models?","Which dependencies are abstracted, what changes require an ADR, and how do we validate that a replacement still meets requirements?","Rešenje je šire od modela","table",[293,296],{"id":294,"label":295},"model","Pitanje usmereno na model",{"id":297,"label":298},"solution","Pitanje arhitekture rešenja","comparison",{"id":301,"data":302,"type":42},"h-simple",{"text":303,"level":242},"Najjednostavniji primer",{"id":305,"data":306,"type":218},"p-simple-1",{"text":307},"Zamislite da kompanija želi internog asistenta koji odgovara na pitanja tehničara na osnovu priručnika za održavanje i operativnih procedura. Vidljiva funkcija zvuči jednostavno: ukucajte pitanje i primite odgovor sa izvorima.",{"id":309,"data":310,"type":218},"p-simple-2",{"text":311},"Arhitektonsko pitanje je mnogo veće. Koji dokumenti su autoritativni? Kako se korisnici autentifikuju? Da li pretraga mora poštovati dozvole odeljenja ili lokacije? Da li odgovor sme koristiti samo pronađene dokaze? Koji model je prihvatljiv za klasifikaciju podataka? Može li cloud provajder primiti sadržaj? Šta se dešava kada pretraga ne pronađe ništa? Kako se proizvode citati? Kako se evaluira kvalitet odgovora? Koje kašnjenje i trošak su prihvatljivi? Ko može videti logove i šta se u njima sme čuvati?",{"id":313,"data":314,"type":339},"simple-flow",{"steps":315,"title":337,"orientation":338},[316,319,322,325,328,331,334],{"label":317,"description":318},"1. Definišite ishod","Razjasnite korisnika, poslovnu vrednost, granicu zadatka i šta znači uspešan odgovor ili akcija.",{"label":320,"description":321},"2. Prikupite zahteve","Učinite eksplicitnim funkcionalne zahteve, nefunkcionalne zahteve, ograničenja, pravila o podacima, toleranciju rizika i kriterijume prihvatanja.",{"label":323,"description":324},"3. Uspostavite granice","Identifikujte korisnike, identitete, aplikacije, autoritativne podatke, zavisnosti od modela\u002Fprovajdera, alate, eksterne sisteme i zone poverenja.",{"label":326,"description":327},"4. Projektujte arhitekturu","Izaberite obrasce za podatke\u002Fpretragu, model, orkestraciju, alate, dozvole, izvršavanje, implementaciju, rezervne opcije i nadzor.",{"label":329,"description":330},"5. Zabeležite značajne odluke","Sačuvajte arhitektonske izbore, alternative, kompromise i posledice kako bi kasnije promene ostale razumljive.",{"label":332,"description":333},"6. Implementirajte i integrišite","Pretvorite arhitekturu u aplikacioni kod, API-je, politike, infrastrukturu, radne tokove i operativne kontrole.",{"label":335,"description":336},"7. Validirajte i operišite","Testirajte kvalitet, bezbednost, pouzdanost, troškove i korisničke ishode; nadzirite stvarno radno opterećenje i vraćajte dokaze u odluke.","Od potrebe do operabilnog AI rešenja","auto","processFlow",{"id":341,"data":342,"type":42},"h-where-simple-stops",{"text":343,"level":242},"Gde se jednostavan primer zaustavlja",{"id":345,"data":346,"type":218},"p-stop-1",{"text":347},"Proof of concept često može preskočiti arhitekturu koju produkcija ne može. Programer može hardkodirati jednog provajdera, koristiti deljeni API ključ, smestiti sve dokumente u jedan indeks, pokretati pretragu bez filtriranja po korisničkom kontekstu, logovati upite doslovno i procenjivati kvalitet ručno. To može demonstrirati izvodljivost, ali ne uspostavlja produkcijsku arhitekturu.",{"id":349,"data":350,"type":218},"p-stop-2",{"text":351},"Produkcija uvodi ograničenja koja interaguju: izolaciju zakupaca ili korisnika, privatnost, rezidentnost podataka, propusnost, kašnjenje, troškove, kvote provajdera, rezervno ponašanje, revizibilnost, promene verzija modela, kvalitet pretrage, dozvole alata, odgovor na incidente i životni ciklus implementacije. Posao arhitekte nije da maksimizuje svaki kvalitet odjednom; već da kompromise učini eksplicitnim i projektuje rešenje koje zadovoljava stvarni skup prioriteta.",{"id":353,"data":354,"type":42},"h-responsibility-map",{"text":355,"level":242},"Mapa arhitektonskih odgovornosti",{"id":357,"data":358,"type":218},"p-resp-intro",{"text":359},"Tačna podela varira od organizacije do organizacije, ali sledeća mapa obuhvata ponavljajuće odgovornosti arhitekture AI rešenja. Arhitekta možda neće lično implementirati svaki sloj; odgovornost je da slojevi funkcionišu koherentno i da kritične odluke ostanu sledljive.",{"id":361,"data":362,"type":291},"responsibility-table",{"content":363,"stretched":43,"withHeadings":14},[364,368,372,376,380,384,388,392,396,400,404],[365,366,367],"Arhitektonska oblast","Pitanja koja AI Solution Architect mora da reši","Tipični izlazi",[369,370,371],"Ishod i obim","Ko je korisnik? Koji zadatak je u obimu? Šta sistem ne sme da radi? Šta predstavlja uspeh?","Kontekst rešenja, granica sposobnosti, kriterijumi prihvatanja",[373,374,375],"Zahtevi i NFR-ovi","Koja ograničenja kvaliteta, bezbednosti, dostupnosti, latencije, troškova, rezidentnosti i usklađenosti se primenjuju?","Mapa zahteva, NFR-ovi, ograničenja, kriterijumi validacije",[377,378,379],"Aplikacija i orkestracija","Gde se završava deterministička aplikaciona logika, a počinje AI ponašanje? Kako se koordinišu tokovi rada?","Model komponenti, API-ji, granice orkestracije, putevi otkaza",[381,382,383],"Autoritativni podaci i pretraga","Šta je izvor istine? Kako se podaci unose, autorizuju, pronalaze, filtriraju, rangiraju i citiraju?","Tokovi podataka, arhitektura pretrage, metapodaci i pravila autorizacije",[385,386,387],"Sloj modela i provajdera","Koje sposobnosti su potrebne? Koja ograničenja provajdera\u002Fruntime-a su važna? Šta treba apstrahovati?","Odluka o modelu\u002Fprovajderu, politika rutiranja\u002Ffallback-a, granica apstrakcije",[389,390,391],"Alati i agenti","Koje radnje sistem može da preduzme? Koje radnje zahtevaju odobrenje? Kako se sprovode identiteti i dozvole alata?","Ugovori alata, granice agenata, pravila odobravanja i najmanjih privilegija",[393,394,395],"Identitet i bezbednost","Koji ljudski i mašinski identiteti postoje? Gde se čuvaju tajne? Koje granice poverenja se prelaze?","Model pretnji\u002Fgranica poverenja, propagacija identiteta, dizajn tajni i autorizacije",[397,398,399],"Runtime i implementacija","Gde se komponente izvršavaju? Šta je lokalno, cloud, edge ili hibridno? Koje pretpostavke o mreži i dostupnosti postoje?","Prikaz implementacije, runtime topologija, odluke o okruženju i povezivanju",[401,402,403],"Evaluacija i observabilnost","Kako se kvalitet meri pre i posle izdanja? Koji tragovi, metrike, logovi i dokazi su potrebni?","Plan evaluacije, telemetrija, revizorski trag, kapije izdanja",[405,406,407],"Operacije i promene","Kako se verzije modela\u002Fpromptova\u002Fkonfiguracije\u002Fpodataka menjaju, vraćaju i podržavaju?","Operativni model, kontrole životnog ciklusa, ADR-ovi, runbook-ovi, pravila promena",{"id":409,"data":410,"type":42},"h-requirements",{"text":411,"level":241},"1. Pretvorite potrebu proizvoda u arhitektonske zahteve",{"id":413,"data":414,"type":218},"p-requirements-1",{"text":415},"AI arhitektura počinje pre izbora modela. Arhitekta prvo utvrđuje šta se od rešenja očekuje da postigne i pod kojim ograničenjima. To uključuje funkcionalno ponašanje, ali i NFR-ove i politike koje sužavaju prostor dizajna: bezbednost, pouzdanost, latenciju, privatnost, rezidentnost, održivost, troškove i operativnu podršku.",{"id":417,"data":418,"type":218},"p-requirements-2",{"text":419},"Ovde je važna razlika iz A02: zahtev kao što je „neovlašćeni korisnici ne smeju da pronađu ograničene dokumente“ nije arhitektonska odluka. To je pokretač. Odluke o propagaciji identiteta, particionisanju indeksa, filtriranju metapodataka, granicama API-ja i sprovođenju autorizacije su arhitektonski odgovori koji kasnije moraju biti validirani.",{"id":421,"data":422,"type":42},"h-data",{"text":423,"level":241},"2. Dizajnirajte autoritativne podatke, pretragu i kontekst",{"id":425,"data":426,"type":218},"p-data-1",{"text":427},"AI sistemi često padaju na granici između ponašanja modela i istine preduzeća. Arhitekta mora da definiše koji izvori su autoritativni, šta znače svežina i poreklo, kako kontrola pristupa stiže do pretrage i kako pronađeni dokazi postaju kontekst modela. Vektorska baza podataka, model za ugrađivanje ili RAG biblioteka nisu sami po sebi arhitektura.",{"id":429,"data":430,"type":218},"p-data-2",{"text":431},"Microsoft-ove trenutne smernice za AI radna opterećenja eksplicitno prave istu razliku: aplikacioni kod ne treba da zaobilazi granice pristupa podacima; kontekst korisnika ili zakupca treba da se propagira u pretragu i filtriranje; podaci za utemeljenje moraju biti dizajnirani za pretraživost, a istovremeno da zadovolje bezbednosne i regulatorne zahteve.",{"id":433,"data":434,"type":42},"h-model",{"text":435,"level":241},"3. Tretirajte modele i provajdere kao zavisnosti, ne kao ceo sistem",{"id":437,"data":438,"type":218},"p-model-1",{"text":439},"Izbor modela je važan, ali treba da bude vođen potrebnim sposobnostima i ograničenjima. Arhitekta razmatra kvalitet rezonovanja ili generisanja, modalitet, ograničenja konteksta, latenciju, rukovanje podacima, lokaciju implementacije, dostupnost provajdera, troškove, observabilnost i rizik zamene.",{"id":441,"data":442,"type":218},"p-model-2",{"text":443},"Apstrakcija provajdera nije automatski „bolja arhitektura“. Dodaje inženjerske troškove i može da sakrije sposobnosti specifične za provajdera. Opravdana je kada su prenosivost, fallback, razdvajanje politika ili rutiranje ka više provajdera eksplicitni zahtevi. U suprotnom, direktna integracija može biti bolja odluka. Poenta je da kompromis bude nameran.",{"id":445,"data":446,"type":42},"h-tools",{"text":447,"level":241},"4. Arhitektirajte alate, radnje i granice agenata",{"id":449,"data":450,"type":218},"p-tools-1",{"text":451},"Kada AI sistem može da poziva alate, menja podatke, šalje poruke, pokreće kod ili upravlja poslovnim sistemima, arhitektonski rizik se menja. Pristup alatima zahteva sopstveni model identiteta i autorizacije. Sposobnost modela da zatraži radnju nije isto što i dozvola da je izvrši.",{"id":453,"data":454,"type":218},"p-tools-2",{"text":455},"Za agentna radna opterećenja, trenutne AWS smernice naglašavaju dodatne dimenzije kao što su identiteti agenata, pristup alatima, orkestracija, ljudski nadzor, praćenje, rukovanje otkazima i troškovi iterativnih petlji rezonovanja. To su pitanja rešenja čak i kada okvir sakriva neke od mehanizama implementacije.",{"id":457,"data":458,"type":42},"h-security",{"text":459,"level":241},"5. Učinite granice poverenja i dozvole eksplicitnim",{"id":461,"data":462,"type":218},"p-security-1",{"text":463},"Produkcijsko AI rešenje ima više granica poverenja: pregledač ili klijent, aplikacioni backend, AI orkestracija, servisi za pretragu\u002Fpodatke, provajderi modela, API-ji alata, lokalni runtime-ovi i eksterni sistemi. Svaka granica treba da odgovori: ko poziva, u čije ime, sa kojim akreditivom, za koji resurs, sa kojim revizorskim tragom i sa kojim ograničavanjem otkaza?",{"id":465,"data":466,"type":218},"p-security-2",{"text":467},"Bezbednost se ne može odložiti na „guardrail“ oko modela. Microsoft-ove smernice za AI radna opterećenja eksplicitno postavljaju bezbednost kroz sve arhitektonske slojeve i zahtevaju upravljanje identitetom\u002Fpristupom, zaštitu podataka, kontrolu sadržaja i bezbednost životnog ciklusa. NIST takođe tretira upravljanje i upravljanje rizikom kao kontinuirano kroz AI životni ciklus.",{"id":469,"data":470,"type":42},"h-runtime",{"text":471,"level":241},"6. Odlučite gde sistem zaista radi",{"id":473,"data":474,"type":218},"p-runtime-1",{"text":475},"„Lokalni AI“, „cloud AI“ i „hibridni AI“ su arhitektonske izjave samo kada su putevi izvršavanja i podataka precizni. Lokalni desktop proces i dalje može da poziva cloud model. Aplikacija hostovana u cloudu može da pronalazi podatke iz on-premises izvora podataka. Air-gapped rešenje ima potpuno drugačija ograničenja ažuriranja, distribucije modela i observabilnosti.",{"id":477,"data":478,"type":218},"p-runtime-2",{"text":479},"Arhitekta stoga razdvaja \u003Cstrong>lokaciju izvršavanja\u003C\u002Fstrong>, \u003Cstrong>lokaciju zaključivanja\u003C\u002Fstrong>, \u003Cstrong>lokaciju podataka\u003C\u002Fstrong> i \u003Cstrong>kontrolnu ravan\u003C\u002Fstrong>. Njihovo poistovećivanje stvara lažnu sigurnost i pretpostavke o raspoređivanju.",{"id":481,"data":482,"type":42},"h-eval",{"text":483,"level":241},"7. Definišite evaluaciju, observabilnost i operativno prihvatanje",{"id":485,"data":486,"type":218},"p-eval-1",{"text":487},"AI ponašanje je delimično nedeterminističko, pa definicija izdanja ne može da se osloni samo na konvencionalne unit testove. Arhitekturi su potrebni merljivi kriterijumi prihvatanja: uspeh zadatka, utemeljenost ili ispravnost citiranja gde je relevantno, ponašanje odbijanja, bezbednost alata, latencija, trošak, pouzdanost i bezbednosni testovi. Tačne metrike zavise od slučaja upotrebe.",{"id":489,"data":490,"type":218},"p-eval-2",{"text":491},"Microsoft-ove trenutne Well-Architected AI smernice tretiraju monitoring kao kontinuiran i primenjuju ga na ponašanje modela, promptove\u002Fkompletiranja, anomalije, bezbednost i kapije kvaliteta u produkciji. AWS na sličan način tretira observabilnost, upravljanje životnim ciklusom i sledljivost modela\u002Fpromptova kao pitanja operativne arhitekture.",{"id":493,"data":494,"type":42},"h-artifacts",{"text":495,"level":242},"Šta bi ova uloga trebalo da proizvede?",{"id":497,"data":498,"type":218},"p-artifacts-1",{"text":499},"Arhitektura nije slajd prezentacija. Korisni izlazi su artefakti koji omogućavaju inženjeringu, bezbednosti, proizvodu i operacijama da donose konzistentne odluke i kasnije razumeju zašto sistem postoji u svom trenutnom obliku.",{"id":501,"data":502,"type":291},"artifacts-table",{"content":503,"stretched":43,"withHeadings":14},[504,507,510,513,516,519,522,525,528],[505,506],"Artefakt","Svrha",[508,509],"Kontekst i granice rešenja","Prikazuje korisnike, eksterne sisteme, glavne odgovornosti i šta je van obima",[511,512],"Mapa zahteva\u002FNFR","Povezuje potrebe proizvoda i ograničenja sa arhitektonskim radom i validacijom",[514,515],"Prikazi komponenti i tokova podataka","Prikazuje interakcije aplikacije, podataka\u002Fpretrage, modela, alata, identiteta i izvršnog okruženja",[517,518],"Model poverenja i dozvola","Eksplicitno prikazuje identitete, tajne, autorizaciju, osetljive podatke i radnje visokog rizika",[520,521],"Zapisi o arhitektonskim odlukama","Čuva značajne izbore, alternative, kompromise, status i posledice",[523,524],"Plan evaluacije i prihvatanja","Definiše dokaze potrebne da bi se tvrdilo da rešenje ispunjava očekivanja kvaliteta i bezbednosti",[526,527],"Prikaz raspoređivanja i operacija","Definiše okruženja, lokacije izvršavanja, observabilnost, vraćanje unazad, incidente i odgovornosti životnog ciklusa",[529,530],"Veze sledljivosti","Povezuje zahteve, odluke, implementacioni rad, testove i operativne dokaze",{"id":532,"data":533,"type":42},"h-tradeoffs",{"text":534,"level":242},"Rad se uglavnom sastoji od kompromisa, a ne od izbora „najbolje prakse“",{"id":536,"data":537,"type":218},"p-tradeoffs-1",{"text":538},"Arhitektura postoji jer su poželjni kvaliteti u sukobu. Model sa nižim troškovima može smanjiti kvalitet. Sposobniji model može povećati latenciju ili ograničenja upravljanja podacima. Agresivno keširanje može poboljšati troškove i brzinu, ali otežati svežinu. Autonomniji agenti mogu smanjiti ljudski napor, ali povećati domet štete i zahteve za revizijom.",{"id":540,"data":541,"type":291},"tradeoff-table",{"content":542,"stretched":43,"withHeadings":14},[543,548,553,558,563,568,573,578],[544,545,546,547],"Odluka","Potencijalna korist","Potencijalni trošak \u002F rizik","Arhitektonsko pitanje",[549,550,551,552],"Upravljani cloud model","Brzo usvajanje, jake upravljane mogućnosti","Spoljna zavisnost, ograničenja podataka i troškova","Da li radno opterećenje dozvoljava provajdera\u002Fputanju podataka i ispunjava potrebe otpornosti?",[554,555,556,557],"Lokalno\u002Fsamostalno hostovano zaključivanje","Kontrola, offline\u002Fprivatne opcije","Hardver, operacije, teret životnog ciklusa modela","Da li je korist od kontrole vredna operativne odgovornosti?",[559,560,561,562],"Integracija sa jednim provajderom","Jednostavnija implementacija, pune mogućnosti provajdera","Veća koncentracija prebacivanja\u002Fotkaza","Da li su prenosivost ili rezervna opcija zaista potrebni?",[564,565,566,567],"Apstrakcija provajdera","Prenosivost, rutiranje i razdvajanje politika","Rizik najnižeg zajedničkog imenioca, više koda\u002Ftestova","Koje razlike moraju ostati vidljive, a ne apstrahovane?",[569,570,571,572],"Veliki kontekst","Više informacija po zahtevu","Latencija, trošak, razblaživanje pažnje, površina curenja","Da li podatke treba preuzimati\u002Ffiltrirati umesto uvek ubacivati?",[574,575,576,577],"Moćni alati \u002F autonomija","Više automatizacije od početka do kraja","Veći privilegiji i domet štete pri otkazu","Koje radnje zahtevaju najmanje privilegije, potvrdu ili ljudsko odobrenje?",[579,580,581,582],"Stroga validacija i evidentiranje","Bolji dokazi i operacije","Latencija, skladištenje, privatnost i trošak složenosti","Koji dokazi su potrebni za ovaj nivo rizika?",{"id":584,"data":585,"type":42},"h-adjacent",{"text":586,"level":242},"Kako se ovo razlikuje od srodnih uloga?",{"id":588,"data":589,"type":218},"p-adjacent-intro",{"text":590},"Titule se u velikoj meri preklapaju među kompanijama. Korisna razlika je \u003Cstrong>obim arhitektonske odgovornosti\u003C\u002Fstrong>, a ne HR oznaka.",{"id":592,"data":593,"type":299},"role-comparison",{"rows":594,"title":631,"layout":291,"columns":632},[595,601,607,613,619,625],{"id":596,"label":597,"values":598},"r1","AI Solution Architect",{"role":599,"focus":600},"One concrete AI-enabled solution\u002Fworkload","How requirements, data, models, tools, security, runtime and operations fit together to deliver the target outcome",{"id":602,"label":603,"values":604},"r2","AI Platform Architect",{"role":605,"focus":606},"Reusable AI platform capabilities across many solutions","Shared provider gateways, model access, identity, evaluation, retrieval services, observability, deployment patterns and developer experience",{"id":608,"label":609,"values":610},"r3","Enterprise AI Architect",{"role":611,"focus":612},"Organization\u002Fportfolio-level target architecture","Capability landscape, governance, integration principles, shared platforms, standards, sourcing and strategic constraints across domains",{"id":614,"label":615,"values":616},"r4","AI \u002F ML Engineer",{"role":617,"focus":618},"Implementation of AI\u002FML behavior and pipelines","Models, data, inference, evaluation, application logic and engineering tasks within the architecture",{"id":620,"label":621,"values":622},"r5","Security Architect",{"role":623,"focus":624},"Security architecture across systems","Threats, identity, authorization, data protection, controls, assurance and compliance boundaries",{"id":626,"label":627,"values":628},"r6","Product \u002F Delivery Lead",{"role":629,"focus":630},"Outcome, scope, prioritization and delivery system","Why\u002Fwhat to build, sequencing, stakeholders, milestones, acceptance and value realization","Srodne uloge odgovaraju na različita primarna pitanja",[633,636],{"id":634,"label":635},"role","Uloga",{"id":637,"label":638},"focus","Primarni arhitektonski fokus",{"id":640,"data":641,"type":218},"p-adjacent-2",{"text":642},"U malom produktnom timu, jedna osoba može pokrivati nekoliko ovih obima. U velikom preduzeću, to mogu biti odvojene uloge sa formalnim odborima za pregled. Arhitektonska odgovornost ne nestaje kada se titula promeni.",{"id":644,"data":645,"type":42},"h-implementation",{"text":646,"level":242},"Dokazi implementacije: kako se ove granice pojavljuju u mom radu",{"id":648,"data":649,"type":225},"implementation-boundary",{"body":650,"title":651,"variant":652},"Primeri ispod su \u003Cstrong>originalni dokazi implementacije\u002Fprojekta\u003C\u002Fstrong>. Oni pokazuju kako sam razdvojio potrebe proizvoda, zahteve, arhitekturu, izvršno okruženje, model\u002Fprovajdera, dozvole i validaciju u stvarnom projektnom radu. Oni nisu tvrdnje da svaka organizacija mora koristiti istu strukturu i ne impliciraju usvajanje od strane kupaca ili raspoređivanje na nivou preduzeća.","Dokazi implementacije, ne univerzalno pravilo","success",{"id":654,"data":655,"type":42},"h-senseflow",{"text":656,"level":241},"SenseFlow: potreba → zahtevi → arhitektura → validacija",{"id":658,"data":659,"type":218},"p-senseflow-1",{"text":660},"U projektu SenseFlow Source of Truth, tehnologija je eksplicitno podređena Product Vision-u. Razvojna struktura se kreće od problema i vizije proizvoda kroz korisničke potrebe, vrednost, obim, epove, priče i kriterijume prihvatanja do arhitekture, implementacije, validacije i iteracije.",{"id":662,"data":663,"type":218},"p-senseflow-2",{"text":664},"Zahtevi su dizajnirani tako da budu sledljivi od Cilja proizvoda → Sposobnosti → Epika → Korisničke priče → Kriterijuma prihvatanja → Tehničkih zadataka. Gde je praktično, uključuju funkcionalne zahteve, nefunkcionalne zahteve, zavisnosti, rizike, pretpostavke, kriterijume prihvatanja i metode validacije. Značajne odluke čuvaju odluku, razlog, alternative, kompromise, status i datum\u002Fverziju.",{"id":666,"data":667,"type":218},"p-senseflow-3",{"text":668},"To je arhitektonski rad pre nego što se izabere konkretan AI okvir ili model: štiti vezu između namere proizvoda i tehničkih odluka i čini kasnije promene preglednim umesto implicitnim.",{"id":670,"data":671,"type":42},"h-client",{"text":672,"level":241},"Aaasaasa AI klijent: razdvojite koncepte pre njihovog integrisanja",{"id":674,"data":675,"type":218},"p-client-1",{"text":676},"Aaasaasa AI klijent pruža primer na nivou implementacije. Njegov AI Hub namerno razdvaja \u003Cstrong>agenta\u002Fklijenta\u003C\u002Fstrong>, \u003Cstrong>provajdera\u003C\u002Fstrong>, \u003Cstrong>model\u003C\u002Fstrong>, \u003Cstrong>lokaciju izvršavanja\u002Fkonekcije\u003C\u002Fstrong>, \u003Cstrong>dozvole\u003C\u002Fstrong> i \u003Cstrong>veb klijenta\u003C\u002Fstrong>. Lokalno izvršavanje se ne pretpostavlja da znači lokalno zaključivanje, a dozvole se tretiraju kao politika izvršavanja\u002Falata, a ne kao svojstvo modela.",{"id":678,"data":679,"type":218},"p-client-2",{"text":680},"Desktop arhitektura takođe definiše granicu poverenja: Nuxt renderer nije pouzdan u odnosu na Electron main. Uzak preload i validirani IPC posreduju u pristupu AI servisima, podešavanjima, enkriptovanim tajnama, servisima radnog prostora\u002Fpodataka i izvršnim okruženjima. Cloud akreditivi ostaju u privilegovanom glavnom procesu; renderer kod prima normalizovano stanje umesto sirovih tajni ili neograničenog pristupa operativnom sistemu.",{"id":682,"data":683,"type":218},"p-client-3",{"text":684},"Odluke o rutiranju su takođe arhitektonske. Implementacija ne prelazi tiho sa lokalne rute na plaćeno cloud zaključivanje; cloud ruta zahteva eksplicitnu potvrdu. Direktan chat nema podrazumevano alate za fajl sistem ili shell, dok izvršavanje agenta primenjuje izabrani radni prostor i profil dozvola. To su odluke na nivou rešenja o poverenju, troškovima, izvršavanju i očekivanjima korisnika—ne karakteristike modela.",{"id":686,"data":687,"type":42},"h-current-frameworks",{"text":688,"level":242},"Kako trenutni arhitektonski okviri podržavaju ovaj širi obim",{"id":690,"data":691,"type":218},"p-frameworks-1",{"text":692},"ISO\u002FIEC\u002FIEEE 42010:2022 pruža opštu disciplinu za opise arhitekture kroz softver, sisteme i preduzeća. Namerno je širi od AI i ne propisuje jednu metodu arhitekture ili naziv radnog mesta. To ga čini korisnim ovde kao granicu: AI arhitektura rešenja je i dalje arhitektura, sa interesima zainteresovanih strana, više pogleda i značajnim odnosima koji moraju biti jasno izraženi.",{"id":694,"data":695,"type":218},"p-frameworks-2",{"text":696},"NIST AI RMF 1.0 uokviruje upravljanje AI rizikom kroz \u003Cstrong>Govern, Map, Measure i Manage\u003C\u002Fstrong> i naglašava da upravljanje rizikom treba da bude kontinuirano tokom životnog ciklusa AI sistema. Generativni AI profil (NIST AI 600-1) prilagođava taj okvir GAI rizicima i organizacionim prioritetima. To pojačava da arhitektura ne može da se zaustavi na funkcionalnim performansama modela.",{"id":698,"data":699,"type":218},"p-frameworks-3",{"text":700},"Microsoft-ove trenutne Azure Well-Architected AI smernice razdvajaju dizajn aplikacije, platformu aplikacije, podatke za obuku, podatke za utemeljenje i pitanja platforme podataka i više puta ih povezuju sa pouzdanošću, bezbednošću, operativnom izvrsnošću, performansama i troškovima. AWS-ova Generativna AI i Agentic AI sočiva takođe tretiraju observabilnost, bezbednost, pouzdanost, životni ciklus modela\u002Falata, troškove i ljudski nadzor kao arhitektonske brige.",{"id":702,"data":703,"type":42},"h-misconceptions",{"text":704,"level":242},"Uobičajene zablude",{"id":706,"data":707,"type":291},"misconceptions-table",{"content":708,"stretched":43,"withHeadings":14},[709,712,715,718,721,724,727,730],[710,711],"Zabluda","Ispravka",[713,714],"„Arhitekta bira LLM.“","Izbor modela je jedna odluka unutar veće arhitekture rešenja.",[716,717],"„Prompt inženjering je arhitektura.“","Promptovi utiču na ponašanje, ali ne definišu identitet, pristup podacima, granice poverenja, implementaciju, dozvole alata ili operacije.",[719,720],"„RAG rešava znanje u preduzeću.“","Pretraga je samo jedan podsistem; autorizacija, poreklo, svežina, dokazi, indeksiranje, evaluacija i upravljanje izvorima i dalje zahtevaju dizajn.",[722,723],"„Lokalno izvršavanje znači privatni\u002Flokalni AI.“","Lokacije izvršavanja, zaključivanja, podataka i kontrolne ravni su odvojena arhitektonska svojstva.",[725,726],"„Ako dobavljač nudi zaštitne mere, bezbednost je pokrivena.“","Bezbednost obuhvata identitet, autorizaciju, tajne, tokove podataka, alate, evidentiranje, implementaciju, ljudsko odobrenje i granice provajdera.",[728,729],"„Arhitekta mora da napiše svaku komponentu.“","Praktična implementacija može poboljšati arhitektonski kvalitet, ali uloga je definisana odgovornošću za integrisane odluke, a ne ličnim kodiranjem svakog sloja.",[731,732],"„Dijagram arhitekture dokazuje spremnost za produkciju.“","Spremnost zahteva implementirane kontrole i dokaze validacije kroz kvalitet, bezbednost, operacije i poslovno prihvatanje.",{"id":734,"data":735,"type":42},"h-failures",{"text":736,"level":242},"Načini neuspeha koje AI arhitekta rešenja treba da spreči",{"id":738,"data":739,"type":291},"failures-table",{"content":740,"stretched":43,"withHeadings":14},[741,745,749,753,757,761,765,769,773],[742,743,744],"Način neuspeha","Zašto se javlja","Arhitektonska korekcija",[746,747,748],"Dizajn sa modelom na prvom mestu","Obećavajuća demonstracija modela postaje nacrt sistema","Počnite od ishoda, ograničenja i validacije; izaberite model unutar tog okvira",[750,751,752],"Dozvole prototipa u produkciji","Deljeni akreditivi i širok pristup prežive PoC","Definišite propagaciju identiteta, najmanje privilegije, opsege alata i granice odobrenja rano",[754,755,756],"Pretraga bez autorizacije","Kvalitet pretrage se dizajnira pre pravila pristupa podacima","Prenesite kontekst korisnika\u002Ftenanta u pretragu i sprovedite autorizaciju na granicama pristupa podacima",[758,759,760],"Tihe pretpostavke o provajderu\u002Fizvršavanju","„Lokalno“, „cloud“ i „offline“ se koriste neprecizno","Dokumentujte odvojeno lokaciju izvršavanja, zaključivanja, podataka i kontrolne ravni",[762,763,764],"Nema ugovora o neuspehu","Dizajniran je srećan put, ali ne i ponašanje odbijanja\u002Frezerve\u002Fgreške","Specifikujte ponašanje kada je pretraga prazna, model nedostupan, alat neuspešan i politika odbija",[766,767,768],"Evaluacija posle implementacije","Kvalitet se procenjuje ručno pred lansiranje","Definišite merljivo prihvatanje i reprezentativne skupove evaluacije pre zamrzavanja arhitekture",[770,771,772],"Nesledljiva promena","Modeli, promptovi, pretraga ili dozvole se menjaju bez arhitektonske istorije","Verzionirajte kritičnu konfiguraciju i evidentirajte značajne odluke\u002Fdokaze validacije",[774,775,776],"Operacije se tretiraju samo kao infrastruktura","AI ponašanje nije vidljivo nakon implementacije","Dizajnirajte tragove, metrike kvaliteta, bezbednosne događaje, telemetriju troškova i vraćanje zajedno",{"id":778,"data":779,"type":42},"h-decision-framework",{"text":780,"level":242},"Praktičan redosled odlučivanja",{"id":782,"data":783,"type":339},"decision-flow",{"steps":784,"title":809,"orientation":338},[785,788,791,794,797,800,803,806],{"label":786,"description":787},"Ishod","Definišite korisnički\u002Fposlovni rezultat i eksplicitne ne-ciljeve.",{"label":789,"description":790},"Dokazi i ograničenja","Identifikujte autoritativne podatke, politike, nefunkcionalne zahteve, rizike i uslove prihvatanja.",{"label":792,"description":793},"Granica sistema","Mapirajte korisnike, identitete, aplikacije, podatke, modele\u002Fprovajdere, alate i eksterne sisteme.",{"label":795,"description":796},"Opcije arhitekture","Uporedite obrasce za pretragu, pristup modelu, orkestraciju, implementaciju, dozvole, evaluaciju i observabilnost.",{"label":798,"description":799},"Odluke o kompromisima","Izaberite značajne opcije i sačuvajte obrazloženje, alternative i posledice.",{"label":801,"description":802},"Ugovori implementacije","Pretvorite odluke u API-je, šeme, pravila dozvola, definicije implementacije i inženjerske zadatke.",{"label":804,"description":805},"Validacija","Testirajte implementirani sistem prema originalnim funkcionalnim i nefunkcionalnim zahtevima.",{"label":807,"description":808},"Operativne povratne informacije","Koristite produkcijske dokaze, incidente, metrike kvaliteta i signale troškova\u002Fbezbednosti da pokrenete kontrolisanu promenu.","Redosled odlučivanja u arhitekturi AI rešenja",{"id":811,"data":812,"type":42},"h-edge",{"text":813,"level":242},"Rubni slučajevi i granice uloge",{"id":815,"data":816,"type":218},"p-edge-1",{"text":817},"Neki AI proizvodi su dominirani obukom modela, naučnim eksperimentisanjem ili specijalizovanim hardverom. U tim slučajevima, model\u002Fdata nauka i arhitektura ML sistema mogu postati mnogo dublji od mape na nivou rešenja prikazane ovde. AI arhitekta rešenja i dalje treba integracione i operativne granice, ali specijalistička arhitektura može posedovati samu platformu za obuku.",{"id":819,"data":820,"type":218},"p-edge-2",{"text":821},"Na drugom kraju spektra, jednostavna SaaS integracija možda ne opravdava posvećenog arhitektu. Senior inženjer ili tehnički vođa proizvoda može nositi istu arhitektonsku odgovornost. Koristan test nije titula, već da li se značajne odluke koje prelaze slojeve donose namerno i validiraju.",{"id":823,"data":824,"type":218},"p-edge-3",{"text":825},"Regulisani, suvereni, vazdušno izolovani, bezbednosno kritični, visoko autonomni ili multi-tenant sistemi takođe pomeraju težište. Identitet, izolacija, rezidentnost, garancija, mehanizmi ažuriranja, ljudski nadzor i mogućnost revizije mogu dominirati kvalitetom modela u arhitekturi.",{"id":827,"data":828,"type":42},"h-change-answer",{"text":829,"level":242},"Šta bi promenilo ovaj odgovor?",{"id":831,"data":832,"type":218},"p-change-1",{"text":833},"Tačna granica odgovornosti se menja kada arhitektura pređe sa jedne aplikacije na platformu koja se može ponovo koristiti ili na ciljnu arhitekturu na nivou preduzeća. Zato \u003Cstrong>AI Platform Architect\u003C\u002Fstrong> i \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> zaslužuju odvojen kanonski tretman umesto da budu spojeni u ovu ulogu.",{"id":835,"data":836,"type":218},"p-change-2",{"text":837},"Promene tehnologije su takođe važne. Nove mogućnosti modela, protokoli, lokalna izvršna okruženja i upravljane usluge mogu ukloniti deo implementacionog rada dok istovremeno stvaraju nove granice poverenja ili operativne granice. Stabilna odgovornost je razumeti te promene kao promene sistema—a ne tretirati novi okvir kao zamenu za arhitekturu.",{"id":839,"data":840,"type":42},"h-checklist",{"text":841,"level":242},"AI Solution Architect kontrolna lista",{"id":843,"data":844,"type":291},"checklist-table",{"content":845,"stretched":43,"withHeadings":14},[846,849,851,854,856,859,862,865,868,871,874,877,879],[847,848],"Provera","Pitanje",[786,850],"Da li su korisnički\u002Fposlovni rezultat i granica ne-ciljeva eksplicitni?",[852,853],"Zahtevi","Da li su funkcionalni zahtevi, nefunkcionalni zahtevi, ograničenja i kriterijumi prihvatanja sledljivi?",[268,855],"Da li su autoritativni izvori, poreklo, svežina, zadržavanje i pravila pristupa definisani?",[857,858],"Pretraga\u002Fkontekst","Da li autorizacija doseže do pretrage i konstrukcije konteksta?",[860,861],"Model\u002Fprovajder","Da li je izbor modela\u002Fprovajdera vezan za mogućnosti i ograničenja, a ne za preferencije?",[863,864],"Alati\u002Fagenti","Da li su granice akcija, dozvole, odobrenja i ponašanje pri neuspehu eksplicitni?",[866,867],"Identitet\u002Fbezbednost","Da li su ljudski\u002Fmašinski identiteti, tajne i granice poverenja definisani?",[869,870],"Izvršno okruženje","Da li su lokacije izvršnog okruženja, inferencije, podataka i kontrolne ravni razlikovane?",[872,873],"Evaluacija","Da li postoje merljivi dokazi za kvalitet, bezbednost i prihvatanje?",[875,876],"Opservabilnost","Mogu li se ponašanje u produkciji, neuspesi, troškovi i bezbednosni događaji istražiti?",[286,878],"Da li su značajne arhitektonske odluke i zamene sledljive?",[280,880],"Da li je vlasništvo nad implementacijom, vraćanjem, incidentima i životnim ciklusom jasno?",{"id":882,"data":883,"type":42},"h-conclusion",{"text":884,"level":242},"Zaključak",{"id":886,"data":887,"type":218},"p-conclusion-1",{"text":888},"AI Solution Architect je osoba ili arhitektonska funkcija koja pretvara AI priliku u koherentan tehnički sistem. Ključna veština nije poznavanje najvećeg broja imena modela; to je povezivanje potreba proizvoda, zahteva, podataka, arhitekture aplikacije, AI mogućnosti, bezbednosti, izvršnog okruženja, isporuke i validacije bez gubljenja granica između njih.",{"id":890,"data":891,"type":218},"p-conclusion-2",{"text":892},"Jaka arhitektura AI rešenja se stoga može sažeti kao: \u003Cstrong>definiši cilj → uspostavi zahteve i ograničenja → projektuj granice sistema → učini značajne kompromise eksplicitnim → implementiraj kroz jasne ugovore → validiraj na osnovu dokaza → upravljaj i razvijaj namerno.\u003C\u002Fstrong> Model je važan. Rešenje je proizvod.",{"id":894,"data":895,"type":894},"faq",{"items":896,"title":929},[897,901,905,909,913,917,921,925],{"id":898,"answer":899,"question":900},"faq1","AI Solution Architect prevodi poslovnu ili proizvodnu potrebu u arhitekturu konkretnog AI rešenja, definišući kako logika aplikacije, podaci\u002Fpretraga, modeli, alati, identitet, bezbednost, izvršno okruženje, evaluacija i operacije funkcionišu zajedno.","Šta je AI Solution Architect?",{"id":902,"answer":903,"question":904},"faq2","Ne. Uloge se mogu preklapati, posebno u malim timovima, ali AI inženjer je prvenstveno implementaciona uloga dok arhitekta rešenja poseduje ili koordinira arhitektonske odluke i kompromise koji prelaze slojeve za kompletan radni opseg.","Da li je AI Solution Architect isto što i AI inženjer?",{"id":906,"answer":907,"question":908},"faq3","Ne po definiciji, ali praktično znanje implementacije je veoma vredno jer AI arhitektura prelazi preko API-ja, podataka, pretrage, bezbednosti, izvršnih okruženja i operativnog ponašanja. Uloga je definisana arhitektonskom odgovornošću, a ne ličnim pisanjem svake komponente.","Da li AI Solution Architect mora da kodira?",{"id":910,"answer":911,"question":912},"faq4","Ne. Izbor modela je jedna odluka. Produkciona arhitektura takođe zahteva granice podataka i pretrage, dozvole, alate, izbore provajdera\u002Fizvršnog okruženja, opservabilnost, evaluaciju, pouzdanost, troškove i dizajn životnog ciklusa.","Da li je izbor LLM-a glavni posao?",{"id":914,"answer":915,"question":916},"faq5","AI Solution Architect se fokusira na jedno konkretno rešenje ili radni opseg. AI Platform Architect se fokusira na AI mogućnosti i zaštitne mere koje se mogu ponovo koristiti i podržavaju više rešenja.","Koja je razlika između AI Solution Architect i AI Platform Architect?",{"id":918,"answer":919,"question":920},"faq6","Arhitekta rešenja radi na nivou aplikacije\u002Fradnog opsega. Enterprise AI arhitektura radi preko organizacionog portfolija, ciljne arhitekture, upravljanja, deljenih mogućnosti, principa integracije i strateških ograničenja.","Koja je razlika između AI Solution Architect i Enterprise AI Architect?",{"id":922,"answer":923,"question":924},"faq7","Oni su arhitektonski obrasci ili podsistemi unutar rešenja kada to zahtevi opravdavaju. RAG se bavi kontekstom zasnovanim na pretrazi; agenti dodaju planiranje\u002Fizvršavanje alata i stoga dodatne brige o identitetu, dozvolama, orkestraciji i operacijama.","Gde se uklapaju RAG i agenti?",{"id":926,"answer":927,"question":928},"faq8","Implementacija plus dokazi validacije: funkcionalni testovi, rezultati evaluacije, bezbednosni testovi\u002F testovi autorizacije, merenja performansi i pouzdanosti, opservabilnost, operativna proba i prihvatanje u odnosu na prvobitne zahteve.","Šta dokazuje da arhitektura funkcioniše?","AI Solution Architect — Često postavljana pitanja",{"id":931,"data":932,"type":931},"glossary",{"title":933,"entries":934},"Ključni pojmovi",[935,938,941,945,949,952,955,958],{"term":597,"anchor":936,"definition":937},"ai-solution-architect","Arhitektonska odgovornost za jedno konkretno AI rešenje ili radni opseg, integrišući zahteve proizvoda sa aplikacijom, podacima, modelom, alatima, bezbednošću, izvršnim okruženjem i operativnim dizajnom.",{"term":792,"anchor":939,"definition":940},"system-boundary","Eksplicitno razdvajanje između onoga što pripada rešenju i korisnika, sistema, provajdera, izvora podataka i okruženja sa kojima interaguje.",{"term":942,"anchor":943,"definition":944},"Granica poverenja","trust-boundary","Tačka u kojoj podaci, identiteti ili kontrola prelaze između komponenti sa različitim pretpostavkama poverenja i stoga zahtevaju eksplicitne bezbednosne kontrole.",{"term":946,"anchor":947,"definition":948},"Uzemljenje","grounding","Snabdevanje AI modela relevantnim spoljnim informacijama ili dokazima tako da njegov odgovor može biti zasnovan na izvorima izvan parametara modela.",{"term":564,"anchor":950,"definition":951},"provider-abstraction","Granica aplikacije koja razdvaja delove rešenja od interfejsa jednog modela\u002Fprovajdera. Korisna kada je opravdana potrebama rutiranja, prenosivosti ili politike, ali nije bez kompromisa.",{"term":872,"anchor":953,"definition":954},"evaluation","Strukturirano merenje ponašanja AI radnog opsega u odnosu na definisane kriterijume prihvatanja, uključujući kvalitet zadatka i relevantne bezbednosne, sigurnosne, performansne i operativne osobine.",{"term":603,"anchor":956,"definition":957},"ai-platform-architect","Arhitektonska uloga fokusirana na AI mogućnosti platforme koje se mogu ponovo koristiti i koristi ih više rešenja, a ne na arhitekturu jednog radnog opsega.",{"term":959,"anchor":960,"definition":961},"Enterprise AI Architecture","enterprise-ai-architecture","Arhitektura na nivou organizacije koja koordinira AI mogućnosti, platforme, upravljanje, integraciju i strateška ograničenja preko portfolija.",{"id":963,"data":964,"type":42},"h-related",{"text":965,"level":242},"Povezano kanonsko znanje",{"id":967,"data":968,"type":218},"p-related-1",{"text":969},"Ovaj članak pripada klasteru AI Architecture Foundations. Njegovi direktni temelji su \u003Cstrong>Generative AI Explained: Models, Retrieval, Tools and Applications Are Not the Same Thing\u003C\u002Fstrong> i \u003Cstrong>ADR vs NFR: Architecture Decisions and System Quality Are Not the Same Thing\u003C\u002Fstrong>. Susedni kanonski čvorovi uključuju \u003Cstrong>Agentic AI Explained\u003C\u002Fstrong>, \u003Cstrong>Source of Truth in AI Systems\u003C\u002Fstrong>, \u003Cstrong>Vector Databases, Embeddings and Reranking\u003C\u002Fstrong>, \u003Cstrong>What Is Context Engineering?\u003C\u002Fstrong>, \u003Cstrong>RBAC vs Tenant Isolation\u003C\u002Fstrong>, \u003Cstrong>AI Platform Architect\u003C\u002Fstrong>, \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> i \u003Cstrong>AI Governance\u003C\u002Fstrong>. URL-ovi namerno nisu izmišljeni tamo gde ti čvorovi još nisu objavljeni.",{"id":971,"data":972,"type":979},"related-rag",{"link":973,"meta":974},"https:\u002F\u002Fstajic.de\u002Fsr\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works",{"image":975,"title":977,"description":978},{"url":976},"","Šta je RAG? Najjednostavnije objašnjenje kako funkcioniše","Postojeće stajic.de kanonsko objašnjenje generacije sa pretragom, korisno za deo pretrage\u002Fuzemljenja u arhitekturi AI rešenja.","linkTool",{"id":981,"data":982,"type":42},"h-sources",{"text":983,"level":242},"Primarni izvori i trenutne arhitektonske smernice",{"id":985,"data":986,"type":218},"p-sources-note",{"text":987},"Spoljni izvori ispod podržavaju opšte arhitektonske tvrdnje; odeljci SenseFlow i Aaasaasa AI Client su eksplicitno originalni dokazi projekta\u002Fimplementacije. Reference trenutnog stanja su proverene 8. oktobra 2026. NIST napominje da se AI RMF 1.0 revidira, tako da reference upravljanja osetljive na verziju treba ponovo proveriti kada naslednik bude objavljen.",{"id":989,"data":990,"type":979},"src-iso-42010",{"link":991,"meta":992},"https:\u002F\u002Fwww.iso.org\u002Fstandard\u002F74393.html",{"image":993,"title":994,"description":995},{"url":976},"ISO\u002FIEC\u002FIEEE 42010:2022 — Opis arhitekture","Trenutni međunarodni standard za strukturu i izražavanje opisa arhitekture. Razlikuje arhitekturu od njenog opisa i ne propisuje jednu metodu arhitekture, alat ili format snimanja.",{"id":997,"data":998,"type":979},"src-nist-rmf",{"link":999,"meta":1000},"https:\u002F\u002Fwww.nist.gov\u002Fitl\u002Fai-risk-management-framework",{"image":1001,"title":1002,"description":1003},{"url":976},"NIST okvir za upravljanje rizicima veštačke inteligencije","NIST stranica sa resursima o AI RMF. Od oktobra 2026. navodi da se AI RMF 1.0 revidira i povezuje Generativni AI profil i povezane resurse.",{"id":1005,"data":1006,"type":979},"src-nist-core",{"link":1007,"meta":1008},"https:\u002F\u002Fairc.nist.gov\u002Fairmf-resources\u002Fairmf\u002F5-sec-core\u002F",{"image":1009,"title":1010,"description":1011},{"url":976},"NIST AI RMF jezgro — Upravljaj, Mapiraj, Meri, Upravljaj","Zvanična NIST AIRC prezentacija jezgra AI RMF 1.0, uključujući četiri funkcije i okvir za upravljanje rizicima orijentisan na životni ciklus.",{"id":1013,"data":1014,"type":979},"src-nist-gai",{"link":1015,"meta":1016},"https:\u002F\u002Fwww.nist.gov\u002Fpublications\u002Fartificial-intelligence-risk-management-framework-generative-artificial-intelligence",{"image":1017,"title":1018,"description":1019},{"url":976},"NIST AI 600-1 — Generativni AI profil","Međusektorski generativni AI profil za AI RMF 1.0, objavljen 26. jula 2024. i ažuriran od strane NIST-a 2026. godine.",{"id":1021,"data":1022,"type":979},"src-ms-start",{"link":1023,"meta":1024},"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fget-started",{"image":1025,"title":1026,"description":1027},{"url":976},"Microsoft Azure Well-Architected — AI radna opterećenja","Aktuelne smernice za arhitekturu na nivou radnog opterećenja koje pokrivaju dizajn AI aplikacija, platformu aplikacija, podatke za obuku, podatke za utemeljenje, platformu podataka i pitanja spremnosti za produkciju.",{"id":1029,"data":1030,"type":979},"src-ms-app",{"link":1031,"meta":1032},"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fapplication-design",{"image":1033,"title":1034,"description":1035},{"url":976},"Microsoft — Dizajn aplikacija za AI radna opterećenja","Smernice o apstrakciji modela\u002Falata, granicama pristupa podacima, propagaciji identiteta, autorizaciji i razdvajanju slojeva klijenta, inteligencije, znanja i alata.",{"id":1037,"data":1038,"type":979},"src-ms-security",{"link":1039,"meta":1040},"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fdesign-principles",{"image":1041,"title":1042,"description":1043},{"url":976},"Microsoft — Principi dizajna za AI radna opterećenja","Aktuelni principi dizajna AI radnih opterećenja u pogledu pouzdanosti, bezbednosti, troškova, operativne izvrsnosti i performansi, uključujući odgovornosti za identitet i zaštitu podataka.",{"id":1045,"data":1046,"type":979},"src-ms-ops",{"link":1047,"meta":1048},"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fmlops-genaiops",{"image":1049,"title":1050,"description":1051},{"url":976},"Microsoft — MLOps i GenAIOps za AI radna opterećenja","Smernice za životni ciklus u produkciji koje pokrivaju nadzor, kapije kvaliteta, ponašanje modela\u002Fupita, bezbednost i operativno merenje.",{"id":1053,"data":1054,"type":979},"src-aws-genai",{"link":1055,"meta":1056},"https:\u002F\u002Fdocs.aws.amazon.com\u002Fwellarchitected\u002Flatest\u002Fgenerative-ai-lens\u002F",{"image":1057,"title":1058,"description":1059},{"url":976},"AWS Well-Architected sočivo za generativnu veštačku inteligenciju","AWS smernice za arhitekturu generativnih AI radnih opterećenja u pogledu operativne izvrsnosti, bezbednosti, pouzdanosti, efikasnosti performansi, optimizacije troškova i održivosti.",{"id":1061,"data":1062,"type":979},"src-aws-agentic",{"link":1063,"meta":1064},"https:\u002F\u002Fdocs.aws.amazon.com\u002Fwellarchitected\u002Flatest\u002Fagentic-ai-lens\u002F",{"image":1065,"title":1066,"description":1067},{"url":976},"AWS Well-Architected sočivo za agentnu veštačku inteligenciju","Objavljeno 2026. godine, pokriva arhitektonska pitanja specifična za agente, uključujući identitete, alate, orkestraciju, ljudski nadzor, pouzdanost, praćenje i troškove petlje zaključivanja.","2.31","AI Solution Architect pretvara poslovne zahteve u AI sistem spreman za produkciju, obuhvatajući podatke, modele, alate, bezbednost, izvršno okruženje, evaluaciju i operacije.","\u002Fuploads\u002F2026\u002F10\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs-1791476643267-1st5xz.webp","what-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs-1791476643267-1st5xz","PUBLISHED","2026-10-08T12:23:00.000Z","2026-10-08T16:23:08.916Z","2026-10-08T16:31:54.093Z",{"en":1077,"de":1078,"sr":1079,"es":1080,"fr":1081,"it":1082,"ru":1083,"zh":1084},"\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u002Fde\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u002Fsr\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u002Fes\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u002Ffr\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u002Fit\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u002Fru\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs","\u002Fzh\u002Fblog\u002Fwhat-is-an-ai-solution-architect-system-boundaries-responsibilities-and-trade-offs",[1086,1090,1094,1098],{"id":1087,"name":1088,"slug":1089},57,"Granice podataka","data-boundaries",{"id":1091,"name":1092,"slug":1093},84,"Politike i granice podataka","policy-and-data",{"id":1095,"name":1096,"slug":1097},80,"Pristup i identitet","access-and-identity",{"id":1099,"name":1100,"slug":1101},54,"Model prijetnji","threat-model",{"id":1103,"login":1104,"email":1105,"displayName":1106},"20","rooth8233","aleksandar@stajic.de","Aleksandar Stajić",[1108,1788],{"lang":1109,"title":1110,"content":1111,"contentJson":1112,"excerpt":1787},"en","What Is an AI Solution Architect? System Boundaries, Responsibilities and Trade-offs","{\"time\":1791476244367,\"blocks\":[{\"id\":\"intro\",\"type\":\"paragraph\",\"data\":{\"text\":\"An \u003Cstrong>AI Solution Architect\u003C\u002Fstrong> translates a business or product need into the architecture of a concrete AI-enabled solution. The role defines system boundaries and the significant choices across application logic, authoritative data, retrieval and context, models and providers, tools or agents, identity and permissions, security, runtime and deployment, observability, evaluation, cost and operational behavior. It is not simply model selection or prompt engineering: the architectural responsibility is to make the whole solution implementable, governable, testable and operable.\"}},{\"id\":\"direct\",\"type\":\"callout\",\"data\":{\"variant\":\"info\",\"title\":\"Direct answer\",\"body\":\"\u003Cstrong>An AI Solution Architect designs the complete AI-enabled solution, not just the AI model.\u003C\u002Fstrong> The role connects requirements and non-functional requirements to architecture decisions, composes the necessary application\u002Fdata\u002Fmodel\u002Ftool\u002Fruntime layers, makes trust and failure boundaries explicit, and defines how the implemented system will be validated and operated.\"}},{\"id\":\"role-note\",\"type\":\"callout\",\"data\":{\"variant\":\"note\",\"title\":\"Terminology note\",\"body\":\"\u003Cstrong>AI Solution Architect is a practical role label, not a universally standardized job title.\u003C\u002Fstrong> ISO\u002FIEC\u002FIEEE 42010:2022 standardizes concepts for architecture descriptions; it does not define this job role. Organizations can distribute the responsibilities across several people. In this article, the term means the architecture responsibility for one concrete AI-enabled solution or workload.\"}},{\"id\":\"version-note\",\"type\":\"callout\",\"data\":{\"variant\":\"note\",\"title\":\"Current-source note — 8 October 2026\",\"body\":\"The architecture principles here are intentionally vendor-neutral, while current vendor guidance is used as implementation evidence. NIST AI RMF 1.0 is currently under revision; NIST AI 600-1 remains the published Generative AI Profile. Microsoft and AWS guidance cited below reflects current production concerns such as identity, data boundaries, model abstraction, security, observability, evaluation, reliability and cost.\"}},{\"id\":\"toc\",\"type\":\"tableOfContents\",\"data\":{\"title\":\"Contents\",\"minLevel\":2,\"maxLevel\":3}},{\"id\":\"h-meaning\",\"type\":\"header\",\"data\":{\"text\":\"What does an AI Solution Architect actually architect?\",\"level\":2}},{\"id\":\"p-meaning-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"The object of the work is the \u003Cstrong>solution\u003C\u002Fstrong>: the complete socio-technical system that turns a need into useful, controlled behavior. A model may be central to that system, but it is still only one dependency. The same model can participate in a safe internal search assistant, an unsafe over-privileged agent, a low-latency customer feature, or a high-cost prototype that cannot be operated economically. Architecture determines those differences.\"}},{\"id\":\"p-meaning-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"A useful boundary is therefore: \u003Cstrong>business outcome → requirements → system responsibilities → architecture decisions → implementation → validation → operation\u003C\u002Fstrong>. The AI Solution Architect works across this chain while collaborating with product, engineering, data, security, infrastructure, governance and domain specialists.\"}},{\"id\":\"solution-vs-model\",\"type\":\"comparison\",\"data\":{\"title\":\"The solution is wider than the model\",\"layout\":\"table\",\"columns\":[{\"id\":\"model\",\"label\":\"Model-centric question\"},{\"id\":\"solution\",\"label\":\"Solution-architecture question\"}],\"rows\":[{\"id\":\"m1\",\"label\":\"Capability\",\"values\":{\"model\":\"Which model can generate or reason well enough?\",\"solution\":\"Which combination of model, data, application logic, retrieval, tools and controls produces the required behavior?\"}},{\"id\":\"m2\",\"label\":\"Data\",\"values\":{\"model\":\"What context can fit in the prompt?\",\"solution\":\"What is authoritative, who may access it, how is it retrieved, versioned, filtered and cited?\"}},{\"id\":\"m3\",\"label\":\"Security\",\"values\":{\"model\":\"Does the provider offer security features?\",\"solution\":\"What are the trust boundaries, identities, permissions, secrets, data flows and failure containment mechanisms?\"}},{\"id\":\"m4\",\"label\":\"Operations\",\"values\":{\"model\":\"What is the token latency?\",\"solution\":\"How is the complete workload deployed, observed, evaluated, recovered, versioned and cost-controlled?\"}},{\"id\":\"m5\",\"label\":\"Change\",\"values\":{\"model\":\"Can we switch models?\",\"solution\":\"Which dependencies are abstracted, what changes require an ADR, and how do we validate that a replacement still meets requirements?\"}}]}},{\"id\":\"h-simple\",\"type\":\"header\",\"data\":{\"text\":\"The simplest example\",\"level\":2}},{\"id\":\"p-simple-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Imagine a company wants an internal assistant that answers technicians’ questions from maintenance manuals and operating procedures. The visible feature sounds simple: type a question and receive an answer with sources.\"}},{\"id\":\"p-simple-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"The architecture question is much larger. Which documents are authoritative? How are users authenticated? Must retrieval respect department or site permissions? Is the answer allowed to use only retrieved evidence? Which model is acceptable for the data classification? Can a cloud provider receive the content? What happens when retrieval finds nothing? How are citations produced? How is answer quality evaluated? What latency and cost are acceptable? Who can see logs, and what may be stored in them?\"}},{\"id\":\"simple-flow\",\"type\":\"processFlow\",\"data\":{\"title\":\"From need to an operable AI solution\",\"orientation\":\"auto\",\"steps\":[{\"label\":\"1. Define the outcome\",\"description\":\"Clarify the user, business value, task boundary and what a successful answer or action means.\"},{\"label\":\"2. Capture requirements\",\"description\":\"Make functional requirements, NFRs, constraints, data rules, risk tolerance and acceptance criteria explicit.\"},{\"label\":\"3. Establish boundaries\",\"description\":\"Identify users, identities, applications, authoritative data, model\u002Fprovider dependencies, tools, external systems and trust zones.\"},{\"label\":\"4. Design the architecture\",\"description\":\"Choose data\u002Fretrieval, model, orchestration, tool, permission, runtime, deployment, fallback and observability patterns.\"},{\"label\":\"5. Record significant decisions\",\"description\":\"Preserve architectural choices, alternatives, trade-offs and consequences so later changes remain understandable.\"},{\"label\":\"6. Implement and integrate\",\"description\":\"Turn the architecture into application code, APIs, policies, infrastructure, workflows and operational controls.\"},{\"label\":\"7. Validate and operate\",\"description\":\"Test quality, security, reliability, cost and user outcomes; monitor the real workload and feed evidence back into decisions.\"}]}},{\"id\":\"h-where-simple-stops\",\"type\":\"header\",\"data\":{\"text\":\"Where the simple example stops\",\"level\":2}},{\"id\":\"p-stop-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"A proof of concept can often skip architecture that production cannot. A developer may hard-code one provider, use a shared API key, place all documents in one index, run retrieval without user-context filtering, log prompts verbatim and judge quality manually. That can demonstrate feasibility, but it does not establish a production architecture.\"}},{\"id\":\"p-stop-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Production introduces constraints that interact: tenant or user isolation, privacy, data residency, throughput, latency, cost, provider quotas, fallback behavior, auditability, model version changes, retrieval quality, tool permissions, incident response and deployment lifecycle. The architect’s job is not to maximize every quality at once; it is to make the trade-offs explicit and design a solution that satisfies the actual priority set.\"}},{\"id\":\"h-responsibility-map\",\"type\":\"header\",\"data\":{\"text\":\"Architecture responsibility map\",\"level\":2}},{\"id\":\"p-resp-intro\",\"type\":\"paragraph\",\"data\":{\"text\":\"The exact split varies by organization, but the following map captures the recurring responsibilities of solution-level AI architecture. The architect may not personally implement every layer; the responsibility is to make the layers fit together coherently and to keep the critical decisions traceable.\"}},{\"id\":\"responsibility-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Architecture area\",\"Questions the AI Solution Architect must resolve\",\"Typical outputs\"],[\"Outcome and scope\",\"Who is the user? What task is in scope? What must the system not do? What constitutes success?\",\"Solution context, capability boundary, acceptance criteria\"],[\"Requirements and NFRs\",\"What quality, security, availability, latency, cost, residency and compliance constraints apply?\",\"Requirement map, NFRs, constraints, validation criteria\"],[\"Application and orchestration\",\"Where does deterministic application logic end and AI behavior begin? How are workflows coordinated?\",\"Component model, APIs, orchestration boundaries, failure paths\"],[\"Authoritative data and retrieval\",\"What is the Source of Truth? How is data ingested, authorized, retrieved, filtered, ranked and cited?\",\"Data flows, retrieval architecture, metadata and authorization rules\"],[\"Model and provider layer\",\"Which capabilities are required? Which provider\u002Fruntime constraints matter? What should be abstracted?\",\"Model\u002Fprovider decision, routing\u002Ffallback policy, abstraction boundary\"],[\"Tools and agents\",\"What actions can the system take? Which actions require approval? How are tool identities and permissions enforced?\",\"Tool contracts, agent boundaries, approval and least-privilege rules\"],[\"Identity and security\",\"Which human and machine identities exist? Where are secrets held? Which trust boundaries are crossed?\",\"Threat\u002Ftrust boundary model, identity propagation, secrets and authorization design\"],[\"Runtime and deployment\",\"Where do components execute? What is local, cloud, edge or hybrid? What network and availability assumptions exist?\",\"Deployment view, runtime topology, environment and connectivity decisions\"],[\"Evaluation and observability\",\"How is quality measured before and after release? What traces, metrics, logs and evidence are needed?\",\"Evaluation plan, telemetry, audit trail, release gates\"],[\"Operations and change\",\"How are models\u002Fprompts\u002Fconfiguration\u002Fdata versions changed, rolled back and supported?\",\"Operational model, lifecycle controls, ADRs, runbooks, change rules\"]]}},{\"id\":\"h-requirements\",\"type\":\"header\",\"data\":{\"text\":\"1. Turn product need into architectural requirements\",\"level\":3}},{\"id\":\"p-requirements-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"AI architecture begins before model selection. The architect first determines what the solution is expected to achieve and under which constraints. This includes functional behavior, but also the NFRs and policies that narrow the design space: security, reliability, latency, privacy, residency, maintainability, cost and operational support.\"}},{\"id\":\"p-requirements-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"This is where A02’s distinction matters: a requirement such as “unauthorized users must not retrieve restricted documents” is not an architecture decision. It is a driver. Decisions about identity propagation, index partitioning, metadata filtering, API boundaries and authorization enforcement are architectural responses that must later be validated.\"}},{\"id\":\"h-data\",\"type\":\"header\",\"data\":{\"text\":\"2. Design authoritative data, retrieval and context\",\"level\":3}},{\"id\":\"p-data-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"AI systems often fail at the boundary between model behavior and enterprise truth. An architect must define which sources are authoritative, what freshness and provenance mean, how access control reaches retrieval, and how retrieved evidence becomes model context. A vector database, embedding model or RAG library is not the architecture by itself.\"}},{\"id\":\"p-data-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Microsoft’s current AI workload guidance makes the same separation explicit: application code should not bypass data-access boundaries; user or tenant context should propagate into retrieval and filtering; grounding data must be designed for searchability while still meeting security and compliance requirements.\"}},{\"id\":\"h-model\",\"type\":\"header\",\"data\":{\"text\":\"3. Treat models and providers as dependencies, not the whole system\",\"level\":3}},{\"id\":\"p-model-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Model selection matters, but it should be driven by required capability and constraints. The architect considers reasoning or generation quality, modality, context limits, latency, data handling, deployment location, provider availability, cost, observability and replacement risk.\"}},{\"id\":\"p-model-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Provider abstraction is not automatically “better architecture.” It adds engineering cost and can hide provider-specific capabilities. It is justified when portability, fallback, policy separation or multi-provider routing is an explicit requirement. Otherwise a direct integration can be the better decision. The point is to make the trade-off intentional.\"}},{\"id\":\"h-tools\",\"type\":\"header\",\"data\":{\"text\":\"4. Architect tools, actions and agent boundaries\",\"level\":3}},{\"id\":\"p-tools-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"When an AI system can call tools, modify data, send messages, run code or operate business systems, the architectural risk changes. Tool access needs its own identity and authorization model. The model’s ability to request an action is not the same as permission to execute it.\"}},{\"id\":\"p-tools-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"For agentic workloads, current AWS guidance emphasizes additional dimensions such as agent identities, tool access, orchestration, human oversight, tracing, failure handling and cost of iterative reasoning loops. These are solution concerns even when a framework hides some of the implementation mechanics.\"}},{\"id\":\"h-security\",\"type\":\"header\",\"data\":{\"text\":\"5. Make trust boundaries and permissions explicit\",\"level\":3}},{\"id\":\"p-security-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"A production AI solution has multiple trust boundaries: browser or client, application backend, AI orchestration, retrieval\u002Fdata services, model providers, tool APIs, local runtimes and external systems. Each boundary should answer: who is calling, on whose behalf, with what credential, for which resource, with what audit trail, and with what failure containment?\"}},{\"id\":\"p-security-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Security cannot be deferred to a “guardrail” around the model. Microsoft’s AI workload guidance explicitly places security across all architecture layers and calls for identity\u002Faccess management, data protection, content controls and lifecycle security. NIST likewise treats governance and risk management as continuous across the AI lifecycle.\"}},{\"id\":\"h-runtime\",\"type\":\"header\",\"data\":{\"text\":\"6. Decide where the system actually runs\",\"level\":3}},{\"id\":\"p-runtime-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"“Local AI,” “cloud AI,” and “hybrid AI” are architectural statements only when the execution and data paths are precise. A local desktop process can still call a cloud model. A cloud-hosted application can retrieve from an on-premises data source. An air-gapped solution has entirely different update, model-distribution and observability constraints.\"}},{\"id\":\"p-runtime-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"The architect therefore separates \u003Cstrong>runtime location\u003C\u002Fstrong>, \u003Cstrong>inference location\u003C\u002Fstrong>, \u003Cstrong>data location\u003C\u002Fstrong> and \u003Cstrong>control plane\u003C\u002Fstrong>. Conflating them creates false security and deployment assumptions.\"}},{\"id\":\"h-eval\",\"type\":\"header\",\"data\":{\"text\":\"7. Define evaluation, observability and operational acceptance\",\"level\":3}},{\"id\":\"p-eval-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"AI behavior is partly nondeterministic, so the release definition cannot rely only on conventional unit tests. The architecture needs measurable acceptance: task success, groundedness or citation correctness where relevant, refusal behavior, tool safety, latency, cost, reliability and security tests. The exact metrics depend on the use case.\"}},{\"id\":\"p-eval-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Microsoft’s current Well-Architected AI guidance treats monitoring as continuous and applies it across model behavior, prompts\u002Fcompletions, anomalies, security and production quality gates. AWS similarly treats observability, lifecycle management and model\u002Fprompt traceability as operational architecture concerns.\"}},{\"id\":\"h-artifacts\",\"type\":\"header\",\"data\":{\"text\":\"What should the role produce?\",\"level\":2}},{\"id\":\"p-artifacts-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Architecture is not the slide deck. The useful outputs are the artifacts that let engineering, security, product and operations make consistent decisions and later understand why the system exists in its current form.\"}},{\"id\":\"artifacts-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Artifact\",\"Purpose\"],[\"Solution context and boundary\",\"Shows users, external systems, major responsibilities and what is outside scope\"],[\"Requirement\u002FNFR map\",\"Connects product need and constraints to architecture work and validation\"],[\"Component and data-flow views\",\"Shows application, data\u002Fretrieval, model, tools, identity and runtime interactions\"],[\"Trust and permission model\",\"Makes identities, secrets, authorization, sensitive data and high-risk actions explicit\"],[\"Architecture Decision Records\",\"Preserves significant choices, alternatives, trade-offs, status and consequences\"],[\"Evaluation and acceptance plan\",\"Defines evidence required to claim that the solution meets quality and safety expectations\"],[\"Deployment and operational view\",\"Defines environments, runtime locations, observability, rollback, incident and lifecycle responsibilities\"],[\"Traceability links\",\"Connects requirements, decisions, implementation work, tests and operational evidence\"]]}},{\"id\":\"h-tradeoffs\",\"type\":\"header\",\"data\":{\"text\":\"The work is mostly trade-offs, not “best practice” selection\",\"level\":2}},{\"id\":\"p-tradeoffs-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Architecture exists because desirable qualities conflict. A lower-cost model may reduce quality. A more capable model may increase latency or data-governance constraints. Aggressive caching can improve cost and speed while complicating freshness. More autonomous agents can reduce human effort while increasing blast radius and audit requirements.\"}},{\"id\":\"tradeoff-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Decision\",\"Potential benefit\",\"Potential cost \u002F risk\",\"Architectural question\"],[\"Managed cloud model\",\"Fast adoption, strong managed capabilities\",\"External dependency, data and cost constraints\",\"Does the workload permit the provider\u002Fdata path and meet resilience needs?\"],[\"Local\u002Fself-hosted inference\",\"Control, offline\u002Fprivate options\",\"Hardware, operations, model lifecycle burden\",\"Is the control benefit worth the operational responsibility?\"],[\"Single provider integration\",\"Simpler implementation, full provider features\",\"Higher switching\u002Ffailure concentration\",\"Is portability or fallback actually required?\"],[\"Provider abstraction\",\"Portability, routing and policy separation\",\"Lowest-common-denominator risk, more code\u002Ftests\",\"Which differences must remain visible rather than abstracted?\"],[\"Large context\",\"More information per request\",\"Latency, cost, attention dilution, leakage surface\",\"Should data be retrieved\u002Ffiltered instead of always injected?\"],[\"Powerful tools \u002F autonomy\",\"More end-to-end automation\",\"Higher privilege and failure blast radius\",\"Which actions require least privilege, confirmation or human approval?\"],[\"Strict validation and logging\",\"Better evidence and operations\",\"Latency, storage, privacy and complexity cost\",\"What evidence is required for this risk level?\"]]}},{\"id\":\"h-adjacent\",\"type\":\"header\",\"data\":{\"text\":\"How is this different from adjacent roles?\",\"level\":2}},{\"id\":\"p-adjacent-intro\",\"type\":\"paragraph\",\"data\":{\"text\":\"Titles overlap heavily across companies. The useful distinction is the \u003Cstrong>scope of architecture responsibility\u003C\u002Fstrong>, not the HR label.\"}},{\"id\":\"role-comparison\",\"type\":\"comparison\",\"data\":{\"title\":\"Adjacent roles answer different primary questions\",\"layout\":\"table\",\"columns\":[{\"id\":\"role\",\"label\":\"Role\"},{\"id\":\"focus\",\"label\":\"Primary architecture focus\"}],\"rows\":[{\"id\":\"r1\",\"label\":\"AI Solution Architect\",\"values\":{\"role\":\"One concrete AI-enabled solution\u002Fworkload\",\"focus\":\"How requirements, data, models, tools, security, runtime and operations fit together to deliver the target outcome\"}},{\"id\":\"r2\",\"label\":\"AI Platform Architect\",\"values\":{\"role\":\"Reusable AI platform capabilities across many solutions\",\"focus\":\"Shared provider gateways, model access, identity, evaluation, retrieval services, observability, deployment patterns and developer experience\"}},{\"id\":\"r3\",\"label\":\"Enterprise AI Architect\",\"values\":{\"role\":\"Organization\u002Fportfolio-level target architecture\",\"focus\":\"Capability landscape, governance, integration principles, shared platforms, standards, sourcing and strategic constraints across domains\"}},{\"id\":\"r4\",\"label\":\"AI \u002F ML Engineer\",\"values\":{\"role\":\"Implementation of AI\u002FML behavior and pipelines\",\"focus\":\"Models, data, inference, evaluation, application logic and engineering tasks within the architecture\"}},{\"id\":\"r5\",\"label\":\"Security Architect\",\"values\":{\"role\":\"Security architecture across systems\",\"focus\":\"Threats, identity, authorization, data protection, controls, assurance and compliance boundaries\"}},{\"id\":\"r6\",\"label\":\"Product \u002F Delivery Lead\",\"values\":{\"role\":\"Outcome, scope, prioritization and delivery system\",\"focus\":\"Why\u002Fwhat to build, sequencing, stakeholders, milestones, acceptance and value realization\"}}]}},{\"id\":\"p-adjacent-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"In a small product team, one person may cover several of these scopes. In a large enterprise, they may be separate roles with formal review boards. The architecture responsibility does not disappear when the title changes.\"}},{\"id\":\"h-implementation\",\"type\":\"header\",\"data\":{\"text\":\"Implementation evidence: how these boundaries appear in my own work\",\"level\":2}},{\"id\":\"implementation-boundary\",\"type\":\"callout\",\"data\":{\"variant\":\"success\",\"title\":\"Implementation evidence, not a universal rule\",\"body\":\"The examples below are \u003Cstrong>original implementation\u002Fproject evidence\u003C\u002Fstrong>. They show how I have separated product need, requirements, architecture, runtime, model\u002Fprovider, permissions and validation in real project work. They are not claims that every organization must use the same structure, and they do not imply customer adoption or enterprise-scale deployment.\"}},{\"id\":\"h-senseflow\",\"type\":\"header\",\"data\":{\"text\":\"SenseFlow: need → requirements → architecture → validation\",\"level\":3}},{\"id\":\"p-senseflow-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"In the SenseFlow project Source of Truth, technology is explicitly subordinate to Product Vision. The development structure moves from problem and product vision through user needs, value, scope, epics, stories and acceptance criteria into architecture, implementation, validation and iteration.\"}},{\"id\":\"p-senseflow-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Requirements are designed to be traceable from Product Goal → Capability → Epic → User Story → Acceptance Criteria → Technical Tasks. Where practical, they include functional requirements, NFRs, dependencies, risks, assumptions, acceptance criteria and validation methods. Significant decisions preserve the decision, reason, alternatives, trade-offs, status and date\u002Fversion.\"}},{\"id\":\"p-senseflow-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"That is architectural work before a specific AI framework or model is chosen: it protects the connection between product intent and technical decisions and makes later change reviewable rather than implicit.\"}},{\"id\":\"h-client\",\"type\":\"header\",\"data\":{\"text\":\"Aaasaasa AI Client: separate concepts before integrating them\",\"level\":3}},{\"id\":\"p-client-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Aaasaasa AI Client provides a more implementation-level example. Its AI Hub deliberately separates \u003Cstrong>agent\u002Fclient\u003C\u002Fstrong>, \u003Cstrong>provider\u003C\u002Fstrong>, \u003Cstrong>model\u003C\u002Fstrong>, \u003Cstrong>connection\u002Fruntime location\u003C\u002Fstrong>, \u003Cstrong>permissions\u003C\u002Fstrong> and \u003Cstrong>web client\u003C\u002Fstrong>. A local runtime is not assumed to mean local inference, and permissions are treated as runtime\u002Ftool policy rather than as a property of the model.\"}},{\"id\":\"p-client-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"The desktop architecture also defines a trust boundary: the Nuxt renderer is untrusted relative to Electron main. A narrow preload and validated IPC mediate access to AI services, settings, encrypted secrets, workspace\u002Fdata services and runtimes. Cloud credentials remain in the privileged main process; renderer code receives normalized state instead of raw secrets or unrestricted operating-system access.\"}},{\"id\":\"p-client-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"Routing decisions are similarly architectural. The implementation does not silently fall back from a local route to paid cloud inference; a cloud route requires explicit confirmation. Direct Chat has no filesystem or shell tools by default, while agent execution applies a selected workspace and permission profile. These are solution-level decisions about trust, cost, execution and user expectation—not model features.\"}},{\"id\":\"h-current-frameworks\",\"type\":\"header\",\"data\":{\"text\":\"How current architecture frameworks support this broader scope\",\"level\":2}},{\"id\":\"p-frameworks-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"ISO\u002FIEC\u002FIEEE 42010:2022 provides a general discipline for architecture descriptions across software, systems and enterprises. It is deliberately broader than AI and does not prescribe one architecting method or job title. That makes it useful here as a boundary: AI solution architecture is still architecture, with stakeholder concerns, multiple views and significant relationships that must be expressed clearly.\"}},{\"id\":\"p-frameworks-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"NIST AI RMF 1.0 frames AI risk management through \u003Cstrong>Govern, Map, Measure and Manage\u003C\u002Fstrong> and emphasizes that risk management should be continuous across the AI system lifecycle. The Generative AI Profile (NIST AI 600-1) adapts that framework to GAI risks and organizational priorities. This reinforces that architecture cannot stop at functional model performance.\"}},{\"id\":\"p-frameworks-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"Microsoft’s current Azure Well-Architected AI guidance separates application design, application platform, training data, grounding data and data platform concerns and repeatedly connects them to reliability, security, operational excellence, performance and cost. AWS’s Generative AI and Agentic AI lenses similarly treat observability, security, reliability, model\u002Ftool lifecycle, cost and human oversight as architecture concerns.\"}},{\"id\":\"h-misconceptions\",\"type\":\"header\",\"data\":{\"text\":\"Common misconceptions\",\"level\":2}},{\"id\":\"misconceptions-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Misconception\",\"Correction\"],[\"“The architect chooses the LLM.”\",\"Model choice is one decision inside a larger solution architecture.\"],[\"“Prompt engineering is the architecture.”\",\"Prompts affect behavior, but they do not define identity, data access, trust boundaries, deployment, tool permissions or operations.\"],[\"“RAG solves enterprise knowledge.”\",\"Retrieval is only one subsystem; authorization, provenance, freshness, evidence, indexing, evaluation and source governance still need design.\"],[\"“Local runtime means private\u002Flocal AI.”\",\"Runtime, inference, data and control-plane locations are separate architectural properties.\"],[\"“If a vendor offers guardrails, security is covered.”\",\"Security spans identity, authorization, secrets, data flows, tools, logging, deployment, human approval and provider boundaries.\"],[\"“The architect must write every component.”\",\"Hands-on implementation can improve architectural quality, but the role is defined by integrated decision responsibility, not by personally coding every layer.\"],[\"“An architecture diagram proves production readiness.”\",\"Readiness requires implemented controls and validation evidence across quality, security, operations and business acceptance.\"]]}},{\"id\":\"h-failures\",\"type\":\"header\",\"data\":{\"text\":\"Failure modes an AI Solution Architect should prevent\",\"level\":2}},{\"id\":\"failures-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Failure mode\",\"Why it happens\",\"Architectural correction\"],[\"Model-first design\",\"A promising model demo becomes the system blueprint\",\"Start from outcome, constraints and validation; select the model inside that frame\"],[\"Prototype permissions in production\",\"Shared credentials and broad access survive the PoC\",\"Define identity propagation, least privilege, tool scopes and approval boundaries early\"],[\"Retrieval without authorization\",\"Search quality is designed before data-access rules\",\"Carry user\u002Ftenant context into retrieval and enforce authorization at data-access boundaries\"],[\"Silent provider\u002Fruntime assumptions\",\"“Local”, “cloud” and “offline” are used imprecisely\",\"Document runtime, inference, data and control-plane location separately\"],[\"No failure contract\",\"The happy path is designed but refusal\u002Ffallback\u002Ferror behavior is not\",\"Specify retrieval-empty, model-unavailable, tool-failure and policy-denied behavior\"],[\"Evaluation after implementation\",\"Quality is judged manually near launch\",\"Define measurable acceptance and representative evaluation sets before architecture freezes\"],[\"Untraceable change\",\"Models, prompts, retrieval or permissions change without architectural history\",\"Version critical configuration and record significant decisions\u002Fvalidation evidence\"],[\"Operations treated as infrastructure only\",\"AI behavior is not observable after deployment\",\"Design traces, quality metrics, security events, cost telemetry and rollback together\"]]}},{\"id\":\"h-decision-framework\",\"type\":\"header\",\"data\":{\"text\":\"A practical decision sequence\",\"level\":2}},{\"id\":\"decision-flow\",\"type\":\"processFlow\",\"data\":{\"title\":\"AI solution architecture decision sequence\",\"orientation\":\"auto\",\"steps\":[{\"label\":\"Outcome\",\"description\":\"Define the user\u002Fbusiness result and explicit non-goals.\"},{\"label\":\"Evidence and constraints\",\"description\":\"Identify authoritative data, policies, NFRs, risks and acceptance conditions.\"},{\"label\":\"System boundary\",\"description\":\"Map users, identities, applications, data, models\u002Fproviders, tools and external systems.\"},{\"label\":\"Architecture options\",\"description\":\"Compare patterns for retrieval, model access, orchestration, deployment, permissions, evaluation and observability.\"},{\"label\":\"Trade-off decisions\",\"description\":\"Select significant options and preserve the rationale, alternatives and consequences.\"},{\"label\":\"Implementation contracts\",\"description\":\"Turn decisions into APIs, schemas, permission rules, deployment definitions and engineering tasks.\"},{\"label\":\"Validation\",\"description\":\"Test the implemented system against the original functional and non-functional requirements.\"},{\"label\":\"Operational feedback\",\"description\":\"Use production evidence, incidents, quality metrics and cost\u002Fsecurity signals to trigger controlled change.\"}]}},{\"id\":\"h-edge\",\"type\":\"header\",\"data\":{\"text\":\"Edge cases and limits of the role\",\"level\":2}},{\"id\":\"p-edge-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"Some AI products are dominated by model training, scientific experimentation or specialized hardware. In those cases, model\u002Fdata science and ML systems architecture can become much deeper than the solution-level map shown here. The AI Solution Architect still needs integration and operational boundaries, but specialist architecture may own the training platform itself.\"}},{\"id\":\"p-edge-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"At the other extreme, a simple SaaS integration may not justify a dedicated architect. A senior engineer or technical product lead can carry the same architecture responsibility. The useful test is not the title but whether significant cross-layer decisions are being made deliberately and validated.\"}},{\"id\":\"p-edge-3\",\"type\":\"paragraph\",\"data\":{\"text\":\"Regulated, sovereign, air-gapped, safety-critical, highly autonomous or multi-tenant systems also shift the center of gravity. Identity, isolation, residency, assurance, update mechanisms, human oversight and auditability may dominate model quality in the architecture.\"}},{\"id\":\"h-change-answer\",\"type\":\"header\",\"data\":{\"text\":\"What would change this answer?\",\"level\":2}},{\"id\":\"p-change-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"The exact responsibility boundary changes when architecture moves from one application to a reusable platform or to enterprise-wide target architecture. That is why \u003Cstrong>AI Platform Architect\u003C\u002Fstrong> and \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> deserve separate canonical treatment rather than being merged into this role.\"}},{\"id\":\"p-change-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"Technology changes also matter. New model capabilities, protocols, local runtimes and managed services can remove some implementation work while creating new trust or operational boundaries. The stable responsibility is to understand those changes as system changes—not to treat a new framework as a replacement for architecture.\"}},{\"id\":\"h-checklist\",\"type\":\"header\",\"data\":{\"text\":\"AI Solution Architect checklist\",\"level\":2}},{\"id\":\"checklist-table\",\"type\":\"table\",\"data\":{\"withHeadings\":true,\"stretched\":false,\"content\":[[\"Check\",\"Question\"],[\"Outcome\",\"Is the user\u002Fbusiness result and non-goal boundary explicit?\"],[\"Requirements\",\"Are functional requirements, NFRs, constraints and acceptance criteria traceable?\"],[\"Data\",\"Are authoritative sources, provenance, freshness, retention and access rules defined?\"],[\"Retrieval\u002Fcontext\",\"Does authorization reach retrieval and context construction?\"],[\"Model\u002Fprovider\",\"Is model\u002Fprovider selection tied to capabilities and constraints rather than preference?\"],[\"Tools\u002Fagents\",\"Are action boundaries, permissions, approvals and failure behavior explicit?\"],[\"Identity\u002Fsecurity\",\"Are human\u002Fmachine identities, secrets and trust boundaries defined?\"],[\"Runtime\",\"Are runtime, inference, data and control-plane locations distinguished?\"],[\"Evaluation\",\"Is there measurable evidence for quality, security and acceptance?\"],[\"Observability\",\"Can production behavior, failures, cost and security events be investigated?\"],[\"Change\",\"Are significant architecture decisions and replacements traceable?\"],[\"Operations\",\"Is ownership for deployment, rollback, incidents and lifecycle clear?\"]]}},{\"id\":\"h-conclusion\",\"type\":\"header\",\"data\":{\"text\":\"Conclusion\",\"level\":2}},{\"id\":\"p-conclusion-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"An AI Solution Architect is the person or architecture function that turns an AI opportunity into a coherent technical system. The key skill is not knowing the most model names; it is connecting product need, requirements, data, application architecture, AI capabilities, security, runtime, delivery and validation without losing the boundaries between them.\"}},{\"id\":\"p-conclusion-2\",\"type\":\"paragraph\",\"data\":{\"text\":\"A strong AI solution architecture can therefore be summarized as: \u003Cstrong>define the target → establish requirements and constraints → design the system boundaries → make significant trade-offs explicit → implement through clear contracts → validate against evidence → operate and evolve deliberately.\u003C\u002Fstrong> The model is important. The solution is the product.\"}},{\"id\":\"faq\",\"type\":\"faq\",\"data\":{\"title\":\"AI Solution Architect — FAQ\",\"items\":[{\"id\":\"faq1\",\"question\":\"What is an AI Solution Architect?\",\"answer\":\"An AI Solution Architect translates a business or product need into the architecture of a concrete AI-enabled solution, defining how application logic, data\u002Fretrieval, models, tools, identity, security, runtime, evaluation and operations work together.\"},{\"id\":\"faq2\",\"question\":\"Is an AI Solution Architect the same as an AI engineer?\",\"answer\":\"No. The roles can overlap, especially in small teams, but an AI engineer is primarily an implementation role while the solution architect owns or coordinates cross-layer architecture decisions and trade-offs for the complete workload.\"},{\"id\":\"faq3\",\"question\":\"Does an AI Solution Architect need to code?\",\"answer\":\"Not by definition, but hands-on implementation knowledge is highly valuable because AI architecture crosses APIs, data, retrieval, security, runtimes and operational behavior. The role is defined by architecture responsibility, not by writing every component personally.\"},{\"id\":\"faq4\",\"question\":\"Is choosing an LLM the main job?\",\"answer\":\"No. Model selection is one decision. Production architecture also needs data and retrieval boundaries, permissions, tools, provider\u002Fruntime choices, observability, evaluation, reliability, cost and lifecycle design.\"},{\"id\":\"faq5\",\"question\":\"What is the difference between an AI Solution Architect and an AI Platform Architect?\",\"answer\":\"An AI Solution Architect focuses on one concrete solution or workload. An AI Platform Architect focuses on reusable AI capabilities and guardrails that support multiple solutions.\"},{\"id\":\"faq6\",\"question\":\"What is the difference between an AI Solution Architect and an Enterprise AI Architect?\",\"answer\":\"The solution architect works at application\u002Fworkload scope. Enterprise AI architecture works across the organizational portfolio, target architecture, governance, shared capabilities, integration principles and strategic constraints.\"},{\"id\":\"faq7\",\"question\":\"Where do RAG and agents fit?\",\"answer\":\"They are architectural patterns or subsystems inside a solution when the requirements justify them. RAG addresses retrieval-grounded context; agents add planning\u002Ftool execution and therefore additional identity, permission, orchestration and operational concerns.\"},{\"id\":\"faq8\",\"question\":\"What proves that the architecture works?\",\"answer\":\"Implementation plus validation evidence: functional tests, evaluation results, security\u002Fauthorization tests, performance and reliability measurements, observability, operational rehearsal and acceptance against the original requirements.\"}]}},{\"id\":\"glossary\",\"type\":\"glossary\",\"data\":{\"title\":\"Core terms\",\"entries\":[{\"term\":\"AI Solution Architect\",\"definition\":\"Architecture responsibility for one concrete AI-enabled solution or workload, integrating product requirements with application, data, model, tool, security, runtime and operational design.\",\"anchor\":\"ai-solution-architect\"},{\"term\":\"System boundary\",\"definition\":\"The explicit separation between what belongs to the solution and the users, systems, providers, data sources and environments it interacts with.\",\"anchor\":\"system-boundary\"},{\"term\":\"Trust boundary\",\"definition\":\"A point where data, identities or control cross between components with different trust assumptions and therefore require explicit security controls.\",\"anchor\":\"trust-boundary\"},{\"term\":\"Grounding\",\"definition\":\"Supplying an AI model with relevant external information or evidence so its response can be based on sources beyond model parameters.\",\"anchor\":\"grounding\"},{\"term\":\"Provider abstraction\",\"definition\":\"An application boundary that decouples parts of the solution from one model\u002Fprovider interface. Useful when justified by routing, portability or policy needs, but not free of trade-offs.\",\"anchor\":\"provider-abstraction\"},{\"term\":\"Evaluation\",\"definition\":\"Structured measurement of AI workload behavior against defined acceptance criteria, including task quality and relevant safety, security, performance and operational properties.\",\"anchor\":\"evaluation\"},{\"term\":\"AI Platform Architect\",\"definition\":\"Architectural role focused on reusable AI platform capabilities used by multiple solutions rather than the architecture of one workload.\",\"anchor\":\"ai-platform-architect\"},{\"term\":\"Enterprise AI Architecture\",\"definition\":\"Organization-level architecture that coordinates AI capabilities, platforms, governance, integration and strategic constraints across a portfolio.\",\"anchor\":\"enterprise-ai-architecture\"}]}},{\"id\":\"h-related\",\"type\":\"header\",\"data\":{\"text\":\"Related canonical knowledge\",\"level\":2}},{\"id\":\"p-related-1\",\"type\":\"paragraph\",\"data\":{\"text\":\"This article sits in the AI Architecture Foundations cluster. Its direct foundations are \u003Cstrong>Generative AI Explained: Models, Retrieval, Tools and Applications Are Not the Same Thing\u003C\u002Fstrong> and \u003Cstrong>ADR vs NFR: Architecture Decisions and System Quality Are Not the Same Thing\u003C\u002Fstrong>. Adjacent canonical nodes include \u003Cstrong>Agentic AI Explained\u003C\u002Fstrong>, \u003Cstrong>Source of Truth in AI Systems\u003C\u002Fstrong>, \u003Cstrong>Vector Databases, Embeddings and Reranking\u003C\u002Fstrong>, \u003Cstrong>What Is Context Engineering?\u003C\u002Fstrong>, \u003Cstrong>RBAC vs Tenant Isolation\u003C\u002Fstrong>, \u003Cstrong>AI Platform Architect\u003C\u002Fstrong>, \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> and \u003Cstrong>AI Governance\u003C\u002Fstrong>. URLs are intentionally not fabricated where those nodes are not yet published.\"}},{\"id\":\"related-rag\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works\",\"meta\":{\"title\":\"What Is RAG? The Simplest Explanation of How It Works\",\"description\":\"Existing stajic.de canonical explanation of retrieval-augmented generation, useful for the retrieval\u002Fgrounding part of AI solution architecture.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"h-sources\",\"type\":\"header\",\"data\":{\"text\":\"Primary sources and current architecture guidance\",\"level\":2}},{\"id\":\"p-sources-note\",\"type\":\"paragraph\",\"data\":{\"text\":\"External sources below support the general architecture claims; the SenseFlow and Aaasaasa AI Client sections are explicitly original project\u002Fimplementation evidence. Current-state references were checked on 8 October 2026. NIST notes that AI RMF 1.0 is being revised, so version-sensitive governance references should be rechecked when a successor is published.\"}},{\"id\":\"src-iso-42010\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fwww.iso.org\u002Fstandard\u002F74393.html\",\"meta\":{\"title\":\"ISO\u002FIEC\u002FIEEE 42010:2022 — Architecture Description\",\"description\":\"Current international standard for the structure and expression of architecture descriptions. It distinguishes architecture from its description and does not prescribe one architecting method, tool or recording format.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-nist-rmf\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fwww.nist.gov\u002Fitl\u002Fai-risk-management-framework\",\"meta\":{\"title\":\"NIST AI Risk Management Framework\",\"description\":\"NIST’s AI RMF resource page. As of October 2026 it states that AI RMF 1.0 is being revised and links the Generative AI Profile and related resources.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-nist-core\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fairc.nist.gov\u002Fairmf-resources\u002Fairmf\u002F5-sec-core\u002F\",\"meta\":{\"title\":\"NIST AI RMF Core — Govern, Map, Measure, Manage\",\"description\":\"Official NIST AIRC presentation of the AI RMF 1.0 Core, including the four functions and lifecycle-oriented risk-management framing.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-nist-gai\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fwww.nist.gov\u002Fpublications\u002Fartificial-intelligence-risk-management-framework-generative-artificial-intelligence\",\"meta\":{\"title\":\"NIST AI 600-1 — Generative AI Profile\",\"description\":\"Cross-sectoral Generative AI profile for AI RMF 1.0, published 26 July 2024 and updated by NIST in 2026.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-ms-start\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fget-started\",\"meta\":{\"title\":\"Microsoft Azure Well-Architected — AI Workloads\",\"description\":\"Current workload-level architecture guidance covering AI application design, application platform, training data, grounding data, data platform and production-readiness concerns.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-ms-app\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fapplication-design\",\"meta\":{\"title\":\"Microsoft — Application Design for AI Workloads\",\"description\":\"Guidance on model\u002Ftool abstraction, data-access boundaries, identity propagation, authorization and separation of client, intelligence, knowledge and tool layers.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-ms-security\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fdesign-principles\",\"meta\":{\"title\":\"Microsoft — Design Principles for AI Workloads\",\"description\":\"Current AI workload design principles across reliability, security, cost, operational excellence and performance, including identity and data-protection responsibilities.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-ms-ops\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fwell-architected\u002Fai\u002Fmlops-genaiops\",\"meta\":{\"title\":\"Microsoft — MLOps and GenAIOps for AI Workloads\",\"description\":\"Production lifecycle guidance covering monitoring, quality gates, model\u002Fprompt behavior, security and operational measurement.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-aws-genai\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fdocs.aws.amazon.com\u002Fwellarchitected\u002Flatest\u002Fgenerative-ai-lens\u002F\",\"meta\":{\"title\":\"AWS Well-Architected Generative AI Lens\",\"description\":\"AWS architectural guidance for generative AI workloads across operational excellence, security, reliability, performance efficiency, cost optimization and sustainability.\",\"image\":{\"url\":\"\"}}}},{\"id\":\"src-aws-agentic\",\"type\":\"linkTool\",\"data\":{\"link\":\"https:\u002F\u002Fdocs.aws.amazon.com\u002Fwellarchitected\u002Flatest\u002Fagentic-ai-lens\u002F\",\"meta\":{\"title\":\"AWS Well-Architected Agentic AI Lens\",\"description\":\"Published in 2026, covering agentic-specific architecture concerns including identities, tools, orchestration, human oversight, reliability, tracing and reasoning-loop cost.\",\"image\":{\"url\":\"\"}}}}],\"version\":\"2.31.0\"}",{"time":1113,"blocks":1114,"version":1786},1791476244367,[1115,1118,1122,1126,1130,1133,1136,1139,1142,1166,1169,1172,1175,1200,1203,1206,1209,1212,1215,1262,1265,1268,1271,1274,1277,1280,1283,1286,1289,1292,1295,1298,1301,1304,1307,1310,1313,1316,1319,1322,1325,1328,1331,1361,1364,1367,1410,1413,1416,1437,1440,1443,1447,1450,1453,1456,1459,1462,1465,1468,1471,1474,1477,1480,1483,1486,1513,1516,1555,1558,1586,1589,1592,1595,1598,1601,1604,1607,1610,1648,1651,1654,1657,1685,1707,1710,1713,1720,1723,1726,1732,1738,1744,1750,1756,1762,1768,1774,1780],{"id":215,"data":1116,"type":218},{"text":1117},"An \u003Cstrong>AI Solution Architect\u003C\u002Fstrong> translates a business or product need into the architecture of a concrete AI-enabled solution. The role defines system boundaries and the significant choices across application logic, authoritative data, retrieval and context, models and providers, tools or agents, identity and permissions, security, runtime and deployment, observability, evaluation, cost and operational behavior. It is not simply model selection or prompt engineering: the architectural responsibility is to make the whole solution implementable, governable, testable and operable.",{"id":220,"data":1119,"type":225},{"body":1120,"title":1121,"variant":224},"\u003Cstrong>An AI Solution Architect designs the complete AI-enabled solution, not just the AI model.\u003C\u002Fstrong> The role connects requirements and non-functional requirements to architecture decisions, composes the necessary application\u002Fdata\u002Fmodel\u002Ftool\u002Fruntime layers, makes trust and failure boundaries explicit, and defines how the implemented system will be validated and operated.","Direct answer",{"id":227,"data":1123,"type":225},{"body":1124,"title":1125,"variant":231},"\u003Cstrong>AI Solution Architect is a practical role label, not a universally standardized job title.\u003C\u002Fstrong> ISO\u002FIEC\u002FIEEE 42010:2022 standardizes concepts for architecture descriptions; it does not define this job role. Organizations can distribute the responsibilities across several people. In this article, the term means the architecture responsibility for one concrete AI-enabled solution or workload.","Terminology note",{"id":233,"data":1127,"type":225},{"body":1128,"title":1129,"variant":231},"The architecture principles here are intentionally vendor-neutral, while current vendor guidance is used as implementation evidence. NIST AI RMF 1.0 is currently under revision; NIST AI 600-1 remains the published Generative AI Profile. Microsoft and AWS guidance cited below reflects current production concerns such as identity, data boundaries, model abstraction, security, observability, evaluation, reliability and cost.","Current-source note — 8 October 2026",{"id":238,"data":1131,"type":243},{"title":1132,"maxLevel":241,"minLevel":242},"Contents",{"id":245,"data":1134,"type":42},{"text":1135,"level":242},"What does an AI Solution Architect actually architect?",{"id":249,"data":1137,"type":218},{"text":1138},"The object of the work is the \u003Cstrong>solution\u003C\u002Fstrong>: the complete socio-technical system that turns a need into useful, controlled behavior. A model may be central to that system, but it is still only one dependency. The same model can participate in a safe internal search assistant, an unsafe over-privileged agent, a low-latency customer feature, or a high-cost prototype that cannot be operated economically. Architecture determines those differences.",{"id":253,"data":1140,"type":218},{"text":1141},"A useful boundary is therefore: \u003Cstrong>business outcome → requirements → system responsibilities → architecture decisions → implementation → validation → operation\u003C\u002Fstrong>. The AI Solution Architect works across this chain while collaborating with product, engineering, data, security, infrastructure, governance and domain specialists.",{"id":257,"data":1143,"type":299},{"rows":1144,"title":1160,"layout":291,"columns":1161},[1145,1148,1151,1154,1157],{"id":261,"label":1146,"values":1147},"Capability",{"model":264,"solution":265},{"id":267,"label":1149,"values":1150},"Data",{"model":270,"solution":271},{"id":273,"label":1152,"values":1153},"Security",{"model":276,"solution":277},{"id":279,"label":1155,"values":1156},"Operations",{"model":282,"solution":283},{"id":285,"label":1158,"values":1159},"Change",{"model":288,"solution":289},"The solution is wider than the model",[1162,1164],{"id":294,"label":1163},"Model-centric question",{"id":297,"label":1165},"Solution-architecture question",{"id":301,"data":1167,"type":42},{"text":1168,"level":242},"The simplest example",{"id":305,"data":1170,"type":218},{"text":1171},"Imagine a company wants an internal assistant that answers technicians’ questions from maintenance manuals and operating procedures. The visible feature sounds simple: type a question and receive an answer with sources.",{"id":309,"data":1173,"type":218},{"text":1174},"The architecture question is much larger. Which documents are authoritative? How are users authenticated? Must retrieval respect department or site permissions? Is the answer allowed to use only retrieved evidence? Which model is acceptable for the data classification? Can a cloud provider receive the content? What happens when retrieval finds nothing? How are citations produced? How is answer quality evaluated? What latency and cost are acceptable? Who can see logs, and what may be stored in them?",{"id":313,"data":1176,"type":339},{"steps":1177,"title":1199,"orientation":338},[1178,1181,1184,1187,1190,1193,1196],{"label":1179,"description":1180},"1. Define the outcome","Clarify the user, business value, task boundary and what a successful answer or action means.",{"label":1182,"description":1183},"2. Capture requirements","Make functional requirements, NFRs, constraints, data rules, risk tolerance and acceptance criteria explicit.",{"label":1185,"description":1186},"3. Establish boundaries","Identify users, identities, applications, authoritative data, model\u002Fprovider dependencies, tools, external systems and trust zones.",{"label":1188,"description":1189},"4. Design the architecture","Choose data\u002Fretrieval, model, orchestration, tool, permission, runtime, deployment, fallback and observability patterns.",{"label":1191,"description":1192},"5. Record significant decisions","Preserve architectural choices, alternatives, trade-offs and consequences so later changes remain understandable.",{"label":1194,"description":1195},"6. Implement and integrate","Turn the architecture into application code, APIs, policies, infrastructure, workflows and operational controls.",{"label":1197,"description":1198},"7. Validate and operate","Test quality, security, reliability, cost and user outcomes; monitor the real workload and feed evidence back into decisions.","From need to an operable AI solution",{"id":341,"data":1201,"type":42},{"text":1202,"level":242},"Where the simple example stops",{"id":345,"data":1204,"type":218},{"text":1205},"A proof of concept can often skip architecture that production cannot. A developer may hard-code one provider, use a shared API key, place all documents in one index, run retrieval without user-context filtering, log prompts verbatim and judge quality manually. That can demonstrate feasibility, but it does not establish a production architecture.",{"id":349,"data":1207,"type":218},{"text":1208},"Production introduces constraints that interact: tenant or user isolation, privacy, data residency, throughput, latency, cost, provider quotas, fallback behavior, auditability, model version changes, retrieval quality, tool permissions, incident response and deployment lifecycle. The architect’s job is not to maximize every quality at once; it is to make the trade-offs explicit and design a solution that satisfies the actual priority set.",{"id":353,"data":1210,"type":42},{"text":1211,"level":242},"Architecture responsibility map",{"id":357,"data":1213,"type":218},{"text":1214},"The exact split varies by organization, but the following map captures the recurring responsibilities of solution-level AI architecture. The architect may not personally implement every layer; the responsibility is to make the layers fit together coherently and to keep the critical decisions traceable.",{"id":361,"data":1216,"type":291},{"content":1217,"stretched":43,"withHeadings":14},[1218,1222,1226,1230,1234,1238,1242,1246,1250,1254,1258],[1219,1220,1221],"Architecture area","Questions the AI Solution Architect must resolve","Typical outputs",[1223,1224,1225],"Outcome and scope","Who is the user? What task is in scope? What must the system not do? What constitutes success?","Solution context, capability boundary, acceptance criteria",[1227,1228,1229],"Requirements and NFRs","What quality, security, availability, latency, cost, residency and compliance constraints apply?","Requirement map, NFRs, constraints, validation criteria",[1231,1232,1233],"Application and orchestration","Where does deterministic application logic end and AI behavior begin? How are workflows coordinated?","Component model, APIs, orchestration boundaries, failure paths",[1235,1236,1237],"Authoritative data and retrieval","What is the Source of Truth? How is data ingested, authorized, retrieved, filtered, ranked and cited?","Data flows, retrieval architecture, metadata and authorization rules",[1239,1240,1241],"Model and provider layer","Which capabilities are required? Which provider\u002Fruntime constraints matter? What should be abstracted?","Model\u002Fprovider decision, routing\u002Ffallback policy, abstraction boundary",[1243,1244,1245],"Tools and agents","What actions can the system take? Which actions require approval? How are tool identities and permissions enforced?","Tool contracts, agent boundaries, approval and least-privilege rules",[1247,1248,1249],"Identity and security","Which human and machine identities exist? Where are secrets held? Which trust boundaries are crossed?","Threat\u002Ftrust boundary model, identity propagation, secrets and authorization design",[1251,1252,1253],"Runtime and deployment","Where do components execute? What is local, cloud, edge or hybrid? What network and availability assumptions exist?","Deployment view, runtime topology, environment and connectivity decisions",[1255,1256,1257],"Evaluation and observability","How is quality measured before and after release? What traces, metrics, logs and evidence are needed?","Evaluation plan, telemetry, audit trail, release gates",[1259,1260,1261],"Operations and change","How are models\u002Fprompts\u002Fconfiguration\u002Fdata versions changed, rolled back and supported?","Operational model, lifecycle controls, ADRs, runbooks, change rules",{"id":409,"data":1263,"type":42},{"text":1264,"level":241},"1. Turn product need into architectural requirements",{"id":413,"data":1266,"type":218},{"text":1267},"AI architecture begins before model selection. The architect first determines what the solution is expected to achieve and under which constraints. This includes functional behavior, but also the NFRs and policies that narrow the design space: security, reliability, latency, privacy, residency, maintainability, cost and operational support.",{"id":417,"data":1269,"type":218},{"text":1270},"This is where A02’s distinction matters: a requirement such as “unauthorized users must not retrieve restricted documents” is not an architecture decision. It is a driver. Decisions about identity propagation, index partitioning, metadata filtering, API boundaries and authorization enforcement are architectural responses that must later be validated.",{"id":421,"data":1272,"type":42},{"text":1273,"level":241},"2. Design authoritative data, retrieval and context",{"id":425,"data":1275,"type":218},{"text":1276},"AI systems often fail at the boundary between model behavior and enterprise truth. An architect must define which sources are authoritative, what freshness and provenance mean, how access control reaches retrieval, and how retrieved evidence becomes model context. A vector database, embedding model or RAG library is not the architecture by itself.",{"id":429,"data":1278,"type":218},{"text":1279},"Microsoft’s current AI workload guidance makes the same separation explicit: application code should not bypass data-access boundaries; user or tenant context should propagate into retrieval and filtering; grounding data must be designed for searchability while still meeting security and compliance requirements.",{"id":433,"data":1281,"type":42},{"text":1282,"level":241},"3. Treat models and providers as dependencies, not the whole system",{"id":437,"data":1284,"type":218},{"text":1285},"Model selection matters, but it should be driven by required capability and constraints. The architect considers reasoning or generation quality, modality, context limits, latency, data handling, deployment location, provider availability, cost, observability and replacement risk.",{"id":441,"data":1287,"type":218},{"text":1288},"Provider abstraction is not automatically “better architecture.” It adds engineering cost and can hide provider-specific capabilities. It is justified when portability, fallback, policy separation or multi-provider routing is an explicit requirement. Otherwise a direct integration can be the better decision. The point is to make the trade-off intentional.",{"id":445,"data":1290,"type":42},{"text":1291,"level":241},"4. Architect tools, actions and agent boundaries",{"id":449,"data":1293,"type":218},{"text":1294},"When an AI system can call tools, modify data, send messages, run code or operate business systems, the architectural risk changes. Tool access needs its own identity and authorization model. The model’s ability to request an action is not the same as permission to execute it.",{"id":453,"data":1296,"type":218},{"text":1297},"For agentic workloads, current AWS guidance emphasizes additional dimensions such as agent identities, tool access, orchestration, human oversight, tracing, failure handling and cost of iterative reasoning loops. These are solution concerns even when a framework hides some of the implementation mechanics.",{"id":457,"data":1299,"type":42},{"text":1300,"level":241},"5. Make trust boundaries and permissions explicit",{"id":461,"data":1302,"type":218},{"text":1303},"A production AI solution has multiple trust boundaries: browser or client, application backend, AI orchestration, retrieval\u002Fdata services, model providers, tool APIs, local runtimes and external systems. Each boundary should answer: who is calling, on whose behalf, with what credential, for which resource, with what audit trail, and with what failure containment?",{"id":465,"data":1305,"type":218},{"text":1306},"Security cannot be deferred to a “guardrail” around the model. Microsoft’s AI workload guidance explicitly places security across all architecture layers and calls for identity\u002Faccess management, data protection, content controls and lifecycle security. NIST likewise treats governance and risk management as continuous across the AI lifecycle.",{"id":469,"data":1308,"type":42},{"text":1309,"level":241},"6. Decide where the system actually runs",{"id":473,"data":1311,"type":218},{"text":1312},"“Local AI,” “cloud AI,” and “hybrid AI” are architectural statements only when the execution and data paths are precise. A local desktop process can still call a cloud model. A cloud-hosted application can retrieve from an on-premises data source. An air-gapped solution has entirely different update, model-distribution and observability constraints.",{"id":477,"data":1314,"type":218},{"text":1315},"The architect therefore separates \u003Cstrong>runtime location\u003C\u002Fstrong>, \u003Cstrong>inference location\u003C\u002Fstrong>, \u003Cstrong>data location\u003C\u002Fstrong> and \u003Cstrong>control plane\u003C\u002Fstrong>. Conflating them creates false security and deployment assumptions.",{"id":481,"data":1317,"type":42},{"text":1318,"level":241},"7. Define evaluation, observability and operational acceptance",{"id":485,"data":1320,"type":218},{"text":1321},"AI behavior is partly nondeterministic, so the release definition cannot rely only on conventional unit tests. The architecture needs measurable acceptance: task success, groundedness or citation correctness where relevant, refusal behavior, tool safety, latency, cost, reliability and security tests. The exact metrics depend on the use case.",{"id":489,"data":1323,"type":218},{"text":1324},"Microsoft’s current Well-Architected AI guidance treats monitoring as continuous and applies it across model behavior, prompts\u002Fcompletions, anomalies, security and production quality gates. AWS similarly treats observability, lifecycle management and model\u002Fprompt traceability as operational architecture concerns.",{"id":493,"data":1326,"type":42},{"text":1327,"level":242},"What should the role produce?",{"id":497,"data":1329,"type":218},{"text":1330},"Architecture is not the slide deck. The useful outputs are the artifacts that let engineering, security, product and operations make consistent decisions and later understand why the system exists in its current form.",{"id":501,"data":1332,"type":291},{"content":1333,"stretched":43,"withHeadings":14},[1334,1337,1340,1343,1346,1349,1352,1355,1358],[1335,1336],"Artifact","Purpose",[1338,1339],"Solution context and boundary","Shows users, external systems, major responsibilities and what is outside scope",[1341,1342],"Requirement\u002FNFR map","Connects product need and constraints to architecture work and validation",[1344,1345],"Component and data-flow views","Shows application, data\u002Fretrieval, model, tools, identity and runtime interactions",[1347,1348],"Trust and permission model","Makes identities, secrets, authorization, sensitive data and high-risk actions explicit",[1350,1351],"Architecture Decision Records","Preserves significant choices, alternatives, trade-offs, status and consequences",[1353,1354],"Evaluation and acceptance plan","Defines evidence required to claim that the solution meets quality and safety expectations",[1356,1357],"Deployment and operational view","Defines environments, runtime locations, observability, rollback, incident and lifecycle responsibilities",[1359,1360],"Traceability links","Connects requirements, decisions, implementation work, tests and operational evidence",{"id":532,"data":1362,"type":42},{"text":1363,"level":242},"The work is mostly trade-offs, not “best practice” selection",{"id":536,"data":1365,"type":218},{"text":1366},"Architecture exists because desirable qualities conflict. A lower-cost model may reduce quality. A more capable model may increase latency or data-governance constraints. Aggressive caching can improve cost and speed while complicating freshness. More autonomous agents can reduce human effort while increasing blast radius and audit requirements.",{"id":540,"data":1368,"type":291},{"content":1369,"stretched":43,"withHeadings":14},[1370,1375,1380,1385,1390,1395,1400,1405],[1371,1372,1373,1374],"Decision","Potential benefit","Potential cost \u002F risk","Architectural question",[1376,1377,1378,1379],"Managed cloud model","Fast adoption, strong managed capabilities","External dependency, data and cost constraints","Does the workload permit the provider\u002Fdata path and meet resilience needs?",[1381,1382,1383,1384],"Local\u002Fself-hosted inference","Control, offline\u002Fprivate options","Hardware, operations, model lifecycle burden","Is the control benefit worth the operational responsibility?",[1386,1387,1388,1389],"Single provider integration","Simpler implementation, full provider features","Higher switching\u002Ffailure concentration","Is portability or fallback actually required?",[1391,1392,1393,1394],"Provider abstraction","Portability, routing and policy separation","Lowest-common-denominator risk, more code\u002Ftests","Which differences must remain visible rather than abstracted?",[1396,1397,1398,1399],"Large context","More information per request","Latency, cost, attention dilution, leakage surface","Should data be retrieved\u002Ffiltered instead of always injected?",[1401,1402,1403,1404],"Powerful tools \u002F autonomy","More end-to-end automation","Higher privilege and failure blast radius","Which actions require least privilege, confirmation or human approval?",[1406,1407,1408,1409],"Strict validation and logging","Better evidence and operations","Latency, storage, privacy and complexity cost","What evidence is required for this risk level?",{"id":584,"data":1411,"type":42},{"text":1412,"level":242},"How is this different from adjacent roles?",{"id":588,"data":1414,"type":218},{"text":1415},"Titles overlap heavily across companies. The useful distinction is the \u003Cstrong>scope of architecture responsibility\u003C\u002Fstrong>, not the HR label.",{"id":592,"data":1417,"type":299},{"rows":1418,"title":1431,"layout":291,"columns":1432},[1419,1421,1423,1425,1427,1429],{"id":596,"label":597,"values":1420},{"role":599,"focus":600},{"id":602,"label":603,"values":1422},{"role":605,"focus":606},{"id":608,"label":609,"values":1424},{"role":611,"focus":612},{"id":614,"label":615,"values":1426},{"role":617,"focus":618},{"id":620,"label":621,"values":1428},{"role":623,"focus":624},{"id":626,"label":627,"values":1430},{"role":629,"focus":630},"Adjacent roles answer different primary questions",[1433,1435],{"id":634,"label":1434},"Role",{"id":637,"label":1436},"Primary architecture focus",{"id":640,"data":1438,"type":218},{"text":1439},"In a small product team, one person may cover several of these scopes. In a large enterprise, they may be separate roles with formal review boards. The architecture responsibility does not disappear when the title changes.",{"id":644,"data":1441,"type":42},{"text":1442,"level":242},"Implementation evidence: how these boundaries appear in my own work",{"id":648,"data":1444,"type":225},{"body":1445,"title":1446,"variant":652},"The examples below are \u003Cstrong>original implementation\u002Fproject evidence\u003C\u002Fstrong>. They show how I have separated product need, requirements, architecture, runtime, model\u002Fprovider, permissions and validation in real project work. They are not claims that every organization must use the same structure, and they do not imply customer adoption or enterprise-scale deployment.","Implementation evidence, not a universal rule",{"id":654,"data":1448,"type":42},{"text":1449,"level":241},"SenseFlow: need → requirements → architecture → validation",{"id":658,"data":1451,"type":218},{"text":1452},"In the SenseFlow project Source of Truth, technology is explicitly subordinate to Product Vision. The development structure moves from problem and product vision through user needs, value, scope, epics, stories and acceptance criteria into architecture, implementation, validation and iteration.",{"id":662,"data":1454,"type":218},{"text":1455},"Requirements are designed to be traceable from Product Goal → Capability → Epic → User Story → Acceptance Criteria → Technical Tasks. Where practical, they include functional requirements, NFRs, dependencies, risks, assumptions, acceptance criteria and validation methods. Significant decisions preserve the decision, reason, alternatives, trade-offs, status and date\u002Fversion.",{"id":666,"data":1457,"type":218},{"text":1458},"That is architectural work before a specific AI framework or model is chosen: it protects the connection between product intent and technical decisions and makes later change reviewable rather than implicit.",{"id":670,"data":1460,"type":42},{"text":1461,"level":241},"Aaasaasa AI Client: separate concepts before integrating them",{"id":674,"data":1463,"type":218},{"text":1464},"Aaasaasa AI Client provides a more implementation-level example. Its AI Hub deliberately separates \u003Cstrong>agent\u002Fclient\u003C\u002Fstrong>, \u003Cstrong>provider\u003C\u002Fstrong>, \u003Cstrong>model\u003C\u002Fstrong>, \u003Cstrong>connection\u002Fruntime location\u003C\u002Fstrong>, \u003Cstrong>permissions\u003C\u002Fstrong> and \u003Cstrong>web client\u003C\u002Fstrong>. A local runtime is not assumed to mean local inference, and permissions are treated as runtime\u002Ftool policy rather than as a property of the model.",{"id":678,"data":1466,"type":218},{"text":1467},"The desktop architecture also defines a trust boundary: the Nuxt renderer is untrusted relative to Electron main. A narrow preload and validated IPC mediate access to AI services, settings, encrypted secrets, workspace\u002Fdata services and runtimes. Cloud credentials remain in the privileged main process; renderer code receives normalized state instead of raw secrets or unrestricted operating-system access.",{"id":682,"data":1469,"type":218},{"text":1470},"Routing decisions are similarly architectural. The implementation does not silently fall back from a local route to paid cloud inference; a cloud route requires explicit confirmation. Direct Chat has no filesystem or shell tools by default, while agent execution applies a selected workspace and permission profile. These are solution-level decisions about trust, cost, execution and user expectation—not model features.",{"id":686,"data":1472,"type":42},{"text":1473,"level":242},"How current architecture frameworks support this broader scope",{"id":690,"data":1475,"type":218},{"text":1476},"ISO\u002FIEC\u002FIEEE 42010:2022 provides a general discipline for architecture descriptions across software, systems and enterprises. It is deliberately broader than AI and does not prescribe one architecting method or job title. That makes it useful here as a boundary: AI solution architecture is still architecture, with stakeholder concerns, multiple views and significant relationships that must be expressed clearly.",{"id":694,"data":1478,"type":218},{"text":1479},"NIST AI RMF 1.0 frames AI risk management through \u003Cstrong>Govern, Map, Measure and Manage\u003C\u002Fstrong> and emphasizes that risk management should be continuous across the AI system lifecycle. The Generative AI Profile (NIST AI 600-1) adapts that framework to GAI risks and organizational priorities. This reinforces that architecture cannot stop at functional model performance.",{"id":698,"data":1481,"type":218},{"text":1482},"Microsoft’s current Azure Well-Architected AI guidance separates application design, application platform, training data, grounding data and data platform concerns and repeatedly connects them to reliability, security, operational excellence, performance and cost. AWS’s Generative AI and Agentic AI lenses similarly treat observability, security, reliability, model\u002Ftool lifecycle, cost and human oversight as architecture concerns.",{"id":702,"data":1484,"type":42},{"text":1485,"level":242},"Common misconceptions",{"id":706,"data":1487,"type":291},{"content":1488,"stretched":43,"withHeadings":14},[1489,1492,1495,1498,1501,1504,1507,1510],[1490,1491],"Misconception","Correction",[1493,1494],"“The architect chooses the LLM.”","Model choice is one decision inside a larger solution architecture.",[1496,1497],"“Prompt engineering is the architecture.”","Prompts affect behavior, but they do not define identity, data access, trust boundaries, deployment, tool permissions or operations.",[1499,1500],"“RAG solves enterprise knowledge.”","Retrieval is only one subsystem; authorization, provenance, freshness, evidence, indexing, evaluation and source governance still need design.",[1502,1503],"“Local runtime means private\u002Flocal AI.”","Runtime, inference, data and control-plane locations are separate architectural properties.",[1505,1506],"“If a vendor offers guardrails, security is covered.”","Security spans identity, authorization, secrets, data flows, tools, logging, deployment, human approval and provider boundaries.",[1508,1509],"“The architect must write every component.”","Hands-on implementation can improve architectural quality, but the role is defined by integrated decision responsibility, not by personally coding every layer.",[1511,1512],"“An architecture diagram proves production readiness.”","Readiness requires implemented controls and validation evidence across quality, security, operations and business acceptance.",{"id":734,"data":1514,"type":42},{"text":1515,"level":242},"Failure modes an AI Solution Architect should prevent",{"id":738,"data":1517,"type":291},{"content":1518,"stretched":43,"withHeadings":14},[1519,1523,1527,1531,1535,1539,1543,1547,1551],[1520,1521,1522],"Failure mode","Why it happens","Architectural correction",[1524,1525,1526],"Model-first design","A promising model demo becomes the system blueprint","Start from outcome, constraints and validation; select the model inside that frame",[1528,1529,1530],"Prototype permissions in production","Shared credentials and broad access survive the PoC","Define identity propagation, least privilege, tool scopes and approval boundaries early",[1532,1533,1534],"Retrieval without authorization","Search quality is designed before data-access rules","Carry user\u002Ftenant context into retrieval and enforce authorization at data-access boundaries",[1536,1537,1538],"Silent provider\u002Fruntime assumptions","“Local”, “cloud” and “offline” are used imprecisely","Document runtime, inference, data and control-plane location separately",[1540,1541,1542],"No failure contract","The happy path is designed but refusal\u002Ffallback\u002Ferror behavior is not","Specify retrieval-empty, model-unavailable, tool-failure and policy-denied behavior",[1544,1545,1546],"Evaluation after implementation","Quality is judged manually near launch","Define measurable acceptance and representative evaluation sets before architecture freezes",[1548,1549,1550],"Untraceable change","Models, prompts, retrieval or permissions change without architectural history","Version critical configuration and record significant decisions\u002Fvalidation evidence",[1552,1553,1554],"Operations treated as infrastructure only","AI behavior is not observable after deployment","Design traces, quality metrics, security events, cost telemetry and rollback together",{"id":778,"data":1556,"type":42},{"text":1557,"level":242},"A practical decision sequence",{"id":782,"data":1559,"type":339},{"steps":1560,"title":1585,"orientation":338},[1561,1564,1567,1570,1573,1576,1579,1582],{"label":1562,"description":1563},"Outcome","Define the user\u002Fbusiness result and explicit non-goals.",{"label":1565,"description":1566},"Evidence and constraints","Identify authoritative data, policies, NFRs, risks and acceptance conditions.",{"label":1568,"description":1569},"System boundary","Map users, identities, applications, data, models\u002Fproviders, tools and external systems.",{"label":1571,"description":1572},"Architecture options","Compare patterns for retrieval, model access, orchestration, deployment, permissions, evaluation and observability.",{"label":1574,"description":1575},"Trade-off decisions","Select significant options and preserve the rationale, alternatives and consequences.",{"label":1577,"description":1578},"Implementation contracts","Turn decisions into APIs, schemas, permission rules, deployment definitions and engineering tasks.",{"label":1580,"description":1581},"Validation","Test the implemented system against the original functional and non-functional requirements.",{"label":1583,"description":1584},"Operational feedback","Use production evidence, incidents, quality metrics and cost\u002Fsecurity signals to trigger controlled change.","AI solution architecture decision sequence",{"id":811,"data":1587,"type":42},{"text":1588,"level":242},"Edge cases and limits of the role",{"id":815,"data":1590,"type":218},{"text":1591},"Some AI products are dominated by model training, scientific experimentation or specialized hardware. In those cases, model\u002Fdata science and ML systems architecture can become much deeper than the solution-level map shown here. The AI Solution Architect still needs integration and operational boundaries, but specialist architecture may own the training platform itself.",{"id":819,"data":1593,"type":218},{"text":1594},"At the other extreme, a simple SaaS integration may not justify a dedicated architect. A senior engineer or technical product lead can carry the same architecture responsibility. The useful test is not the title but whether significant cross-layer decisions are being made deliberately and validated.",{"id":823,"data":1596,"type":218},{"text":1597},"Regulated, sovereign, air-gapped, safety-critical, highly autonomous or multi-tenant systems also shift the center of gravity. Identity, isolation, residency, assurance, update mechanisms, human oversight and auditability may dominate model quality in the architecture.",{"id":827,"data":1599,"type":42},{"text":1600,"level":242},"What would change this answer?",{"id":831,"data":1602,"type":218},{"text":1603},"The exact responsibility boundary changes when architecture moves from one application to a reusable platform or to enterprise-wide target architecture. That is why \u003Cstrong>AI Platform Architect\u003C\u002Fstrong> and \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> deserve separate canonical treatment rather than being merged into this role.",{"id":835,"data":1605,"type":218},{"text":1606},"Technology changes also matter. New model capabilities, protocols, local runtimes and managed services can remove some implementation work while creating new trust or operational boundaries. The stable responsibility is to understand those changes as system changes—not to treat a new framework as a replacement for architecture.",{"id":839,"data":1608,"type":42},{"text":1609,"level":242},"AI Solution Architect checklist",{"id":843,"data":1611,"type":291},{"content":1612,"stretched":43,"withHeadings":14},[1613,1616,1618,1621,1623,1626,1629,1632,1635,1638,1641,1644,1646],[1614,1615],"Check","Question",[1562,1617],"Is the user\u002Fbusiness result and non-goal boundary explicit?",[1619,1620],"Requirements","Are functional requirements, NFRs, constraints and acceptance criteria traceable?",[1149,1622],"Are authoritative sources, provenance, freshness, retention and access rules defined?",[1624,1625],"Retrieval\u002Fcontext","Does authorization reach retrieval and context construction?",[1627,1628],"Model\u002Fprovider","Is model\u002Fprovider selection tied to capabilities and constraints rather than preference?",[1630,1631],"Tools\u002Fagents","Are action boundaries, permissions, approvals and failure behavior explicit?",[1633,1634],"Identity\u002Fsecurity","Are human\u002Fmachine identities, secrets and trust boundaries defined?",[1636,1637],"Runtime","Are runtime, inference, data and control-plane locations distinguished?",[1639,1640],"Evaluation","Is there measurable evidence for quality, security and acceptance?",[1642,1643],"Observability","Can production behavior, failures, cost and security events be investigated?",[1158,1645],"Are significant architecture decisions and replacements traceable?",[1155,1647],"Is ownership for deployment, rollback, incidents and lifecycle clear?",{"id":882,"data":1649,"type":42},{"text":1650,"level":242},"Conclusion",{"id":886,"data":1652,"type":218},{"text":1653},"An AI Solution Architect is the person or architecture function that turns an AI opportunity into a coherent technical system. The key skill is not knowing the most model names; it is connecting product need, requirements, data, application architecture, AI capabilities, security, runtime, delivery and validation without losing the boundaries between them.",{"id":890,"data":1655,"type":218},{"text":1656},"A strong AI solution architecture can therefore be summarized as: \u003Cstrong>define the target → establish requirements and constraints → design the system boundaries → make significant trade-offs explicit → implement through clear contracts → validate against evidence → operate and evolve deliberately.\u003C\u002Fstrong> The model is important. The solution is the product.",{"id":894,"data":1658,"type":894},{"items":1659,"title":1684},[1660,1663,1666,1669,1672,1675,1678,1681],{"id":898,"answer":1661,"question":1662},"An AI Solution Architect translates a business or product need into the architecture of a concrete AI-enabled solution, defining how application logic, data\u002Fretrieval, models, tools, identity, security, runtime, evaluation and operations work together.","What is an AI Solution Architect?",{"id":902,"answer":1664,"question":1665},"No. The roles can overlap, especially in small teams, but an AI engineer is primarily an implementation role while the solution architect owns or coordinates cross-layer architecture decisions and trade-offs for the complete workload.","Is an AI Solution Architect the same as an AI engineer?",{"id":906,"answer":1667,"question":1668},"Not by definition, but hands-on implementation knowledge is highly valuable because AI architecture crosses APIs, data, retrieval, security, runtimes and operational behavior. The role is defined by architecture responsibility, not by writing every component personally.","Does an AI Solution Architect need to code?",{"id":910,"answer":1670,"question":1671},"No. Model selection is one decision. Production architecture also needs data and retrieval boundaries, permissions, tools, provider\u002Fruntime choices, observability, evaluation, reliability, cost and lifecycle design.","Is choosing an LLM the main job?",{"id":914,"answer":1673,"question":1674},"An AI Solution Architect focuses on one concrete solution or workload. An AI Platform Architect focuses on reusable AI capabilities and guardrails that support multiple solutions.","What is the difference between an AI Solution Architect and an AI Platform Architect?",{"id":918,"answer":1676,"question":1677},"The solution architect works at application\u002Fworkload scope. Enterprise AI architecture works across the organizational portfolio, target architecture, governance, shared capabilities, integration principles and strategic constraints.","What is the difference between an AI Solution Architect and an Enterprise AI Architect?",{"id":922,"answer":1679,"question":1680},"They are architectural patterns or subsystems inside a solution when the requirements justify them. RAG addresses retrieval-grounded context; agents add planning\u002Ftool execution and therefore additional identity, permission, orchestration and operational concerns.","Where do RAG and agents fit?",{"id":926,"answer":1682,"question":1683},"Implementation plus validation evidence: functional tests, evaluation results, security\u002Fauthorization tests, performance and reliability measurements, observability, operational rehearsal and acceptance against the original requirements.","What proves that the architecture works?","AI Solution Architect — FAQ",{"id":931,"data":1686,"type":931},{"title":1687,"entries":1688},"Core terms",[1689,1691,1693,1696,1699,1701,1703,1705],{"term":597,"anchor":936,"definition":1690},"Architecture responsibility for one concrete AI-enabled solution or workload, integrating product requirements with application, data, model, tool, security, runtime and operational design.",{"term":1568,"anchor":939,"definition":1692},"The explicit separation between what belongs to the solution and the users, systems, providers, data sources and environments it interacts with.",{"term":1694,"anchor":943,"definition":1695},"Trust boundary","A point where data, identities or control cross between components with different trust assumptions and therefore require explicit security controls.",{"term":1697,"anchor":947,"definition":1698},"Grounding","Supplying an AI model with relevant external information or evidence so its response can be based on sources beyond model parameters.",{"term":1391,"anchor":950,"definition":1700},"An application boundary that decouples parts of the solution from one model\u002Fprovider interface. Useful when justified by routing, portability or policy needs, but not free of trade-offs.",{"term":1639,"anchor":953,"definition":1702},"Structured measurement of AI workload behavior against defined acceptance criteria, including task quality and relevant safety, security, performance and operational properties.",{"term":603,"anchor":956,"definition":1704},"Architectural role focused on reusable AI platform capabilities used by multiple solutions rather than the architecture of one workload.",{"term":959,"anchor":960,"definition":1706},"Organization-level architecture that coordinates AI capabilities, platforms, governance, integration and strategic constraints across a portfolio.",{"id":963,"data":1708,"type":42},{"text":1709,"level":242},"Related canonical knowledge",{"id":967,"data":1711,"type":218},{"text":1712},"This article sits in the AI Architecture Foundations cluster. Its direct foundations are \u003Cstrong>Generative AI Explained: Models, Retrieval, Tools and Applications Are Not the Same Thing\u003C\u002Fstrong> and \u003Cstrong>ADR vs NFR: Architecture Decisions and System Quality Are Not the Same Thing\u003C\u002Fstrong>. Adjacent canonical nodes include \u003Cstrong>Agentic AI Explained\u003C\u002Fstrong>, \u003Cstrong>Source of Truth in AI Systems\u003C\u002Fstrong>, \u003Cstrong>Vector Databases, Embeddings and Reranking\u003C\u002Fstrong>, \u003Cstrong>What Is Context Engineering?\u003C\u002Fstrong>, \u003Cstrong>RBAC vs Tenant Isolation\u003C\u002Fstrong>, \u003Cstrong>AI Platform Architect\u003C\u002Fstrong>, \u003Cstrong>Enterprise AI Architecture\u003C\u002Fstrong> and \u003Cstrong>AI Governance\u003C\u002Fstrong>. URLs are intentionally not fabricated where those nodes are not yet published.",{"id":971,"data":1714,"type":979},{"link":1715,"meta":1716},"https:\u002F\u002Fstajic.de\u002Fblog\u002Fwhat-is-rag-the-simplest-explanation-of-how-it-works",{"image":1717,"title":1718,"description":1719},{"url":976},"What Is RAG? The Simplest Explanation of How It Works","Existing stajic.de canonical explanation of retrieval-augmented generation, useful for the retrieval\u002Fgrounding part of AI solution architecture.",{"id":981,"data":1721,"type":42},{"text":1722,"level":242},"Primary sources and current architecture guidance",{"id":985,"data":1724,"type":218},{"text":1725},"External sources below support the general architecture claims; the SenseFlow and Aaasaasa AI Client sections are explicitly original project\u002Fimplementation evidence. Current-state references were checked on 8 October 2026. NIST notes that AI RMF 1.0 is being revised, so version-sensitive governance references should be rechecked when a successor is published.",{"id":989,"data":1727,"type":979},{"link":991,"meta":1728},{"image":1729,"title":1730,"description":1731},{"url":976},"ISO\u002FIEC\u002FIEEE 42010:2022 — Architecture Description","Current international standard for the structure and expression of architecture descriptions. It distinguishes architecture from its description and does not prescribe one architecting method, tool or recording format.",{"id":997,"data":1733,"type":979},{"link":999,"meta":1734},{"image":1735,"title":1736,"description":1737},{"url":976},"NIST AI Risk Management Framework","NIST’s AI RMF resource page. As of October 2026 it states that AI RMF 1.0 is being revised and links the Generative AI Profile and related resources.",{"id":1005,"data":1739,"type":979},{"link":1007,"meta":1740},{"image":1741,"title":1742,"description":1743},{"url":976},"NIST AI RMF Core — Govern, Map, Measure, Manage","Official NIST AIRC presentation of the AI RMF 1.0 Core, including the four functions and lifecycle-oriented risk-management framing.",{"id":1013,"data":1745,"type":979},{"link":1015,"meta":1746},{"image":1747,"title":1748,"description":1749},{"url":976},"NIST AI 600-1 — Generative AI Profile","Cross-sectoral Generative AI profile for AI RMF 1.0, published 26 July 2024 and updated by NIST in 2026.",{"id":1021,"data":1751,"type":979},{"link":1023,"meta":1752},{"image":1753,"title":1754,"description":1755},{"url":976},"Microsoft Azure Well-Architected — AI Workloads","Current workload-level architecture guidance covering AI application design, application platform, training data, grounding data, data platform and production-readiness concerns.",{"id":1029,"data":1757,"type":979},{"link":1031,"meta":1758},{"image":1759,"title":1760,"description":1761},{"url":976},"Microsoft — Application Design for AI Workloads","Guidance on model\u002Ftool abstraction, data-access boundaries, identity propagation, authorization and separation of client, intelligence, knowledge and tool layers.",{"id":1037,"data":1763,"type":979},{"link":1039,"meta":1764},{"image":1765,"title":1766,"description":1767},{"url":976},"Microsoft — Design Principles for AI Workloads","Current AI workload design principles across reliability, security, cost, operational excellence and performance, including identity and data-protection responsibilities.",{"id":1045,"data":1769,"type":979},{"link":1047,"meta":1770},{"image":1771,"title":1772,"description":1773},{"url":976},"Microsoft — MLOps and GenAIOps for AI Workloads","Production lifecycle guidance covering monitoring, quality gates, model\u002Fprompt behavior, security and operational measurement.",{"id":1053,"data":1775,"type":979},{"link":1055,"meta":1776},{"image":1777,"title":1778,"description":1779},{"url":976},"AWS Well-Architected Generative AI Lens","AWS architectural guidance for generative AI workloads across operational excellence, security, reliability, performance efficiency, cost optimization and sustainability.",{"id":1061,"data":1781,"type":979},{"link":1063,"meta":1782},{"image":1783,"title":1784,"description":1785},{"url":976},"AWS Well-Architected Agentic AI Lens","Published in 2026, covering agentic-specific architecture concerns including identities, tools, orchestration, human oversight, reliability, tracing and reasoning-loop cost.","2.31.0","An AI Solution Architect turns business requirements into a production-ready AI system across data, models, tools, security, runtime, evaluation and operations.",{"lang":7,"title":208,"content":210,"contentJson":1789,"excerpt":1069},{"time":212,"blocks":1790,"version":1068},[1791,1793,1795,1797,1799,1801,1803,1805,1807,1823,1825,1827,1829,1839,1841,1843,1845,1847,1849,1863,1865,1867,1869,1871,1873,1875,1877,1879,1881,1883,1885,1887,1889,1891,1893,1895,1897,1899,1901,1903,1905,1907,1909,1921,1923,1925,1936,1938,1940,1958,1960,1962,1964,1966,1968,1970,1972,1974,1976,1978,1980,1982,1984,1986,1988,1990,2001,2003,2015,2017,2028,2030,2032,2034,2036,2038,2040,2042,2044,2060,2062,2064,2066,2077,2088,2090,2092,2096,2098,2100,2104,2108,2112,2116,2120,2124,2128,2132,2136],{"id":215,"data":1792,"type":218},{"text":217},{"id":220,"data":1794,"type":225},{"body":222,"title":223,"variant":224},{"id":227,"data":1796,"type":225},{"body":229,"title":230,"variant":231},{"id":233,"data":1798,"type":225},{"body":235,"title":236,"variant":231},{"id":238,"data":1800,"type":243},{"title":240,"maxLevel":241,"minLevel":242},{"id":245,"data":1802,"type":42},{"text":247,"level":242},{"id":249,"data":1804,"type":218},{"text":251},{"id":253,"data":1806,"type":218},{"text":255},{"id":257,"data":1808,"type":299},{"rows":1809,"title":290,"layout":291,"columns":1820},[1810,1812,1814,1816,1818],{"id":261,"label":262,"values":1811},{"model":264,"solution":265},{"id":267,"label":268,"values":1813},{"model":270,"solution":271},{"id":273,"label":274,"values":1815},{"model":276,"solution":277},{"id":279,"label":280,"values":1817},{"model":282,"solution":283},{"id":285,"label":286,"values":1819},{"model":288,"solution":289},[1821,1822],{"id":294,"label":295},{"id":297,"label":298},{"id":301,"data":1824,"type":42},{"text":303,"level":242},{"id":305,"data":1826,"type":218},{"text":307},{"id":309,"data":1828,"type":218},{"text":311},{"id":313,"data":1830,"type":339},{"steps":1831,"title":337,"orientation":338},[1832,1833,1834,1835,1836,1837,1838],{"label":317,"description":318},{"label":320,"description":321},{"label":323,"description":324},{"label":326,"description":327},{"label":329,"description":330},{"label":332,"description":333},{"label":335,"description":336},{"id":341,"data":1840,"type":42},{"text":343,"level":242},{"id":345,"data":1842,"type":218},{"text":347},{"id":349,"data":1844,"type":218},{"text":351},{"id":353,"data":1846,"type":42},{"text":355,"level":242},{"id":357,"data":1848,"type":218},{"text":359},{"id":361,"data":1850,"type":291},{"content":1851,"stretched":43,"withHeadings":14},[1852,1853,1854,1855,1856,1857,1858,1859,1860,1861,1862],[365,366,367],[369,370,371],[373,374,375],[377,378,379],[381,382,383],[385,386,387],[389,390,391],[393,394,395],[397,398,399],[401,402,403],[405,406,407],{"id":409,"data":1864,"type":42},{"text":411,"level":241},{"id":413,"data":1866,"type":218},{"text":415},{"id":417,"data":1868,"type":218},{"text":419},{"id":421,"data":1870,"type":42},{"text":423,"level":241},{"id":425,"data":1872,"type":218},{"text":427},{"id":429,"data":1874,"type":218},{"text":431},{"id":433,"data":1876,"type":42},{"text":435,"level":241},{"id":437,"data":1878,"type":218},{"text":439},{"id":441,"data":1880,"type":218},{"text":443},{"id":445,"data":1882,"type":42},{"text":447,"level":241},{"id":449,"data":1884,"type":218},{"text":451},{"id":453,"data":1886,"type":218},{"text":455},{"id":457,"data":1888,"type":42},{"text":459,"level":241},{"id":461,"data":1890,"type":218},{"text":463},{"id":465,"data":1892,"type":218},{"text":467},{"id":469,"data":1894,"type":42},{"text":471,"level":241},{"id":473,"data":1896,"type":218},{"text":475},{"id":477,"data":1898,"type":218},{"text":479},{"id":481,"data":1900,"type":42},{"text":483,"level":241},{"id":485,"data":1902,"type":218},{"text":487},{"id":489,"data":1904,"type":218},{"text":491},{"id":493,"data":1906,"type":42},{"text":495,"level":242},{"id":497,"data":1908,"type":218},{"text":499},{"id":501,"data":1910,"type":291},{"content":1911,"stretched":43,"withHeadings":14},[1912,1913,1914,1915,1916,1917,1918,1919,1920],[505,506],[508,509],[511,512],[514,515],[517,518],[520,521],[523,524],[526,527],[529,530],{"id":532,"data":1922,"type":42},{"text":534,"level":242},{"id":536,"data":1924,"type":218},{"text":538},{"id":540,"data":1926,"type":291},{"content":1927,"stretched":43,"withHeadings":14},[1928,1929,1930,1931,1932,1933,1934,1935],[544,545,546,547],[549,550,551,552],[554,555,556,557],[559,560,561,562],[564,565,566,567],[569,570,571,572],[574,575,576,577],[579,580,581,582],{"id":584,"data":1937,"type":42},{"text":586,"level":242},{"id":588,"data":1939,"type":218},{"text":590},{"id":592,"data":1941,"type":299},{"rows":1942,"title":631,"layout":291,"columns":1955},[1943,1945,1947,1949,1951,1953],{"id":596,"label":597,"values":1944},{"role":599,"focus":600},{"id":602,"label":603,"values":1946},{"role":605,"focus":606},{"id":608,"label":609,"values":1948},{"role":611,"focus":612},{"id":614,"label":615,"values":1950},{"role":617,"focus":618},{"id":620,"label":621,"values":1952},{"role":623,"focus":624},{"id":626,"label":627,"values":1954},{"role":629,"focus":630},[1956,1957],{"id":634,"label":635},{"id":637,"label":638},{"id":640,"data":1959,"type":218},{"text":642},{"id":644,"data":1961,"type":42},{"text":646,"level":242},{"id":648,"data":1963,"type":225},{"body":650,"title":651,"variant":652},{"id":654,"data":1965,"type":42},{"text":656,"level":241},{"id":658,"data":1967,"type":218},{"text":660},{"id":662,"data":1969,"type":218},{"text":664},{"id":666,"data":1971,"type":218},{"text":668},{"id":670,"data":1973,"type":42},{"text":672,"level":241},{"id":674,"data":1975,"type":218},{"text":676},{"id":678,"data":1977,"type":218},{"text":680},{"id":682,"data":1979,"type":218},{"text":684},{"id":686,"data":1981,"type":42},{"text":688,"level":242},{"id":690,"data":1983,"type":218},{"text":692},{"id":694,"data":1985,"type":218},{"text":696},{"id":698,"data":1987,"type":218},{"text":700},{"id":702,"data":1989,"type":42},{"text":704,"level":242},{"id":706,"data":1991,"type":291},{"content":1992,"stretched":43,"withHeadings":14},[1993,1994,1995,1996,1997,1998,1999,2000],[710,711],[713,714],[716,717],[719,720],[722,723],[725,726],[728,729],[731,732],{"id":734,"data":2002,"type":42},{"text":736,"level":242},{"id":738,"data":2004,"type":291},{"content":2005,"stretched":43,"withHeadings":14},[2006,2007,2008,2009,2010,2011,2012,2013,2014],[742,743,744],[746,747,748],[750,751,752],[754,755,756],[758,759,760],[762,763,764],[766,767,768],[770,771,772],[774,775,776],{"id":778,"data":2016,"type":42},{"text":780,"level":242},{"id":782,"data":2018,"type":339},{"steps":2019,"title":809,"orientation":338},[2020,2021,2022,2023,2024,2025,2026,2027],{"label":786,"description":787},{"label":789,"description":790},{"label":792,"description":793},{"label":795,"description":796},{"label":798,"description":799},{"label":801,"description":802},{"label":804,"description":805},{"label":807,"description":808},{"id":811,"data":2029,"type":42},{"text":813,"level":242},{"id":815,"data":2031,"type":218},{"text":817},{"id":819,"data":2033,"type":218},{"text":821},{"id":823,"data":2035,"type":218},{"text":825},{"id":827,"data":2037,"type":42},{"text":829,"level":242},{"id":831,"data":2039,"type":218},{"text":833},{"id":835,"data":2041,"type":218},{"text":837},{"id":839,"data":2043,"type":42},{"text":841,"level":242},{"id":843,"data":2045,"type":291},{"content":2046,"stretched":43,"withHeadings":14},[2047,2048,2049,2050,2051,2052,2053,2054,2055,2056,2057,2058,2059],[847,848],[786,850],[852,853],[268,855],[857,858],[860,861],[863,864],[866,867],[869,870],[872,873],[875,876],[286,878],[280,880],{"id":882,"data":2061,"type":42},{"text":884,"level":242},{"id":886,"data":2063,"type":218},{"text":888},{"id":890,"data":2065,"type":218},{"text":892},{"id":894,"data":2067,"type":894},{"items":2068,"title":929},[2069,2070,2071,2072,2073,2074,2075,2076],{"id":898,"answer":899,"question":900},{"id":902,"answer":903,"question":904},{"id":906,"answer":907,"question":908},{"id":910,"answer":911,"question":912},{"id":914,"answer":915,"question":916},{"id":918,"answer":919,"question":920},{"id":922,"answer":923,"question":924},{"id":926,"answer":927,"question":928},{"id":931,"data":2078,"type":931},{"title":933,"entries":2079},[2080,2081,2082,2083,2084,2085,2086,2087],{"term":597,"anchor":936,"definition":937},{"term":792,"anchor":939,"definition":940},{"term":942,"anchor":943,"definition":944},{"term":946,"anchor":947,"definition":948},{"term":564,"anchor":950,"definition":951},{"term":872,"anchor":953,"definition":954},{"term":603,"anchor":956,"definition":957},{"term":959,"anchor":960,"definition":961},{"id":963,"data":2089,"type":42},{"text":965,"level":242},{"id":967,"data":2091,"type":218},{"text":969},{"id":971,"data":2093,"type":979},{"link":973,"meta":2094},{"image":2095,"title":977,"description":978},{"url":976},{"id":981,"data":2097,"type":42},{"text":983,"level":242},{"id":985,"data":2099,"type":218},{"text":987},{"id":989,"data":2101,"type":979},{"link":991,"meta":2102},{"image":2103,"title":994,"description":995},{"url":976},{"id":997,"data":2105,"type":979},{"link":999,"meta":2106},{"image":2107,"title":1002,"description":1003},{"url":976},{"id":1005,"data":2109,"type":979},{"link":1007,"meta":2110},{"image":2111,"title":1010,"description":1011},{"url":976},{"id":1013,"data":2113,"type":979},{"link":1015,"meta":2114},{"image":2115,"title":1018,"description":1019},{"url":976},{"id":1021,"data":2117,"type":979},{"link":1023,"meta":2118},{"image":2119,"title":1026,"description":1027},{"url":976},{"id":1029,"data":2121,"type":979},{"link":1031,"meta":2122},{"image":2123,"title":1034,"description":1035},{"url":976},{"id":1037,"data":2125,"type":979},{"link":1039,"meta":2126},{"image":2127,"title":1042,"description":1043},{"url":976},{"id":1045,"data":2129,"type":979},{"link":1047,"meta":2130},{"image":2131,"title":1050,"description":1051},{"url":976},{"id":1053,"data":2133,"type":979},{"link":1055,"meta":2134},{"image":2135,"title":1058,"description":1059},{"url":976},{"id":1061,"data":2137,"type":979},{"link":1063,"meta":2138},{"image":2139,"title":1066,"description":1067},{"url":976},"Post erfolgreich abgerufen",{"items":2142,"source":2227,"manualIds":2228,"manualMatchedIds":2229},[2143,2150,2157,2164,2171,2178,2185,2192,2199,2206,2213,2220],{"id":2144,"slug":2145,"title":2146,"excerpt":2147,"featuredImage":2148,"publishedAt":2149},"486","source-of-truth-in-ai-systems-where-reliable-knowledge-actually-comes-from","Izvor istine u AI sistemima: Odakle pouzdano znanje zaista dolazi","Izvor istine definiše koji je izvor merodavan za određenu činjenicu ili stanje. Saznajte kako se razlikuje od RAG-a, porekla, memorije, konteksta, vektorskih baza podataka i sistema evidencije.","\u002Fuploads\u002F2026\u002F10\u002Fsource-of-truth-in-ai-systems-where-reliable-knowledge-actually-comes-from-1791479103235-6bq9em.webp","2026-10-08T13:02:00.000Z",{"id":2151,"slug":2152,"title":2153,"excerpt":2154,"featuredImage":2155,"publishedAt":2156},"470","what-should-an-ai-agent-remember-forget-recompute-or-retrieve-again","Šta bi AI agent trebalo da zapamti, zaboravi, ponovo izračuna ili ponovo preuzme?","Dugotrajni agenti ne bi trebalo da pamte sve. Ovaj članak pruža praktičan model životnog ciklusa za odlučivanje o tome šta pripada trajnoj memoriji, šta bi trebalo ponovo preuzeti, šta je bezbednije ponovo izračunati i šta bi trebalo da istekne ili bude zamenjeno.","\u002Fuploads\u002F2026\u002F09\u002Fwhat-should-an-ai-agent-remember-forget-recompute-or-retrieve-again-1790351131087-iehz28.webp","2026-09-25T09:43:00.000Z",{"id":2158,"slug":2159,"title":2160,"excerpt":2161,"featuredImage":2162,"publishedAt":2163},"487","vector-databases-embeddings-and-reranking-three-different-parts-of-retrieval","Vektorske baze podataka, ugrađivanja i ponovno rangiranje: Tri različita dela pretraživanja","Embedinzi predstavljaju značenje, vektorske baze podataka pronalaze kandidate, a rerangirači prečišćavaju rezultate. Saznajte kako se ova tri sloja pronalaženja razlikuju i kako rade zajedno u RAG-u.","\u002Fuploads\u002F2026\u002F10\u002Fvector-databases-embeddings-and-reranking-three-different-parts-of-retrieval-1791480129884-9dtasz.webp","2026-10-08T11:21:00.000Z",{"id":2165,"slug":2166,"title":2167,"excerpt":2168,"featuredImage":2169,"publishedAt":2170},"484","what-is-an-ai-platform-architect-models-data-runtime-security-and-operations","Šta je arhitekta AI platforme? Modeli, podaci, izvršno okruženje, bezbednost i operacije","Arhitekta AI platforme projektuje višekratno upotrebljive AI temelje kroz modele, provajdere, pretragu, agente, identitet, bezbednost, evaluaciju, opservabilnost i operacije.","\u002Fuploads\u002F2026\u002F10\u002Fwhat-is-an-ai-platform-architect-models-data-runtime-security-and-operations-1791477229171-ou3zcc.webp","2026-10-08T12:32:00.000Z",{"id":2172,"slug":2173,"title":2174,"excerpt":2175,"featuredImage":2176,"publishedAt":2177},"466","the-gpu-is-not-the-product-future-proof-private-ai-architecture","GPU nije proizvod: Privatna AI arhitektura spremna za budućnost","Privatna AI infrastruktura ne bi trebalo da bude projektovana oko jednog GPU-a ili jednog modela. Otporniji pristup kombinuje brze GPU-ove za inferenciju, memorijski bogate AI sisteme, čvorove za fizički AI i opcione vodeće modele u oblaku iza sloja za rutiranje koji prepoznaje mogućnosti.","\u002Fuploads\u002F2026\u002F09\u002Fthe-gpu-is-not-the-product-future-proof-private-ai-architecture-1790140878812-8hsl39.webp","2026-09-23T01:19:00.000Z",{"id":2179,"slug":2180,"title":2181,"excerpt":2182,"featuredImage":2183,"publishedAt":2184},"485","enterprise-ai-architecture-what-changes-when-ai-enters-a-company","Enterprise AI arhitektura: Šta se menja kada AI uđe u kompaniju","Enterprise AI arhitektura objašnjava kako AI menja korporativne sisteme kroz autoritet podataka, identitet, dozvole, provajdere, rizik, upravljanje, evaluaciju, usklađenost i operacije.","\u002Fuploads\u002F2026\u002F10\u002Fenterprise-ai-architecture-what-changes-when-ai-enters-a-company-1791478161363-czrwaq.webp","2026-10-08T10:48:00.000Z",{"id":2186,"slug":2187,"title":2188,"excerpt":2189,"featuredImage":2190,"publishedAt":2191},"490","rbac-vs-tenant-isolation-two-different-security-boundaries","RBAC naspram izolacije zakupaca: dve različite bezbednosne granice","RBAC kontroliše šta korisnik sme da radi; izolacija zakupaca kontroliše kojim resursima tog zakupca ta radnja može da pristupi. Saznajte zašto bezbednost višekorisničkog SaaS-a zahteva obe granice.","\u002Fuploads\u002F2026\u002F10\u002Frbac-vs-tenant-isolation-two-different-security-boundaries-1791485111528-qqtzby.webp","2026-10-08T14:43:00.000Z",{"id":2193,"slug":2194,"title":2195,"excerpt":2196,"featuredImage":2197,"publishedAt":2198},"489","agentic-ai-explained-when-an-ai-system-can-plan-use-tools-and-act","Agentna AI objašnjena: Kada AI sistem može da planira, koristi alate i deluje","Agentna AI koristi modele unutar višekoračnih izvršnih petlji gde mogu da biraju alate, posmatraju rezultate, ažuriraju stanje i prilagode svoju sledeću akciju unutar eksplicitnih granica izvršavanja i dozvola.","\u002Fuploads\u002F2026\u002F10\u002Fagentic-ai-explained-when-an-ai-system-can-plan-use-tools-and-act-1791481499084-wnji2a.webp","2026-10-08T11:43:00.000Z",{"id":2200,"slug":2201,"title":2202,"excerpt":2203,"featuredImage":2204,"publishedAt":2205},"481","generative-ai-explained-models-retrieval-tools-and-applications-are-not-the-same-thing","Generativna veštačka inteligencija objašnjena: modeli, pretraga, alati i aplikacije nisu ista stvar","Generativna AI je više od modela. Saznajte kako se modeli, pretraga, alati, kontekst, okruženja i aplikacije uklapaju u produkcione AI sisteme.","\u002Fuploads\u002F2026\u002F10\u002Fgenerative-ai-explained-models-retrieval-tools-and-applications-are-not-the-same-thing-1791475411822-pp0dvz.webp","2026-10-08T12:00:00.000Z",{"id":2207,"slug":2208,"title":2209,"excerpt":2210,"featuredImage":2211,"publishedAt":2212},"468","ai-agent-memory-is-not-rag-how-to-separate-memory-retrieval-state-and-context","Memorija AI agenta nije RAG: Kako razdvojiti memoriju, pronalaženje, stanje i kontekst","Memorija agenta, RAG, stanje i kontekst često se koriste kao da su međusobno zamenjivi. Oni to nisu. Ovaj praktični arhitektonski model razdvaja ova četiri sloja, pokazuje gde svaki pripada i objašnjava šta se kvari kada ih sistemi stope u jedno.","\u002Fuploads\u002F2026\u002F09\u002Fai-agent-memory-is-not-rag-how-to-separate-memory-retrieval-state-and-context-1790350560308-np0xy6.webp","2026-09-25T11:34:00.000Z",{"id":2214,"slug":2215,"title":2216,"excerpt":2217,"featuredImage":2218,"publishedAt":2219},"492","mcp-explained-what-it-connects-what-it-does-not-do-and-where-it-fits","MCP objašnjen: Šta povezuje, šta ne radi i gde se uklapa","Model Context Protocol povezuje AI aplikacije sa eksternim alatima, resursima i promptovima kroz standardnu granicu klijent-server. Saznajte šta MCP radi, šta ne radi i gde se uklapa u arhitekturu agenata.","\u002Fuploads\u002F2026\u002F10\u002Fmcp-explained-what-it-connects-what-it-does-not-do-and-where-it-fits-1791486640275-7ub1cq.webp","2026-10-08T15:09:00.000Z",{"id":2221,"slug":2222,"title":2223,"excerpt":2224,"featuredImage":2225,"publishedAt":2226},"476","mcp-vs-a2a-vs-ucp-vs-ap2-vs-a2ui-the-agent-protocol-stack-explained","MCP vs A2A vs UCP vs AP2 vs A2UI: Objašnjen stek agentskih protokola","MCP, A2A, UCP, AP2 i A2UI se često predstavljaju kao konkurentski standardi za agente. Oni uglavnom rešavaju različite probleme interoperabilnosti. Ovaj vodič mapira svaki protokol na granicu koju zapravo standardizuje—i pokazuje kako oni mogu da rade zajedno u jednom produkcionom sistemu.","\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","fallback",[],[]]