{"id":11396,"date":"2026-10-04T00:40:47","date_gmt":"2026-10-04T00:40:47","guid":{"rendered":"https:\/\/bitep.net\/blog\/2026\/10\/04\/etcjournalcom\/"},"modified":"2026-10-04T00:40:47","modified_gmt":"2026-10-04T00:40:47","slug":"etcjournalcom","status":"publish","type":"post","link":"https:\/\/bitep.net\/blog\/2026\/10\/04\/etcjournalcom\/","title":{"rendered":"etcjournal.com"},"content":{"rendered":"<h2>Reaktiv \u00c7atbotlardan Avtonom \u018fm\u0259kda\u015flara: &#8220;Agentic AI&#8221; \u0130nqilab\u0131<\/h2>\n<p>S\u00fcni intellekt s\u0259nayesi son iki il \u0259rzind\u0259 \u0259sas\u0259n istifad\u0259\u00e7i sor\u011fular\u0131na cavab ver\u0259n (reaktiv) b\u00f6y\u00fck dil modell\u0259rinin (LLM) t\u0259siri alt\u0131nda idi. \u0130stifad\u0259\u00e7i sual verir, model is\u0259 m\u0259tn generasiya edirdi. Lakin bu paradiqma art\u0131q k\u00f6kl\u00fc \u015f\u0259kild\u0259 d\u0259yi\u015fir. Texnologiya d\u00fcnyas\u0131 sad\u0259 dialoq interfeysl\u0259rind\u0259n m\u00fcr\u0259kk\u0259b m\u00fchakim\u0259 qabiliyy\u0259tin\u0259 malik, s\u0259rb\u0259st q\u0259rar ver\u0259 bil\u0259n v\u0259 m\u00fc\u0259yy\u0259n olunmu\u015f m\u0259qs\u0259dl\u0259ri m\u00fcst\u0259qil \u015f\u0259kild\u0259 icra ed\u0259n <strong>Agentik S\u00fcni \u0130ntellekt (Agentic AI)<\/strong> sisteml\u0259rin\u0259 ke\u00e7id edir.<\/p>\n<p>Agentik AI art\u0131q sad\u0259c\u0259 bir k\u00f6m\u0259k\u00e7i v\u0259 ya m\u0259tn yazma al\u0259ti deyil, t\u0259\u015fkilat daxilind\u0259 tamh\u00fcquqlu &#8220;r\u0259q\u0259msal h\u0259mkar&#8221; (digital coworker) rolunu \u00f6z \u00fcz\u0259rin\u0259 g\u00f6t\u00fcr\u00fcr. Bu sisteml\u0259r veril\u0259n tap\u015f\u0131r\u0131\u011f\u0131 alt m\u0259rh\u0259l\u0259l\u0259r\u0259 b\u00f6l\u00fcr, laz\u0131mi xarici al\u0259tl\u0259ri (API-l\u0259r, veril\u0259nl\u0259r bazalar\u0131, skriptl\u0259r) i\u015f\u0259 sal\u0131r, x\u0259talar\u0131 analiz ed\u0259r\u0259k \u00f6z\u00fcn\u00fc d\u00fcz\u0259ldir v\u0259 prosesi insan m\u00fcdaxil\u0259si olmadan sona \u00e7atd\u0131r\u0131r.<\/p>\n<h2>Agentik S\u00fcni \u0130ntellektin Texniki Arxitekturas\u0131<\/h2>\n<p>\u018fn\u0259n\u0259vi LLM-l\u0259rl\u0259 agentik sisteml\u0259r aras\u0131ndak\u0131 \u0259sas f\u0259rq onlar\u0131n arxitektur qurulu\u015fundad\u0131r. T\u0259k bir modelin imkanlar\u0131 kontekst p\u0259nc\u0259r\u0259si v\u0259 statik m\u0259lumat bazas\u0131 il\u0259 m\u0259hdudla\u015fd\u0131\u011f\u0131 halda, Agentik AI d\u00f6rd \u0259sas s\u00fctun \u00fcz\u0259rind\u0259 qurulur:<\/p>\n<ul>\n<li><strong>Planla\u015fd\u0131rma v\u0259 M\u00fchakim\u0259 (Planning &#038; Reasoning):<\/strong> Agent qar\u015f\u0131ya qoyulan m\u0259qs\u0259di d\u0259rk edir v\u0259 <em>Chain-of-Thought<\/em> (Fikir z\u0259nciri) v\u0259 ya <em>Tree-of-Thoughts<\/em> yana\u015fmalar\u0131 il\u0259 problemi m\u0259ntiqi add\u0131mlara ay\u0131r\u0131r.<\/li>\n<li><strong>Yadda\u015f Mexanizmi (Memory):<\/strong> Q\u0131sam\u00fcdd\u0259tli (cari sessiya konteksti) v\u0259 uzunm\u00fcdd\u0259tli yadda\u015f (vektor bazalar\u0131 &#8211; RAG vasit\u0259sil\u0259 ke\u00e7mi\u015f t\u0259cr\u00fcb\u0259l\u0259rin v\u0259 bilikl\u0259rin saxlanmas\u0131).<\/li>\n<li><strong>Al\u0259tl\u0259rd\u0259n \u0130stifad\u0259 (Tool Calling):<\/strong> Modelin real d\u00fcnya il\u0259 \u0259laq\u0259si. Buraya SQL sor\u011fular\u0131 icra etm\u0259k, brauzeri idar\u0259 ed\u0259r\u0259k veb-skreypinq aparmaq, API sor\u011fular\u0131 g\u00f6nd\u0259rm\u0259k v\u0259 terminalda kod i\u015f\u0259 salmaq daxildir.<\/li>\n<li><strong>Avtonom \u018fks-\u018flaq\u0259 D\u00f6vr\u0259si (Reflection &#038; Execution Loop):<\/strong> \u0130cra olunan h\u0259r add\u0131m\u0131n n\u0259tic\u0259sini qiym\u0259tl\u0259ndirm\u0259k v\u0259 x\u0259ta ba\u015f verdikd\u0259 f\u0259rqli strategiya se\u00e7\u0259r\u0259k prosesi davam etdirm\u0259k.<\/li>\n<\/ul>\n<table border=\"1\" cellpadding=\"8\" cellspacing=\"0\">\n<thead>\n<tr>\n<th>X\u00fcsusiyy\u0259t<\/th>\n<th>\u018fn\u0259n\u0259vi Generativ AI (M\u0259s. ChatGPT-3.5)<\/th>\n<th>Agentik AI Sisteml\u0259ri (M\u0259s. CrewAI, AutoGen)<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><strong>\u0130\u015f Rejimi<\/strong><\/td>\n<td>Reaktiv (Prompt daxil edilir, cavab al\u0131n\u0131r)<\/td>\n<td>Proaktiv \/ M\u0259qs\u0259dy\u00f6nl\u00fc (M\u0259qs\u0259d verilir, prosesi idar\u0259 edir)<\/td>\n<\/tr>\n<tr>\n<td><strong>Avtonomiya<\/strong><\/td>\n<td>A\u015fa\u011f\u0131 (H\u0259r add\u0131m \u00fc\u00e7\u00fcn istifad\u0259\u00e7i t\u0259l\u0259b olunur)<\/td>\n<td>Y\u00fcks\u0259k (B\u00fct\u00fcn i\u015f ax\u0131n\u0131n\u0131 t\u0259kba\u015f\u0131na icra edir)<\/td>\n<\/tr>\n<tr>\n<td><strong>Al\u0259t \u0130nteqrasiyas\u0131<\/strong><\/td>\n<td>M\u0259hdud v\u0259 ya yoxdur<\/td>\n<td>Geni\u015f (API, Bash, Vektor bazalar\u0131, Brauzer)<\/td>\n<\/tr>\n<tr>\n<td><strong>X\u0259ta \u0130dar\u0259etm\u0259si<\/strong><\/td>\n<td>\u0130stifad\u0259\u00e7i x\u0259tan\u0131 bildirib yenid\u0259n soru\u015fmal\u0131d\u0131r<\/td>\n<td>Daxili t\u0259nziml\u0259m\u0259 v\u0259 \u00f6z\u00fcn\u00fc-d\u00fcz\u0259ltm\u0259 (Self-healing)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2>Biznes v\u0259 \u0130nformasiya Texnologiyalar\u0131nda T\u0259tbiq Sah\u0259l\u0259ri<\/h2>\n<p>Agentik AI sisteml\u0259ri \u0259n\u0259n\u0259vi avtomatla\u015fd\u0131rma (RPA) texnologiyalar\u0131ndan f\u0259rqli olaraq qeyri-m\u00fc\u0259yy\u0259nlikl\u0259rl\u0259 i\u015fl\u0259y\u0259 bilir. \u018fsas t\u0259tbiq istiqam\u0259tl\u0259ri bunlard\u0131r:<\/p>\n<ul>\n<li><strong>IT D\u0259st\u0259k v\u0259 \u0130nsident \u0130dar\u0259etm\u0259si:<\/strong> Daxil olan biletl\u0259rin (tickets) loglar\u0131n\u0131 oxumaq, sistem konfiqurasiyalar\u0131n\u0131 yoxlamaq, problemi a\u015fkarlay\u0131b serveri yenid\u0259n ba\u015flatmaq v\u0259 n\u0259tic\u0259 bar\u0259d\u0259 komandaya m\u0259lumat verm\u0259k.<\/li>\n<li><strong>Maliyy\u0259 \u018fm\u0259liyyatlar\u0131:<\/strong> M\u00fcxt\u0259lif formatl\u0131 invoyslar\u0131n qar\u015f\u0131la\u015fd\u0131r\u0131lmas\u0131 (reconciliation), anomaliyalar\u0131n a\u015fkarlanmas\u0131 v\u0259 ERP sisteml\u0259rin\u0259 m\u0259lumatlar\u0131n avtonom daxil edilm\u0259si.<\/li>\n<li><strong>Proqram T\u0259minat\u0131n\u0131n \u0130nki\u015faf\u0131:<\/strong> T\u0259l\u0259bl\u0259r \u0259sas\u0131nda arxitekturan\u0131n qurulmas\u0131, kodun yaz\u0131lmas\u0131, testl\u0259rin i\u015f\u0259 sal\u0131nmas\u0131 v\u0259 CI\/CD konveyerin\u0259 inteqrasiya edilm\u0259si (M\u0259s\u0259l\u0259n, Devin v\u0259 b\u0259nz\u0259ri proqram\u00e7\u0131 agentl\u0259r).<\/li>\n<\/ul>\n<h2>Praktik T\u0259limat: Python v\u0259 CrewAI il\u0259 Sad\u0259 Avtonom Agentin Qurulmas\u0131<\/h2>\n<p>Agentik AI konsepsiyas\u0131n\u0131 daha yax\u015f\u0131 anlamaq \u00fc\u00e7\u00fcn \u00e7ox-agentli (multi-agent) arxitekturan\u0131 t\u0259min ed\u0259n m\u00fcasir freymvorklardan biri olan <strong>CrewAI<\/strong> vasit\u0259sil\u0259 sad\u0259 t\u0259dqiqat v\u0259 analiz agentl\u0259ri komandas\u0131 quraq.<\/p>\n<h3>1. M\u00fchitin Haz\u0131rlanmas\u0131<\/h3>\n<p>Terminalda t\u0259l\u0259b olunan kitabxanalar\u0131 qura\u015fd\u0131r\u0131n:<\/p>\n<pre><code class=\"language-bash\">pip install crewai crewai-tools langchain-openai<\/code><\/pre>\n<h3>2. Agentl\u0259rin v\u0259 V\u0259zif\u0259l\u0259rin T\u0259yin Edilm\u0259si<\/h3>\n<p>A\u015fa\u011f\u0131dak\u0131 Python skriptind\u0259 iki agent yarad\u0131l\u0131r: biri texnoloji ara\u015fd\u0131rma apar\u0131r, dig\u0259ri is\u0259 bu m\u0259lumatlar\u0131 texniki x\u00fclas\u0259y\u0259 \u00e7evirir.<\/p>\n<pre><code class=\"language-python\">import os\nfrom crewai import Agent, Task, Crew, Process\nfrom crewai_tools import SerperDevTool\n\n# API a\u00e7arlar\u0131n\u0131 t\u0259yin edin\nos.environ[\"OPENAI_API_KEY\"] = \"sk-...\"\nos.environ[\"SERPER_API_KEY\"] = \"serper_api_key_buraya\"\n\n# Axtar\u0131\u015f al\u0259ti\nsearch_tool = SerperDevTool()\n\n# 1-ci Agent: T\u0259dqiqat\u00e7\u0131\nresearcher = Agent(\n    role=\"Apar\u0131c\u0131 Texnoloji Analitik\",\n    goal=\"2025-ci il \u00fczr\u0259 Agentic AI trendl\u0259rini d\u0259rind\u0259n ara\u015fd\u0131rmaq\",\n    backstory=\"S\u0259n qabaqc\u0131l s\u00fcni intellekt yenilikl\u0259rini izl\u0259y\u0259n t\u0259cr\u00fcb\u0259li analitiks\u0259n.\",\n    tools=[search_tool],\n    verbose=True\n)\n\n# 2-ci Agent: Texniki Redaktor\nwriter = Agent(\n    role=\"Ba\u015f Texniki Redaktor\",\n    goal=\"Ara\u015fd\u0131rma n\u0259tic\u0259l\u0259rini pe\u015f\u0259kar hesabat hal\u0131na salmaq\",\n    backstory=\"S\u0259n m\u00fcr\u0259kk\u0259b texniki konsepsiyalar\u0131 anla\u015f\u0131ql\u0131 v\u0259 strukturla\u015fd\u0131r\u0131lm\u0131\u015f formata salan redaktorsan.\",\n    verbose=True\n)\n\n# V\u0259zif\u0259l\u0259rin t\u0259yini\ntask1 = Task(\n    description=\"Agentic AI-nin \u0259n son 3 real t\u0259tbiq sah\u0259sini ara\u015fd\u0131r v\u0259 \u0259sas x\u00fcsusiyy\u0259tl\u0259rini qeyd et.\",\n    expected_output=\"3 \u0259sas t\u0259tbiq n\u00fcmun\u0259si olan madd\u0259li hesabat.\",\n    agent=researcher\n)\n\ntask2 = Task(\n    description=\"Ara\u015fd\u0131rma n\u0259tic\u0259l\u0259rini BITEP.net standartlar\u0131na uy\u011fun x\u00fclas\u0259 m\u0259qal\u0259y\u0259 \u00e7evir.\",\n    expected_output=\"Giri\u015f, texniki analiz v\u0259 n\u0259tic\u0259 hiss\u0259sind\u0259n ibar\u0259t yekun m\u0259qal\u0259.\",\n    agent=writer\n)\n\n# Komandan\u0131n i\u015f\u0259 sal\u0131nmas\u0131\ncrew = Crew(\n    agents=[researcher, writer],\n    tasks=[task1, task2],\n    process=Process.sequential\n)\n\nresult = crew.kickoff()\nprint(\"\\n--- YEKUN N\u018fT\u0130C\u018f ---\\n\")\nprint(result)<\/code><\/pre>\n<p>Bu n\u00fcmun\u0259d\u0259 agentl\u0259r veril\u0259n m\u0259qs\u0259d\u0259 \u00e7atmaq \u00fc\u00e7\u00fcn bir-birl\u0259ri il\u0259 \u0259laq\u0259 qurur, internetd\u0259 axtar\u0131\u015f apar\u0131r v\u0259 n\u0259tic\u0259ni t\u0259nziml\u0259y\u0259r\u0259k insan m\u00fcdaxil\u0259si olmadan son hesabat\u0131 t\u0259qdim edirl\u0259r.<\/p>\n<h2>T\u0259hl\u00fck\u0259sizlik \u00c7a\u011f\u0131r\u0131\u015flar\u0131 v\u0259 Risk Amill\u0259ri<\/h2>\n<p>Agentik sisteml\u0259r sistem resurslar\u0131na v\u0259 xarici al\u0259tl\u0259r\u0259 birba\u015fa \u00e7\u0131x\u0131\u015f \u0259ld\u0259 etdikc\u0259 yeni kibert\u0259hl\u00fck\u0259sizlik vektorlar\u0131 yaran\u0131r:<\/p>\n<ul>\n<li><strong>Qeyri-m\u00fc\u0259yy\u0259n D\u00f6vr\u0259l\u0259r (Infinite Loops):<\/strong> Modelin bir x\u0259tan\u0131 h\u0259ll ed\u0259 bilm\u0259y\u0259r\u0259k daimi olaraq eyni \u0259m\u0259liyyatlar\u0131 t\u0259krarlamas\u0131 v\u0259 resurslar\u0131 (token v\u0259 hesablama g\u00fcc\u00fc) t\u00fck\u0259tm\u0259si.<\/li>\n<li><strong>Sistem \u0130caz\u0259l\u0259rinin Sui-istifad\u0259si (Excessive Agency):<\/strong> Agent\u0259 veril\u0259n s\u0259lahiyy\u0259tl\u0259rin h\u0259ddind\u0259n art\u0131q geni\u015f olmas\u0131 (m\u0259s\u0259l\u0259n, bazadan sad\u0259c\u0259 oxumaq \u0259v\u0259zin\u0259, s\u0259hv\u0259n c\u0259dv\u0259ll\u0259ri silm\u0259 \u0259mri icra etm\u0259si).<\/li>\n<li><strong>Dolay\u0131 Sor\u011fu \u0130nyeksiyas\u0131 (Indirect Prompt Injection):<\/strong> Agentin oxudu\u011fu xarici veb-s\u0259hif\u0259d\u0259 v\u0259 ya s\u0259n\u0259dd\u0259 gizl\u0259dilmi\u015f z\u0259r\u0259rli t\u0259limatlar\u0131n modeli \u00e7a\u015fd\u0131raraq t\u0259hl\u00fck\u0259li \u0259mrl\u0259r icra etm\u0259sin\u0259 s\u0259b\u0259b olmas\u0131.<\/li>\n<\/ul>\n<h2>N\u0259tic\u0259<\/h2>\n<p>S\u00fcni intellekt &#8220;biziml\u0259 dan\u0131\u015fan&#8221; sisteml\u0259rd\u0259n &#8220;biziml\u0259 birlikd\u0259 i\u015fl\u0259y\u0259n&#8221; platformalara \u00e7evrilir. Agentic AI sad\u0259c\u0259 m\u0259hsuldarl\u0131\u011f\u0131 art\u0131ran \u0259lav\u0259 proqram t\u0259minat\u0131 deyil; o, r\u0259q\u0259msal i\u015f\u00e7i q\u00fcvv\u0259sinin \u0259sas\u0131n\u0131 t\u0259\u015fkil ed\u0259c\u0259k yeni hesablama paradiqmas\u0131d\u0131r. T\u0259\u015fkilatlar \u00fc\u00e7\u00fcn bu ke\u00e7id\u0259 haz\u0131rla\u015fmaq \u2014 infrastrukturu API \u0259sasl\u0131 arxitekturaya uy\u011funla\u015fd\u0131rmaq v\u0259 agentl\u0259rin t\u0259hl\u00fck\u0259siz idar\u0259etm\u0259 \u00e7\u0259r\u00e7iv\u0259l\u0259rini formala\u015fd\u0131rmaq art\u0131q strateji z\u0259rur\u0259tdir.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Reaktiv \u00c7atbotlardan Avtonom \u018fm\u0259kda\u015flara: &#8220;Agentic AI&#8221; \u0130nqilab\u0131 S\u00fcni intellekt s\u0259nayesi son iki il \u0259rzind\u0259 \u0259sas\u0259n istifad\u0259\u00e7i sor\u011fular\u0131na cavab ver\u0259n (reaktiv) b\u00f6y\u00fck dil modell\u0259rinin (LLM) t\u0259siri alt\u0131nda idi. \u0130stifad\u0259\u00e7i sual verir, model is\u0259 m\u0259tn generasiya edirdi. Lakin bu paradiqma art\u0131q k\u00f6kl\u00fc \u015f\u0259kild\u0259 d\u0259yi\u015fir. Texnologiya d\u00fcnyas\u0131 sad\u0259 dialoq interfeysl\u0259rind\u0259n m\u00fcr\u0259kk\u0259b m\u00fchakim\u0259 qabiliyy\u0259tin\u0259 malik, s\u0259rb\u0259st q\u0259rar ver\u0259 bil\u0259n [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":11395,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"om_disable_all_campaigns":false,"_monsterinsights_skip_tracking":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[1],"tags":[],"class_list":["post-11396","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-uncategorized"],"aioseo_notices":[],"jetpack_featured_media_url":"https:\/\/bitep.net\/blog\/wp-content\/uploads\/2026\/10\/bitep-ai-1791074441542-18.jpg","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/posts\/11396","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/comments?post=11396"}],"version-history":[{"count":0,"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/posts\/11396\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/media\/11395"}],"wp:attachment":[{"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/media?parent=11396"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/categories?post=11396"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/bitep.net\/blog\/wp-json\/wp\/v2\/tags?post=11396"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}