{"id":104448,"date":"2026-10-20T16:34:35","date_gmt":"2026-10-20T13:34:35","guid":{"rendered":"https:\/\/twodots.gr\/?p=104448"},"modified":"2026-10-20T16:34:35","modified_gmt":"2026-10-20T13:34:35","slug":"ai-agents-peperasmenes-katastaseis-provlepsi-apotychias","status":"publish","type":"post","link":"https:\/\/twodots.gr\/bg\/ai-agents-peperasmenes-katastaseis-provlepsi-apotychias\/","title":{"rendered":"AI \u0430\u0433\u0435\u043d\u0442\u0438 \u0432 \u043a\u0440\u0430\u0439\u043d\u0438 \u0441\u0438\u0442\u0443\u0430\u0446\u0438\u0438: \u043f\u0440\u043e\u0433\u043d\u043e\u0437\u0438\u0440\u0430\u043d\u0435 \u043d\u0430 \u0441\u043b\u0435\u0434\u0432\u0430\u0449\u0438\u044f \u0445\u043e\u0434 \u0438 \u043d\u0430 \u043f\u0440\u043e\u0432\u0430\u043b"},"content":{"rendered":"<p>\u0388\u03bd\u03b1\u03c2 AI agent \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c8\u03ac\u03c7\u03bd\u03b5\u03b9, \u03bd\u03b1 \u03ba\u03b1\u03bb\u03b5\u03af \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03b1, \u03bd\u03b1 \u03b5\u03c0\u03b5\u03be\u03b5\u03c1\u03b3\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03b1\u03c1\u03c7\u03b5\u03af\u03b1, \u03bd\u03b1 \u03b5\u03bb\u03ad\u03b3\u03c7\u03b5\u03b9 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03ba\u03b1\u03b9 \u03bd\u03b1 \u03b5\u03c0\u03b1\u03bd\u03b1\u03bb\u03b1\u03bc\u03b2\u03ac\u03bd\u03b5\u03b9 \u03c4\u03b1 \u03b2\u03ae\u03bc\u03b1\u03c4\u03ac \u03c4\u03bf\u03c5 \u03bc\u03ad\u03c7\u03c1\u03b9 \u03bd\u03b1 \u03bf\u03bb\u03bf\u03ba\u03bb\u03b7\u03c1\u03ce\u03c3\u03b5\u03b9 \u03bc\u03b9\u03b1 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1. \u038c\u03c3\u03bf \u03cc\u03bc\u03c9\u03c2 \u03c4\u03bf workflow \u03bc\u03b5\u03b3\u03b1\u03bb\u03ce\u03bd\u03b5\u03b9, \u03c4\u03cc\u03c3\u03bf \u03b4\u03c5\u03c3\u03ba\u03bf\u03bb\u03cc\u03c4\u03b5\u03c1\u03bf \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03b1\u03c0\u03b1\u03bd\u03c4\u03ae\u03c3\u03bf\u03c5\u03bc\u03b5 \u03c3\u03b5 \u03c4\u03c1\u03b5\u03b9\u03c2 \u03b1\u03c0\u03bb\u03ad\u03c2 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03bc\u03b1\u03c4\u03b9\u03ba\u03ad\u03c2 \u03b5\u03c1\u03c9\u03c4\u03ae\u03c3\u03b5\u03b9\u03c2: \u03c4\u03b9 \u03ba\u03ac\u03bd\u03b5\u03b9 \u03c4\u03ce\u03c1\u03b1 \u03bf agent, \u03c0\u03bf\u03b9\u03b1 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03b9\u03b8\u03b1\u03bd\u03cc \u03bd\u03b1 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03ae\u03c3\u03b5\u03b9 \u03ba\u03b1\u03b9 \u03c0\u03cc\u03c4\u03b5 \u03b7 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac \u03c4\u03bf\u03c5 \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03bf\u03b4\u03b7\u03b3\u03b5\u03af\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1;<\/p>\n<div class=\"td-note\">\n<p class=\"td-note-title\">\u0397 \u03c3\u03cd\u03bd\u03c4\u03bf\u03bc\u03b7 \u03b1\u03c0\u03ac\u03bd\u03c4\u03b7\u03c3\u03b7<\/p>\n<p>\u039c\u03b9\u03b1 finite-state machine \u03b4\u03b5\u03bd \u03ba\u03ac\u03bd\u03b5\u03b9 \u03c4\u03bf\u03bd AI agent \u03b1\u03c3\u03c6\u03b1\u03bb\u03ae \u03b1\u03c0\u03cc \u03bc\u03cc\u03bd\u03b7 \u03c4\u03b7\u03c2. \u039c\u03b5\u03c4\u03b1\u03c4\u03c1\u03ad\u03c0\u03b5\u03b9 \u03cc\u03bc\u03c9\u03c2 \u03c4\u03b1 agent traces \u03c3\u03b5 \u03bc\u03b9\u03ba\u03c1\u03cc \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03cc \u03c7\u03ac\u03c1\u03c4\u03b7, \u03ce\u03c3\u03c4\u03b5 \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03bd\u03b1 \u03b2\u03bb\u03ad\u03c0\u03b5\u03b9 \u03c0\u03bf\u03cd \u03b2\u03c1\u03af\u03c3\u03ba\u03b5\u03c4\u03b1\u03b9 \u03b7 \u03b5\u03ba\u03c4\u03ad\u03bb\u03b5\u03c3\u03b7, \u03c0\u03bf\u03b9\u03bf \u03b2\u03ae\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03b9\u03b8\u03b1\u03bd\u03cc \u03bd\u03b1 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03ae\u03c3\u03b5\u03b9 \u03ba\u03b1\u03b9 \u03c0\u03bf\u03b9\u03bf\u03b9 \u03b2\u03c1\u03cc\u03c7\u03bf\u03b9 \u03ae \u03b1\u03c3\u03c5\u03bd\u03ae\u03b8\u03b9\u03c3\u03c4\u03b5\u03c2 \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c3\u03c5\u03bd\u03b4\u03ad\u03bf\u03bd\u03c4\u03b1\u03b9 \u03bc\u03b5 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1.<\/p>\n<\/div>\n<p>\u0418\u0437\u0441\u043b\u0435\u0434\u043e\u0432\u0430\u0442\u0435\u043b\u0441\u043a\u0430\u0442\u0430 \u0440\u0430\u0431\u043e\u0442\u0430 <em>Automata from Agent Traces: Failure and Next-Step Prediction<\/em> \u03c0\u03c1\u03bf\u03c4\u03b5\u03af\u03bd\u03b5\u03b9 \u03bc\u03b9\u03b1 \u03b5\u03bd\u03c4\u03c5\u03c0\u03c9\u03c3\u03b9\u03b1\u03ba\u03ac \u03bb\u03b9\u03c4\u03ae \u03b1\u03c0\u03ac\u03bd\u03c4\u03b7\u03c3\u03b7. \u0391\u03bd\u03c4\u03af \u03bd\u03b1 \u03b1\u03bd\u03c4\u03b9\u03bc\u03b5\u03c4\u03c9\u03c0\u03af\u03b6\u03b5\u03b9 \u03ba\u03ac\u03b8\u03b5 \u03b5\u03ba\u03c4\u03ad\u03bb\u03b5\u03c3\u03b7 \u03c9\u03c2 \u03ad\u03bd\u03b1 \u03bc\u03b1\u03ba\u03c1\u03cd, \u03ac\u03bc\u03bf\u03c1\u03c6\u03bf log, \u03bc\u03b5\u03c4\u03b1\u03c4\u03c1\u03ad\u03c0\u03b5\u03b9 \u03ad\u03bd\u03b1 \u03bf\u03bb\u03cc\u03ba\u03bb\u03b7\u03c1\u03bf \u03c3\u03cd\u03bd\u03bf\u03bb\u03bf traces \u03c3\u03b5 \u03bc\u03af\u03b1 \u03c3\u03c5\u03bc\u03c0\u03b1\u03b3\u03ae \u03bc\u03b7\u03c7\u03b1\u03bd\u03ae \u03c0\u03b5\u03c0\u03b5\u03c1\u03b1\u03c3\u03bc\u03ad\u03bd\u03c9\u03bd \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03c9\u03bd, \u03ae finite-state machine (FSM). \u0397 \u03b2\u03b1\u03c3\u03b9\u03ba\u03ae \u03b9\u03b4\u03ad\u03b1 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03bf agent \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ac \u03ad\u03bd\u03b1 \u03b1\u03c0\u03bb\u03cc \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03bf. \u0395\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03b7 \u03b5\u03c0\u03b1\u03bd\u03b1\u03bb\u03b1\u03bc\u03b2\u03b1\u03bd\u03cc\u03bc\u03b5\u03bd\u03b7 \u03b4\u03bf\u03bc\u03ae \u03c4\u03c9\u03bd \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd \u03c4\u03bf\u03c5 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c0\u03c9\u03b8\u03b5\u03af \u03b1\u03c1\u03ba\u03b5\u03c4\u03ac \u03ba\u03b1\u03bb\u03ac \u03ce\u03c3\u03c4\u03b5 \u03bd\u03b1 \u03c5\u03c0\u03bf\u03c3\u03c4\u03b7\u03c1\u03af\u03be\u03b5\u03b9 \u03ad\u03bb\u03b5\u03b3\u03c7\u03bf, \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03ba\u03b1\u03b9 \u03c0\u03b1\u03c1\u03b1\u03ba\u03bf\u03bb\u03bf\u03cd\u03b8\u03b7\u03c3\u03b7.<\/p>\n<p>\u0393\u03b9\u03b1 \u03b5\u03c0\u03b1\u03b3\u03b3\u03b5\u03bb\u03bc\u03b1\u03c4\u03af\u03b5\u03c2, e-commerce \u03bf\u03bc\u03ac\u03b4\u03b5\u03c2 \u03ba\u03b1\u03b9 \u03bf\u03c1\u03b3\u03b1\u03bd\u03b9\u03c3\u03bc\u03bf\u03cd\u03c2 \u03c0\u03bf\u03c5 \u03c7\u03c4\u03af\u03b6\u03bf\u03c5\u03bd agentic workflows, \u03b7 \u03b1\u03be\u03af\u03b1 \u03b2\u03c1\u03af\u03c3\u03ba\u03b5\u03c4\u03b1\u03b9 \u03c3\u03c4\u03b7 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03ae \u03bf\u03c1\u03b1\u03c4\u03cc\u03c4\u03b7\u03c4\u03b1. \u0388\u03bd\u03b1 \u03bc\u03b9\u03ba\u03c1\u03cc FSM \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03b5\u03af\u03be\u03b5\u03b9 \u03b1\u03bd \u03bf agent \u03ba\u03b9\u03bd\u03b5\u03af\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03b1\u03bd\u03b1\u03bc\u03b5\u03bd\u03cc\u03bc\u03b5\u03bd\u03b7 \u03b4\u03b9\u03b1\u03b4\u03c1\u03bf\u03bc\u03ae, \u03b1\u03bd \u03ad\u03c7\u03b5\u03b9 \u03ba\u03bf\u03bb\u03bb\u03ae\u03c3\u03b5\u03b9 \u03c3\u03b5 \u03ba\u03cd\u03ba\u03bb\u03bf, \u03b1\u03bd \u03c0\u03b1\u03c1\u03b1\u03bb\u03b5\u03af\u03c0\u03b5\u03b9 \u03ba\u03c1\u03af\u03c3\u03b9\u03bc\u03b1 \u03b2\u03ae\u03bc\u03b1\u03c4\u03b1 \u03ae \u03b1\u03bd \u03b1\u03c0\u03bf\u03ba\u03bb\u03af\u03bd\u03b5\u03b9 \u03b1\u03c0\u03cc \u03c4\u03b7 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac \u03c0\u03bf\u03c5 \u03ad\u03c7\u03b5\u03b9 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03b7\u03b8\u03b5\u03af \u03c3\u03b5 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b5\u03c2 \u03ba\u03b1\u03b9 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b5\u03c2 \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b5\u03b9\u03c2. \u03a4\u03b1 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c5\u03c0\u03bf\u03c3\u03c7\u03cc\u03bc\u03b5\u03bd\u03b1, \u03b1\u03bb\u03bb\u03ac \u03b7 \u03af\u03b4\u03b9\u03b1 \u03b7 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03bf\u03c1\u03af\u03b6\u03b5\u03b9 \u03c3\u03b1\u03c6\u03ae \u03cc\u03c1\u03b9\u03b1 \u03c0\u03bf\u03c5 \u03b4\u03b5\u03bd \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c7\u03b1\u03b8\u03bf\u03cd\u03bd \u03bc\u03ad\u03c3\u03b1 \u03c3\u03c4\u03bf\u03c5\u03c2 \u03b5\u03bd\u03c4\u03c5\u03c0\u03c9\u03c3\u03b9\u03b1\u03ba\u03bf\u03cd\u03c2 \u03b1\u03c1\u03b9\u03b8\u03bc\u03bf\u03cd\u03c2.<\/p>\n<div class=\"td-article-toc\">\n<div class=\"td-toc-title\">\u0421\u044a\u0434\u044a\u0440\u0436\u0430\u043d\u0438\u0435<\/div>\n<ul>\n<li><a href=\"#provlima-agent-traces\">\u03a4\u03bf \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1 \u03c0\u03bf\u03c5 \u03ba\u03c1\u03cd\u03b2\u03b5\u03c4\u03b1\u03b9 \u03bc\u03ad\u03c3\u03b1 \u03c3\u03c4\u03b1 agent traces<\/a><\/li>\n<li><a href=\"#istoriko-mikro-fsm\">\u03a0\u03ce\u03c2 \u03ad\u03bd\u03b1 \u03bc\u03b1\u03ba\u03c1\u03cd \u03b9\u03c3\u03c4\u03bf\u03c1\u03b9\u03ba\u03cc \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03ad\u03bd\u03b1 \u03bc\u03b9\u03ba\u03c1\u03cc FSM<\/a><\/li>\n<li><a href=\"#datasets-sympiesi\">\u0394\u03ce\u03b4\u03b5\u03ba\u03b1 datasets \u03ba\u03b1\u03b9 \u03bf \u03bb\u03cc\u03b3\u03bf\u03c2 \u03c0\u03bf\u03c5 \u03b7 \u03c3\u03c5\u03bc\u03c0\u03af\u03b5\u03c3\u03b7 \u03ad\u03c7\u03b5\u03b9 \u03c3\u03b7\u03bc\u03b1\u03c3\u03af\u03b1<\/a><\/li>\n<li><a href=\"#provlepsi-epomenis-energeias\">\u0397 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03c4\u03b7\u03c2 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1\u03c2 \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03b4\u03bf\u03bc\u03b9\u03ba\u03cc \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1<\/a><\/li>\n<li><a href=\"#provlepsi-apotychias\">\u0391\u03c0\u03cc \u03c4\u03bf \u00ab\u03c4\u03b9 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03b5\u03af\u00bb \u03c3\u03c4\u03bf \u00ab\u03b8\u03b1 \u03b1\u03c0\u03bf\u03c4\u03cd\u03c7\u03b5\u03b9;\u00bb<\/a><\/li>\n<li><a href=\"#runtime-monitoring-proimo-stamatima\">Runtime monitoring \u03ba\u03b1\u03b9 \u03ad\u03b3\u03ba\u03b1\u03b9\u03c1\u03bf \u03c3\u03c4\u03b1\u03bc\u03ac\u03c4\u03b7\u03bc\u03b1<\/a><\/li>\n<li><a href=\"#harness-vs-montelo\">\u0397 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac \u03b5\u03c0\u03b7\u03c1\u03b5\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c0\u03b5\u03c1\u03b9\u03c3\u03c3\u03cc\u03c4\u03b5\u03c1\u03bf \u03b1\u03c0\u03cc \u03c4\u03bf harness \u03c0\u03b1\u03c1\u03ac \u03b1\u03c0\u03cc \u03c4\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf<\/a><\/li>\n<li><a href=\"#ecommerce-epicheirisiakoi-agents\">\u03a4\u03b9 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 \u03b3\u03b9\u03b1 e-commerce \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03bf\u03cd\u03c2 agents<\/a><\/li>\n<li><a href=\"#periorismoi-fsm\">\u039f\u03b9 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03bf\u03af \u03c0\u03bf\u03c5 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03bc\u03b5\u03af\u03bd\u03bf\u03c5\u03bd \u03c3\u03c4\u03bf \u03c0\u03c1\u03bf\u03c3\u03ba\u03ae\u03bd\u03b9\u03bf<\/a><\/li>\n<li><a href=\"#diakyvernisi-agentic-ai\">\u0388\u03bd\u03b1 \u03c0\u03b9\u03bf \u03c0\u03c1\u03b1\u03ba\u03c4\u03b9\u03ba\u03cc \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03b4\u03b9\u03b1\u03ba\u03c5\u03b2\u03ad\u03c1\u03bd\u03b7\u03c3\u03b7\u03c2 agentic AI<\/a><\/li>\n<\/ul>\n<\/div>\n<h2 id=\"provlima-agent-traces\">\u03a4\u03bf \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1 \u03c0\u03bf\u03c5 \u03ba\u03c1\u03cd\u03b2\u03b5\u03c4\u03b1\u03b9 \u03bc\u03ad\u03c3\u03b1 \u03c3\u03c4\u03b1 agent traces<\/h2>\n<p>\u039f\u03b9 \u03c3\u03cd\u03b3\u03c7\u03c1\u03bf\u03bd\u03bf\u03b9 agents \u03b4\u03b5\u03bd \u03c0\u03b1\u03c1\u03ac\u03b3\u03bf\u03c5\u03bd \u03bc\u03cc\u03bd\u03bf \u03ad\u03bd\u03b1 \u03c4\u03b5\u03bb\u03b9\u03ba\u03cc \u03ba\u03b5\u03af\u03bc\u03b5\u03bd\u03bf. \u0395\u03ba\u03c4\u03b5\u03bb\u03bf\u03cd\u03bd \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03af\u03b5\u03c2 \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd: \u03bb\u03b1\u03bc\u03b2\u03ac\u03bd\u03bf\u03c5\u03bd \u03bf\u03b4\u03b7\u03b3\u03af\u03b5\u03c2, \u03ba\u03ac\u03bd\u03bf\u03c5\u03bd tool calls, \u03b1\u03bd\u03b1\u03b6\u03b7\u03c4\u03bf\u03cd\u03bd \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b5\u03c2, \u03b5\u03c0\u03b5\u03be\u03b5\u03c1\u03b3\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03b1, \u03b3\u03c1\u03ac\u03c6\u03bf\u03c5\u03bd \u03ae \u03b1\u03bb\u03bb\u03ac\u03b6\u03bf\u03c5\u03bd \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1 \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c3\u03c4\u03c1\u03ad\u03c6\u03bf\u03c5\u03bd \u03c3\u03b5 \u03c0\u03c1\u03bf\u03b7\u03b3\u03bf\u03cd\u03bc\u03b5\u03bd\u03b1 \u03c3\u03c4\u03ac\u03b4\u03b9\u03b1. \u0388\u03bd\u03b1\u03c2 coding agent \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03bd\u03b1\u03bb\u03bb\u03ac\u03c3\u03c3\u03b5\u03c4\u03b1\u03b9 \u03bc\u03b5\u03c4\u03b1\u03be\u03cd \u03b1\u03bd\u03b1\u03b6\u03ae\u03c4\u03b7\u03c3\u03b7\u03c2, \u03b5\u03c0\u03b5\u03be\u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1\u03c2 \u03ba\u03b1\u03b9 \u03b5\u03ba\u03c4\u03ad\u03bb\u03b5\u03c3\u03b7\u03c2. \u0388\u03bd\u03b1\u03c2 agent \u03b5\u03be\u03c5\u03c0\u03b7\u03c1\u03ad\u03c4\u03b7\u03c3\u03b7\u03c2 \u03c0\u03b5\u03bb\u03b1\u03c4\u03ce\u03bd \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03ba\u03b9\u03bd\u03b5\u03af\u03c4\u03b1\u03b9 \u03bc\u03b5\u03c4\u03b1\u03be\u03cd \u03b5\u03c0\u03b9\u03ba\u03bf\u03b9\u03bd\u03c9\u03bd\u03af\u03b1\u03c2 \u03bc\u03b5 \u03c4\u03bf\u03bd \u03c7\u03c1\u03ae\u03c3\u03c4\u03b7 \u03ba\u03b1\u03b9 \u03b5\u03c1\u03c9\u03c4\u03b7\u03bc\u03ac\u03c4\u03c9\u03bd \u03c3\u03b5 \u03c3\u03c5\u03c3\u03c4\u03ae\u03bc\u03b1\u03c4\u03b1 \u03c0\u03b1\u03c1\u03b1\u03b3\u03b3\u03b5\u03bb\u03b9\u03ce\u03bd \u03ae \u03bb\u03bf\u03b3\u03b1\u03c1\u03b9\u03b1\u03c3\u03bc\u03ce\u03bd.<\/p>\n<p>\u0391\u03c5\u03c4\u03ae \u03b7 \u03b4\u03bf\u03bc\u03ae \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03c3\u03c5\u03bd\u03ae\u03b8\u03c9\u03c2 \u03b3\u03c1\u03b1\u03bc\u03bc\u03ad\u03bd\u03b7 \u03c3\u03b5 \u03ad\u03bd\u03b1 \u03b5\u03b3\u03c7\u03b5\u03b9\u03c1\u03af\u03b4\u03b9\u03bf. \u03a0\u03c1\u03bf\u03ba\u03cd\u03c0\u03c4\u03b5\u03b9 \u03b1\u03c0\u03cc \u03c4\u03bf system prompt, \u03c4\u03b1 \u03b4\u03b9\u03b1\u03b8\u03ad\u03c3\u03b9\u03bc\u03b1 \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03b1, \u03c4\u03b7 \u03b4\u03b9\u03b5\u03c0\u03b1\u03c6\u03ae \u03c4\u03bf\u03c5 agent \u03ba\u03b1\u03b9 \u03c4\u03bf \u03b5\u03af\u03b4\u03bf\u03c2 \u03c4\u03c9\u03bd \u03b5\u03c1\u03b3\u03b1\u03c3\u03b9\u03ce\u03bd. \u03a4\u03bf \u03b1\u03c0\u03bf\u03c4\u03ad\u03bb\u03b5\u03c3\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03b4\u03cd\u03bf \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b5\u03b9\u03c2 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c6\u03b1\u03af\u03bd\u03bf\u03bd\u03c4\u03b1\u03b9 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03ad\u03c2 \u03c3\u03b5 \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03bf \u03ba\u03b5\u03b9\u03bc\u03ad\u03bd\u03bf\u03c5, \u03b1\u03bb\u03bb\u03ac \u03bd\u03b1 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03bf\u03cd\u03bd \u03c3\u03c7\u03b5\u03b4\u03cc\u03bd \u03c4\u03b7\u03bd \u03af\u03b4\u03b9\u03b1 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03ae \u03c4\u03bf\u03c0\u03bf\u03bb\u03bf\u03b3\u03af\u03b1. \u0391\u03bd\u03c4\u03af\u03c3\u03c4\u03c1\u03bf\u03c6\u03b1, \u03b4\u03cd\u03bf traces \u03af\u03b4\u03b9\u03bf\u03c5 \u03bc\u03ae\u03ba\u03bf\u03c5\u03c2 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03ad\u03c7\u03bf\u03c5\u03bd \u03b5\u03bd\u03c4\u03b5\u03bb\u03ce\u03c2 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03ae \u03c3\u03b7\u03bc\u03b1\u03c3\u03af\u03b1 \u03b5\u03c0\u03b5\u03b9\u03b4\u03ae \u03bf\u03b9 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b5\u03c2 \u03ad\u03b3\u03b9\u03bd\u03b1\u03bd \u03c3\u03b5 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03ae \u03c3\u03b5\u03b9\u03c1\u03ac.<\/p>\n<p>\u0397 \u03c3\u03c5\u03bd\u03b7\u03b8\u03b9\u03c3\u03bc\u03ad\u03bd\u03b7 \u03c0\u03b1\u03c1\u03b1\u03ba\u03bf\u03bb\u03bf\u03cd\u03b8\u03b7\u03c3\u03b7 \u03bc\u03b5 \u03c3\u03c5\u03bd\u03bf\u03bb\u03b9\u03ba\u03ac \u03bc\u03b5\u03b3\u03ad\u03b8\u03b7 ,\u03bc\u03ae\u03ba\u03bf\u03c2 trace, \u03c0\u03bb\u03ae\u03b8\u03bf\u03c2 errors \u03ae \u03b1\u03c1\u03b9\u03b8\u03bc\u03cc\u03c2 tool calls, \u03c7\u03ac\u03bd\u03b5\u03b9 \u03b1\u03c5\u03c4\u03ae \u03c4\u03b7 \u03b4\u03b9\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7. \u0394\u03b5\u03bd \u03b1\u03c0\u03bf\u03ba\u03b1\u03bb\u03cd\u03c0\u03c4\u03b5\u03b9 \u03c3\u03b5 \u03c0\u03bf\u03b9\u03bf \u03c3\u03b7\u03bc\u03b5\u03af\u03bf \u03b4\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03ae\u03b8\u03b7\u03ba\u03b5 \u03c4\u03bf \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1 \u03bf\u03cd\u03c4\u03b5 \u03c0\u03bf\u03b9\u03b1 \u03bc\u03b5\u03c4\u03ac\u03b2\u03b1\u03c3\u03b7 \u03ae\u03c4\u03b1\u03bd \u03b1\u03c3\u03c5\u03bd\u03ae\u03b8\u03b9\u03c3\u03c4\u03b7. \u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b1\u03bd\u03c4\u03b9\u03bc\u03b5\u03c4\u03c9\u03c0\u03af\u03b6\u03b5\u03b9 \u03c4\u03b7\u03bd \u03b1\u03bd\u03ac\u03ba\u03c4\u03b7\u03c3\u03b7 \u03c4\u03b7\u03c2 \u03ba\u03c1\u03c5\u03c6\u03ae\u03c2 \u03b4\u03bf\u03bc\u03ae\u03c2 \u03c9\u03c2 \u03b1\u03bd\u03c4\u03af\u03c3\u03c4\u03c1\u03bf\u03c6\u03bf \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1: \u03b1\u03c0\u03cc \u03c4\u03b1 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03b7\u03bc\u03ad\u03bd\u03b1 traces \u03c0\u03c1\u03bf\u03c3\u03c0\u03b1\u03b8\u03b5\u03af \u03bd\u03b1 \u03ba\u03b1\u03c4\u03b1\u03c3\u03ba\u03b5\u03c5\u03ac\u03c3\u03b5\u03b9 \u03ad\u03bd\u03b1 \u03bc\u03b9\u03ba\u03c1\u03cc \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03c0\u03bf\u03c5 \u03b5\u03be\u03b7\u03b3\u03b5\u03af \u03c4\u03b9\u03c2 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03af\u03b5\u03c2 \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd.<\/p>\n<h2 id=\"istoriko-mikro-fsm\">\u03a0\u03ce\u03c2 \u03ad\u03bd\u03b1 \u03bc\u03b1\u03ba\u03c1\u03cd \u03b9\u03c3\u03c4\u03bf\u03c1\u03b9\u03ba\u03cc \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03ad\u03bd\u03b1 \u03bc\u03b9\u03ba\u03c1\u03cc FSM<\/h2>\n<p>\u0397 \u03b4\u03b9\u03b1\u03b4\u03b9\u03ba\u03b1\u03c3\u03af\u03b1 \u03be\u03b5\u03ba\u03b9\u03bd\u03ac \u03bc\u03b5 \u03c4\u03b7\u03bd \u03b5\u03be\u03b1\u03b3\u03c9\u03b3\u03ae \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03bf\u03c4\u03ae\u03c4\u03c9\u03bd \u03b1\u03c0\u03cc \u03ba\u03ac\u03b8\u03b5 \u03bc\u03ae\u03bd\u03c5\u03bc\u03b1 \u03c4\u03bf\u03c5 trace. \u0397 \u03c0\u03c1\u03bf\u03c4\u03b5\u03c1\u03b1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b4\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03c3\u03c4\u03b1 \u03bf\u03bd\u03cc\u03bc\u03b1\u03c4\u03b1 \u03c4\u03c9\u03bd tool calls. \u038c\u03c4\u03b1\u03bd \u03c5\u03c0\u03ac\u03c1\u03c7\u03bf\u03c5\u03bd action tags, \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03bf\u03cd\u03bd\u03c4\u03b1\u03b9 \u03bf\u03b9 \u03b1\u03bd\u03c4\u03af\u03c3\u03c4\u03bf\u03b9\u03c7\u03b5\u03c2 \u03b5\u03c4\u03b9\u03ba\u03ad\u03c4\u03b5\u03c2, \u03b5\u03bd\u03ce \u03c3\u03b5 \u03c0\u03b5\u03c1\u03b9\u03b2\u03ac\u03bb\u03bb\u03bf\u03bd\u03c4\u03b1 \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1 \u03c4\u03bf \u03c0\u03c1\u03ce\u03c4\u03bf command token \u03b1\u03bd\u03c4\u03b9\u03c3\u03c4\u03bf\u03b9\u03c7\u03af\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03bc\u03b9\u03b1 \u03c3\u03b7\u03bc\u03b1\u03c3\u03b9\u03bf\u03bb\u03bf\u03b3\u03b9\u03ba\u03ae \u03ba\u03b1\u03c4\u03b7\u03b3\u03bf\u03c1\u03af\u03b1. \u0391\u03bd \u03b4\u03b5\u03bd \u03b9\u03c3\u03c7\u03cd\u03b5\u03b9 \u03ba\u03ac\u03c0\u03bf\u03b9\u03bf\u03c2 \u03b1\u03c0\u03cc \u03b1\u03c5\u03c4\u03bf\u03cd\u03c2 \u03c4\u03bf\u03c5\u03c2 \u03ba\u03b1\u03bd\u03cc\u03bd\u03b5\u03c2, \u03b7 \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1 \u03c0\u03b5\u03c1\u03b9\u03b3\u03c1\u03ac\u03c6\u03b5\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03c4\u03bf\u03bd \u03c1\u03cc\u03bb\u03bf \u03ba\u03b1\u03b9 \u03c4\u03bf\u03bd \u03c4\u03cd\u03c0\u03bf \u03c0\u03b5\u03c1\u03b9\u03b5\u03c7\u03bf\u03bc\u03ad\u03bd\u03bf\u03c5, \u03b3\u03b9\u03b1 \u03c0\u03b1\u03c1\u03ac\u03b4\u03b5\u03b9\u03b3\u03bc\u03b1 assistant:text.<\/p>\n<p>\u03a3\u03c4\u03bf \u03c0\u03c1\u03ce\u03c4\u03bf \u03ba\u03b1\u03c4\u03b1\u03c3\u03ba\u03b5\u03c5\u03b1\u03c3\u03c4\u03b9\u03ba\u03cc \u03b2\u03ae\u03bc\u03b1, \u03cc\u03bb\u03b5\u03c2 \u03bf\u03b9 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03af\u03b5\u03c2 \u03b5\u03b9\u03c3\u03ac\u03b3\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03ad\u03bd\u03b1 prefix tree. \u039a\u03ac\u03b8\u03b5 \u03bc\u03bf\u03bd\u03b1\u03b4\u03b9\u03ba\u03cc \u03c0\u03c1\u03cc\u03b8\u03b5\u03bc\u03b1 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03b5\u03af \u03be\u03b5\u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03ae \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7. \u0391\u03c5\u03c4\u03cc \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03ac\u03b3\u03b5\u03b9 \u03c4\u03ad\u03bb\u03b5\u03b9\u03b1 \u03c4\u03b1 \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03b1 \u03b5\u03ba\u03c0\u03b1\u03af\u03b4\u03b5\u03c5\u03c3\u03b7\u03c2, \u03b1\u03bb\u03bb\u03ac \u03b4\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03b5\u03af \u03ad\u03bd\u03b1 \u03bc\u03b5\u03b3\u03ac\u03bb\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03c0\u03bf\u03c5 \u03bf\u03c5\u03c3\u03b9\u03b1\u03c3\u03c4\u03b9\u03ba\u03ac \u03b1\u03c0\u03bf\u03bc\u03bd\u03b7\u03bc\u03bf\u03bd\u03b5\u03cd\u03b5\u03b9 \u03b4\u03b9\u03b1\u03b4\u03c1\u03bf\u03bc\u03ad\u03c2. \u03a3\u03c4\u03bf \u03b4\u03b5\u03cd\u03c4\u03b5\u03c1\u03bf \u03b2\u03ae\u03bc\u03b1, \u03bf\u03b9 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c0\u03bf\u03c5 \u03c0\u03c1\u03bf\u03c3\u03b5\u03b3\u03b3\u03af\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03af\u03b4\u03b9\u03b1 \u03c4\u03b5\u03bb\u03b5\u03c5\u03c4\u03b1\u03af\u03b1 \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1 \u03c3\u03c5\u03b3\u03c7\u03c9\u03bd\u03b5\u03cd\u03bf\u03bd\u03c4\u03b1\u03b9. \u0388\u03c4\u03c3\u03b9, \u03bf\u03b9 \u03b5\u03c0\u03b1\u03bd\u03b1\u03bb\u03b1\u03bc\u03b2\u03b1\u03bd\u03cc\u03bc\u03b5\u03bd\u03b5\u03c2 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03af\u03b5\u03c2 \u03cc\u03c0\u03c9\u03c2 search\u2013edit\u2013execute \u03bc\u03b5\u03c4\u03b1\u03c4\u03c1\u03ad\u03c0\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03ba\u03b1\u03b8\u03b1\u03c1\u03bf\u03cd\u03c2 \u03b2\u03c1\u03cc\u03c7\u03bf\u03c5\u03c2 \u03ba\u03b1\u03b9 \u03b7 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 \u03c3\u03c5\u03bd\u03b4\u03ad\u03b5\u03c4\u03b1\u03b9 \u03bc\u03b5 \u03c4\u03bf \u03b5\u03af\u03b4\u03bf\u03c2 \u03c4\u03b7\u03c2 \u03c4\u03b5\u03bb\u03b5\u03c5\u03c4\u03b1\u03af\u03b1\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1\u03c2.<\/p>\n<p>\u03a3\u03c4\u03bf \u03c4\u03c1\u03af\u03c4\u03bf \u03b2\u03ae\u03bc\u03b1 \u03c6\u03b9\u03bb\u03c4\u03c1\u03ac\u03c1\u03bf\u03bd\u03c4\u03b1\u03b9 \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c0\u03bf\u03c5 \u03b5\u03bc\u03c6\u03b1\u03bd\u03af\u03c3\u03c4\u03b7\u03ba\u03b1\u03bd \u03bc\u03cc\u03bd\u03bf \u03bc\u03af\u03b1 \u03c6\u03bf\u03c1\u03ac, \u03b5\u03ba\u03c4\u03cc\u03c2 \u03b1\u03bd \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03bf\u03cd\u03bd \u03c4\u03b7 \u03bc\u03bf\u03bd\u03b1\u03b4\u03b9\u03ba\u03ae \u03c3\u03c5\u03bd\u03ad\u03c7\u03b5\u03b9\u03b1 \u03c4\u03b7\u03c2 \u03c0\u03b7\u03b3\u03b1\u03af\u03b1\u03c2 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7\u03c2. \u039f \u03c3\u03c4\u03cc\u03c7\u03bf\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bd\u03b1 \u03b1\u03c6\u03b1\u03b9\u03c1\u03b5\u03b8\u03bf\u03cd\u03bd \u03bc\u03b5\u03bc\u03bf\u03bd\u03c9\u03bc\u03ad\u03bd\u03b5\u03c2 \u03c0\u03b1\u03c1\u03b5\u03ba\u03ba\u03bb\u03af\u03c3\u03b5\u03b9\u03c2 \u03c7\u03c9\u03c1\u03af\u03c2 \u03bd\u03b1 \u03b5\u03be\u03b1\u03c6\u03b1\u03bd\u03b9\u03c3\u03c4\u03bf\u03cd\u03bd \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2. \u0397 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 \u03c4\u03bf\u03bd\u03af\u03b6\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03b1\u03c5\u03c4\u03cc \u03b5\u03af\u03bd\u03b1\u03b9 \u03c4\u03bf \u03bc\u03cc\u03bd\u03bf \u03b2\u03ae\u03bc\u03b1 \u03c0\u03bf\u03c5 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03bc\u03b5\u03b9\u03ce\u03c3\u03b5\u03b9 \u03c4\u03bf replay fitness, \u03b4\u03b7\u03bb\u03b1\u03b4\u03ae \u03c4\u03bf \u03c0\u03bf\u03c3\u03bf\u03c3\u03c4\u03cc \u03c4\u03bf\u03c5 trace \u03c0\u03bf\u03c5 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03b1\u03c7\u03b8\u03b5\u03af \u03bc\u03ad\u03c3\u03b1 \u03c3\u03c4\u03bf FSM.<\/p>\n<p>\u0397 \u03ba\u03b1\u03c4\u03b1\u03c3\u03ba\u03b5\u03c5\u03ae \u03b5\u03af\u03bd\u03b1\u03b9 \u03bd\u03c4\u03b5\u03c4\u03b5\u03c1\u03bc\u03b9\u03bd\u03b9\u03c3\u03c4\u03b9\u03ba\u03ae \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af \u03c5\u03c0\u03b5\u03c1\u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03c4\u03c1\u03bf\u03c5\u03c2 \u03b5\u03ba\u03bc\u03ac\u03b8\u03b7\u03c3\u03b7\u03c2. \u0397 \u03b5\u03c0\u03b9\u03bb\u03bf\u03b3\u03ae \u03c0\u03bf\u03c5 \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bf \u03c4\u03c1\u03cc\u03c0\u03bf\u03c2 \u03b5\u03be\u03b1\u03b3\u03c9\u03b3\u03ae\u03c2 \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03bf\u03c4\u03ae\u03c4\u03c9\u03bd. \u03a3\u03c4\u03b1 \u03c0\u03b5\u03b9\u03c1\u03ac\u03bc\u03b1\u03c4\u03b1, \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03ac \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03b1 \u03bb\u03b5\u03c0\u03c4\u03bf\u03bc\u03ad\u03c1\u03b5\u03b9\u03b1\u03c2 \u03b4\u03b9\u03b1\u03c4\u03ae\u03c1\u03b7\u03c3\u03b1\u03bd replay fitness \u03c4\u03bf\u03c5\u03bb\u03ac\u03c7\u03b9\u03c3\u03c4\u03bf\u03bd 0,999 \u03c3\u03b5 \u03cc\u03bb\u03b1 \u03c4\u03b1 datasets, \u03b1\u03bd \u03ba\u03b1\u03b9 \u03b7 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b3\u03b9\u03b1 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03b5\u03be\u03b1\u03c1\u03c4\u03ac\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03c4\u03bf \u03b1\u03bd \u03c4\u03bf alphabet \u03c4\u03c9\u03bd \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03b1\u03c1\u03ba\u03b5\u03c4\u03ac \u03c0\u03bb\u03bf\u03cd\u03c3\u03b9\u03bf.<\/p>\n<h2 id=\"datasets-sympiesi\">\u0394\u03ce\u03b4\u03b5\u03ba\u03b1 datasets \u03ba\u03b1\u03b9 \u03bf \u03bb\u03cc\u03b3\u03bf\u03c2 \u03c0\u03bf\u03c5 \u03b7 \u03c3\u03c5\u03bc\u03c0\u03af\u03b5\u03c3\u03b7 \u03ad\u03c7\u03b5\u03b9 \u03c3\u03b7\u03bc\u03b1\u03c3\u03af\u03b1<\/h2>\n<p>\u0397 \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7 \u03ba\u03ac\u03bb\u03c5\u03c8\u03b5 \u03b4\u03ce\u03b4\u03b5\u03ba\u03b1 \u03b4\u03b7\u03bc\u03cc\u03c3\u03b9\u03b1 datasets \u03ba\u03b1\u03b9 \u03bf\u03ba\u03c4\u03ce \u03c0\u03b5\u03b4\u03af\u03b1 \u03c7\u03c1\u03ae\u03c3\u03b7\u03c2, \u03bc\u03b5\u03c4\u03b1\u03be\u03cd \u03c4\u03c9\u03bd \u03bf\u03c0\u03bf\u03af\u03c9\u03bd coding agents, web navigation, desktop GUI, customer service, telecom \u03ba\u03b1\u03b9 safety tasks. \u03a4\u03b1 activity alphabets \u03b5\u03af\u03c7\u03b1\u03bd \u03b1\u03c0\u03cc 6 \u03ad\u03c9\u03c2 42 \u03c3\u03cd\u03bc\u03b2\u03bf\u03bb\u03b1. \u0395\u03bd\u03bd\u03ad\u03b1 datasets \u03b4\u03b9\u03ad\u03b8\u03b5\u03c4\u03b1\u03bd \u03ba\u03b1\u03c4\u03ac\u03bb\u03bb\u03b7\u03bb\u03b5\u03c2 \u03b5\u03c4\u03b9\u03ba\u03ad\u03c4\u03b5\u03c2 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03bf\u03c2 \u03ba\u03b1\u03b9 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03ae\u03b8\u03b7\u03ba\u03b1\u03bd \u03b3\u03b9\u03b1 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1\u03c2, \u03b5\u03bd\u03ce \u03c4\u03b1 \u03c5\u03c0\u03cc\u03bb\u03bf\u03b9\u03c0\u03b1 \u03c3\u03c5\u03bd\u03ad\u03b2\u03b1\u03bb\u03b1\u03bd \u03ba\u03c5\u03c1\u03af\u03c9\u03c2 \u03c3\u03c4\u03b7\u03bd \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7 \u03c3\u03c5\u03bc\u03c0\u03af\u03b5\u03c3\u03b7\u03c2 \u03ba\u03b1\u03b9 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03bf\u03c5 \u03b2\u03ae\u03bc\u03b1\u03c4\u03bf\u03c2.<\/p>\n<p>\u03a4\u03b1 \u03c4\u03b5\u03bb\u03b9\u03ba\u03ac FSM \u03b5\u03af\u03c7\u03b1\u03bd \u03bc\u03cc\u03bb\u03b9\u03c2 7 \u03ad\u03c9\u03c2 43 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2. \u03a0\u03b1\u03c1\u03ac \u03c4\u03bf \u03bc\u03b9\u03ba\u03c1\u03cc \u03bc\u03ad\u03b3\u03b5\u03b8\u03bf\u03c2, \u03c0\u03ad\u03c4\u03c5\u03c7\u03b1\u03bd test replay fitness \u03c4\u03bf\u03c5\u03bb\u03ac\u03c7\u03b9\u03c3\u03c4\u03bf\u03bd 0,997. \u03a3\u03b5 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03bc\u03b5 \u03c4\u03bf RPNI, \u03b7 \u03c3\u03c5\u03bc\u03c0\u03af\u03b5\u03c3\u03b7 \u03ba\u03c5\u03bc\u03ac\u03bd\u03b8\u03b7\u03ba\u03b5 \u03b1\u03c0\u03cc 15 \u03c6\u03bf\u03c1\u03ad\u03c2 \u03ad\u03c9\u03c2 3.036 \u03c6\u03bf\u03c1\u03ad\u03c2 \u03cc\u03c4\u03b1\u03bd \u03c3\u03c5\u03bd\u03c5\u03c0\u03bf\u03bb\u03bf\u03b3\u03af\u03c3\u03c4\u03b7\u03ba\u03b1\u03bd \u03ba\u03b1\u03b9 \u03c4\u03b1 unlabeled datasets. \u03a3\u03c4\u03bf SWE-agent, \u03b3\u03b9\u03b1 \u03c0\u03b1\u03c1\u03ac\u03b4\u03b5\u03b9\u03b3\u03bc\u03b1, \u03c4\u03bf \u03c0\u03c1\u03bf\u03c4\u03b5\u03b9\u03bd\u03cc\u03bc\u03b5\u03bd\u03bf FSM \u03b5\u03af\u03c7\u03b5 25 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03ba\u03b1\u03b9 fitness 0,999, \u03b5\u03bd\u03ce \u03c4\u03bf RPNI \u03ad\u03c6\u03c4\u03b1\u03c3\u03b5 \u03c4\u03b9\u03c2 59.510 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03ba\u03b1\u03b9 fitness 0,646 \u03bc\u03ad\u03c3\u03b1 \u03c3\u03c4\u03bf\u03c5\u03c2 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03bf\u03cd\u03c2 \u03c4\u03b7\u03c2 \u03b4\u03bf\u03ba\u03b9\u03bc\u03ae\u03c2.<\/p>\n<p>\u0397 \u03c3\u03c5\u03bc\u03c0\u03af\u03b5\u03c3\u03b7 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03b1\u03c0\u03bb\u03ce\u03c2 \u03b8\u03ad\u03bc\u03b1 \u03b1\u03c0\u03bf\u03b8\u03ae\u03ba\u03b5\u03c5\u03c3\u03b7\u03c2. \u039c\u03b5 \u03bb\u03af\u03b3\u03b5\u03c2 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2, \u03ba\u03ac\u03b8\u03b5 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 \u03c3\u03c5\u03b3\u03ba\u03b5\u03bd\u03c4\u03c1\u03ce\u03bd\u03b5\u03b9 \u03b1\u03c1\u03ba\u03b5\u03c4\u03ad\u03c2 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03ae\u03c3\u03b5\u03b9\u03c2 \u03ce\u03c3\u03c4\u03b5 \u03bf\u03b9 \u03c0\u03b9\u03b8\u03b1\u03bd\u03cc\u03c4\u03b7\u03c4\u03b5\u03c2 \u03bc\u03b5\u03c4\u03ac\u03b2\u03b1\u03c3\u03b7\u03c2 \u03ba\u03b1\u03b9 \u03c4\u03b1 \u03c3\u03ae\u03bc\u03b1\u03c4\u03b1 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1\u03c2 \u03bd\u03b1 \u03b5\u03ba\u03c4\u03b9\u03bc\u03ce\u03bd\u03c4\u03b1\u03b9 \u03c0\u03b9\u03bf \u03b1\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03b1. \u0388\u03bd\u03b1 \u03c4\u03b5\u03c1\u03ac\u03c3\u03c4\u03b9\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03c0\u03bf\u03c5 \u03b8\u03c5\u03bc\u03ac\u03c4\u03b1\u03b9 \u03c3\u03c7\u03b5\u03b4\u03cc\u03bd \u03ba\u03ac\u03b8\u03b5 trace \u03be\u03b5\u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03ac \u03b4\u03c5\u03c3\u03ba\u03bf\u03bb\u03b5\u03cd\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03b3\u03b5\u03bd\u03b9\u03ba\u03b5\u03cd\u03c3\u03b5\u03b9. \u0391\u03c0\u03cc \u03c4\u03b7\u03bd \u03ac\u03bb\u03bb\u03b7 \u03c0\u03bb\u03b5\u03c5\u03c1\u03ac, \u03bc\u03b9\u03b1 \u03c5\u03c0\u03b5\u03c1\u03b2\u03bf\u03bb\u03b9\u03ba\u03ae \u03c3\u03c5\u03b3\u03c7\u03ce\u03bd\u03b5\u03c5\u03c3\u03b7 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03ba\u03b1\u03c4\u03b1\u03bb\u03ae\u03be\u03b5\u03b9 \u03c3\u03b5 \u03ad\u03bd\u03b1 \u03ba\u03b1\u03b8\u03bf\u03bb\u03b9\u03ba\u03cc \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03bc\u03b9\u03b1\u03c2 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7\u03c2 \u03c0\u03bf\u03c5 \u03b4\u03ad\u03c7\u03b5\u03c4\u03b1\u03b9 \u03c4\u03b1 \u03c0\u03ac\u03bd\u03c4\u03b1 \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03b5\u03be\u03b7\u03b3\u03b5\u03af \u03c4\u03af\u03c0\u03bf\u03c4\u03b1. \u0397 \u03bc\u03ad\u03b8\u03bf\u03b4\u03bf\u03c2 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b5\u03af \u03bd\u03b1 \u03b9\u03c3\u03bf\u03c1\u03c1\u03bf\u03c0\u03ae\u03c3\u03b5\u03b9 \u03b1\u03bd\u03ac\u03bc\u03b5\u03c3\u03b1 \u03c3\u03b5 \u03b1\u03c5\u03c4\u03ac \u03c4\u03b1 \u03b4\u03cd\u03bf \u03ac\u03ba\u03c1\u03b1.<\/p>\n<div class=\"td-chart td-chart--metrics\">\n<div class=\"td-chart-head\">\n<p class=\"td-chart-title\">\u03a0\u03cc\u03c3\u03bf \u03bc\u03b9\u03ba\u03c1\u03cc \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03c4\u03bf FSM \u03c7\u03c9\u03c1\u03af\u03c2 \u03bd\u03b1 \u03c7\u03ac\u03bd\u03b5\u03b9 \u03c4\u03bf trace<\/p>\n<p class=\"td-chart-subtitle\">\u0391\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03c4\u03b7\u03c2 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7\u03c2 \u03c3\u03b5 \u03b4\u03ce\u03b4\u03b5\u03ba\u03b1 \u03b4\u03b7\u03bc\u03cc\u03c3\u03b9\u03b1 datasets<\/p>\n<\/div>\n<div class=\"td-chart-body\">\n<div class=\"td-metric-grid\">\n<div class=\"td-metric-card\">\n<p class=\"td-metric-value\">12<\/p>\n<p class=\"td-metric-label\">\u03b4\u03b7\u03bc\u03cc\u03c3\u03b9\u03b1 datasets<\/p>\n<p class=\"td-metric-note\">\u039a\u03ac\u03bb\u03c5\u03c8\u03b1\u03bd \u03bf\u03ba\u03c4\u03ce \u03c0\u03b5\u03b4\u03af\u03b1, \u03b1\u03c0\u03cc coding \u03ba\u03b1\u03b9 web navigation \u03ad\u03c9\u03c2 customer service.<\/p>\n<\/div>\n<div class=\"td-metric-card\">\n<p class=\"td-metric-value\">7\u201343<\/p>\n<p class=\"td-metric-label\">\u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03b1\u03bd\u03ac FSM<\/p>\n<p class=\"td-metric-note\">\u03a4\u03bf \u03c4\u03b5\u03bb\u03b9\u03ba\u03cc \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03ad\u03bc\u03b5\u03b9\u03bd\u03b5 \u03bc\u03b9\u03ba\u03c1\u03cc \u03c0\u03b1\u03c1\u03ac \u03c4\u03b7 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03b5\u03c4\u03b9\u03ba\u03ae \u03ba\u03bb\u03af\u03bc\u03b1\u03ba\u03b1 \u03c4\u03c9\u03bd traces.<\/p>\n<\/div>\n<div class=\"td-metric-card\">\n<p class=\"td-metric-value\">\u22650,997<\/p>\n<p class=\"td-metric-label\">held-out replay fitness<\/p>\n<p class=\"td-metric-note\">\u0397 \u03c3\u03c5\u03bc\u03c0\u03b1\u03b3\u03ae\u03c2 \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03ae\u03b3\u03b1\u03b3\u03b5 \u03c3\u03c7\u03b5\u03b4\u03cc\u03bd \u03bf\u03bb\u03cc\u03ba\u03bb\u03b7\u03c1\u03b7 \u03c4\u03b7\u03bd \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03af\u03b1 \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd.<\/p>\n<\/div>\n<div class=\"td-metric-card\">\n<p class=\"td-metric-value\">15\u20133.036\u00d7<\/p>\n<p class=\"td-metric-label\">\u03bb\u03b9\u03b3\u03cc\u03c4\u03b5\u03c1\u03b5\u03c2 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03b1\u03c0\u03cc RPNI<\/p>\n<p class=\"td-metric-note\">\u0397 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03c0\u03b5\u03c1\u03b9\u03bb\u03b1\u03bc\u03b2\u03ac\u03bd\u03b5\u03b9 \u03ba\u03b1\u03b9 \u03c4\u03b1 unlabeled datasets \u03c4\u03b7\u03c2 \u03b1\u03be\u03b9\u03bf\u03bb\u03cc\u03b3\u03b7\u03c3\u03b7\u03c2.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<h2 id=\"provlepsi-epomenis-energeias\">\u0397 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03c4\u03b7\u03c2 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1\u03c2 \u03b3\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03b4\u03bf\u03bc\u03b9\u03ba\u03cc \u03c0\u03c1\u03cc\u03b2\u03bb\u03b7\u03bc\u03b1<\/h2>\n<p>\u0391\u03c0\u03cc \u03ba\u03ac\u03b8\u03b5 \u03c4\u03c1\u03ad\u03c7\u03bf\u03c5\u03c3\u03b1 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7, \u03c4\u03bf FSM \u03b5\u03ba\u03c4\u03b9\u03bc\u03ac \u03c4\u03b7\u03bd \u03c0\u03b9\u03b8\u03b1\u03bd\u03cc\u03c4\u03b7\u03c4\u03b1 \u03c4\u03b7\u03c2 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7\u03c2 \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1\u03c2 \u03b1\u03c0\u03cc \u03c4\u03b9\u03c2 \u03c3\u03c5\u03c7\u03bd\u03cc\u03c4\u03b7\u03c4\u03b5\u03c2 \u03c4\u03c9\u03bd \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03b7\u03bc\u03ad\u03bd\u03c9\u03bd \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03c9\u03bd. \u0397 \u03c0\u03bb\u03b7\u03c1\u03bf\u03c6\u03bf\u03c1\u03af\u03b1 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03bf\u03b8\u03b5\u03af \u03c3\u03b5 \u03ad\u03bd\u03b1\u03bd LLM agent \u03c9\u03c2 \u03c3\u03cd\u03bd\u03c4\u03bf\u03bc\u03bf context: \u03c0\u03b9\u03b8\u03b1\u03bd\u03cc\u03c4\u03b7\u03c4\u03b5\u03c2 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1\u03c2 \u03ba\u03b1\u03b9 \u03bf\u03b9 \u03b4\u03b5\u03ba\u03b1\u03c0\u03ad\u03bd\u03c4\u03b5 \u03b5\u03c0\u03b9\u03ba\u03c1\u03b1\u03c4\u03ad\u03c3\u03c4\u03b5\u03c1\u03b5\u03c2 \u03c3\u03c5\u03bd\u03ad\u03c7\u03b5\u03b9\u03b5\u03c2 \u03c0\u03bf\u03bb\u03bb\u03ce\u03bd \u03b2\u03b7\u03bc\u03ac\u03c4\u03c9\u03bd. \u0391\u03c5\u03c4\u03ae \u03b7 \u00ab\u03b5\u03bb\u03ac\u03c7\u03b9\u03c3\u03c4\u03b7\u00bb \u03bc\u03bf\u03c1\u03c6\u03ae \u03b1\u03c0\u03bf\u03b4\u03b5\u03af\u03c7\u03b8\u03b7\u03ba\u03b5 \u03ba\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b7 \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03c0\u03b1\u03c1\u03bf\u03c5\u03c3\u03af\u03b1\u03c3\u03b7 \u03bf\u03bb\u03cc\u03ba\u03bb\u03b7\u03c1\u03bf\u03c5 \u03c4\u03bf\u03c5 \u03b3\u03c1\u03ac\u03c6\u03bf\u03c5.<\/p>\n<p>\u03a3\u03c4\u03bf tau2-bench retail, \u03c4\u03bf trace prefix \u03c7\u03c9\u03c1\u03af\u03c2 \u03bc\u03bd\u03ae\u03bc\u03b7 \u03c0\u03ad\u03c4\u03c5\u03c7\u03b5 top-1 accuracy 27,6%, \u03c4\u03bf Agent Workflow Memory (AWM) 52,9%, \u03b5\u03bd\u03ce \u03c4\u03bf minimal FSM context \u03ad\u03c6\u03c4\u03b1\u03c3\u03b5 65,1%. \u0397 \u03c0\u03c1\u03bf\u03c3\u03b8\u03ae\u03ba\u03b7 \u03bf\u03bb\u03cc\u03ba\u03bb\u03b7\u03c1\u03b7\u03c2 \u03c4\u03b7\u03c2 \u03b4\u03bf\u03bc\u03ae\u03c2 \u03b4\u03b5\u03bd \u03b2\u03bf\u03ae\u03b8\u03b7\u03c3\u03b5: \u03c4\u03bf ASG-full \u03c3\u03b7\u03bc\u03b5\u03af\u03c9\u03c3\u03b5 52,2% \u03ba\u03b1\u03b9 \u03b7 \u03b1\u03ba\u03cc\u03bc\u03b7 \u03c0\u03bb\u03bf\u03c5\u03c3\u03b9\u03cc\u03c4\u03b5\u03c1\u03b7 \u03b5\u03ba\u03b4\u03bf\u03c7\u03ae ASG++ 49,2%. \u03a4\u03bf \u03b5\u03cd\u03c1\u03b7\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03c1\u03b1\u03ba\u03c4\u03b9\u03ba\u03cc: \u03c0\u03b5\u03c1\u03b9\u03c3\u03c3\u03cc\u03c4\u03b5\u03c1\u03b1 \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03b1 \u03c3\u03c4\u03bf prompt \u03b4\u03b5\u03bd \u03b9\u03c3\u03bf\u03b4\u03c5\u03bd\u03b1\u03bc\u03bf\u03cd\u03bd \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03b1 \u03bc\u03b5 \u03ba\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b7 \u03ba\u03b1\u03b8\u03bf\u03b4\u03ae\u03b3\u03b7\u03c3\u03b7. \u0397 \u03c3\u03c9\u03c3\u03c4\u03ae \u03c3\u03c5\u03bc\u03c0\u03cd\u03ba\u03bd\u03c9\u03c3\u03b7 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03b9\u03bf \u03c7\u03c1\u03ae\u03c3\u03b9\u03bc\u03b7 \u03b1\u03c0\u03cc \u03bc\u03b9\u03b1 \u03c0\u03bb\u03ae\u03c1\u03b7 \u03b1\u03bb\u03bb\u03ac \u03b4\u03cd\u03c3\u03c7\u03c1\u03b7\u03c3\u03c4\u03b7 \u03c0\u03b5\u03c1\u03b9\u03b3\u03c1\u03b1\u03c6\u03ae.<\/p>\n<p>\u03a3\u03b5 \u03bf\u03ba\u03c4\u03ce datasets, \u03c4\u03bf FSM context \u03be\u03b5\u03c0\u03ad\u03c1\u03b1\u03c3\u03b5 \u03c4\u03bf AWM \u03c3\u03c4\u03b7\u03bd top-1 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03bc\u03b5 LLM judge. \u039f\u03b9 \u03b4\u03b9\u03b1\u03c6\u03bf\u03c1\u03ad\u03c2 \u03ba\u03c5\u03bc\u03ac\u03bd\u03b8\u03b7\u03ba\u03b1\u03bd \u03b1\u03c0\u03cc 0,8 \u03c0\u03bf\u03c3\u03bf\u03c3\u03c4\u03b9\u03b1\u03af\u03b5\u03c2 \u03bc\u03bf\u03bd\u03ac\u03b4\u03b5\u03c2 \u03c3\u03c4\u03bf tau2-bench airline \u03ad\u03c9\u03c2 25,3 \u03bc\u03bf\u03bd\u03ac\u03b4\u03b5\u03c2 \u03c3\u03c4\u03bf SWE-smith. \u03a3\u03c4\u03bf WebArena \u03b7 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03ae\u03c4\u03b1\u03bd 81,2% \u03ad\u03bd\u03b1\u03bd\u03c4\u03b9 65,5%, \u03b5\u03bd\u03ce \u03c3\u03c4\u03bf OSWorld 70,7% \u03ad\u03bd\u03b1\u03bd\u03c4\u03b9 55,0%. \u039f\u03b9 \u03c3\u03c5\u03b3\u03b3\u03c1\u03b1\u03c6\u03b5\u03af\u03c2 \u03b1\u03c0\u03bf\u03b4\u03af\u03b4\u03bf\u03c5\u03bd \u03bc\u03ad\u03c1\u03bf\u03c2 \u03c4\u03bf\u03c5 \u03c7\u03ac\u03c3\u03bc\u03b1\u03c4\u03bf\u03c2 \u03c3\u03c4\u03bf \u03cc\u03c4\u03b9 \u03c4\u03bf AWM \u03b5\u03be\u03ac\u03b3\u03b5\u03b9 workflows \u03bc\u03cc\u03bd\u03bf \u03b1\u03c0\u03cc \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b1 traces, \u03ba\u03ac\u03c4\u03b9 \u03c0\u03bf\u03c5 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03af\u03b6\u03b5\u03b9 \u03c4\u03b7\u03bd \u03ba\u03ac\u03bb\u03c5\u03c8\u03ae \u03c4\u03bf\u03c5 \u03cc\u03c4\u03b1\u03bd \u03c4\u03bf \u03c0\u03bf\u03c3\u03bf\u03c3\u03c4\u03cc \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03af\u03b1\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c7\u03b1\u03bc\u03b7\u03bb\u03cc.<\/p>\n<div class=\"td-comparison td-comparison-cards--horizontal\">\n<div class=\"td-chart-head\">\n<p class=\"td-chart-kicker\">tau2-bench retail<\/p>\n<p class=\"td-chart-title\">\u0397 \u03c3\u03c5\u03bc\u03c0\u03cd\u03ba\u03bd\u03c9\u03c3\u03b7 \u03c4\u03bf\u03c5 context \u03ba\u03ad\u03c1\u03b4\u03b9\u03c3\u03b5 \u03c4\u03b7\u03bd \u03c0\u03bb\u03b7\u03c1\u03cc\u03c4\u03b7\u03c4\u03b1<\/p>\n<p class=\"td-chart-intro\">\u03a3\u03c4\u03b7\u03bd top-1 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1\u03c2, \u03c4\u03bf \u03bc\u03b9\u03ba\u03c1\u03cc state context \u03b1\u03c0\u03ad\u03b4\u03c9\u03c3\u03b5 \u03ba\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b1 \u03b1\u03c0\u03cc \u03c4\u03bf success-only workflow memory \u03ba\u03b1\u03b9 \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03c0\u03b1\u03c1\u03bf\u03c5\u03c3\u03af\u03b1\u03c3\u03b7 \u03bf\u03bb\u03cc\u03ba\u03bb\u03b7\u03c1\u03bf\u03c5 \u03c4\u03bf\u03c5 \u03b3\u03c1\u03ac\u03c6\u03bf\u03c5.<\/p>\n<\/div>\n<div class=\"td-comparison-grid\">\n<div class=\"td-platform-card\">\n<p class=\"td-platform-name\">Agent Workflow Memory<\/p>\n<p><strong>52,9%<\/strong> top-1 accuracy \u03bc\u03b5 workflows \u03c0\u03bf\u03c5 \u03b5\u03be\u03ac\u03b3\u03bf\u03bd\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b1 traces.<\/p>\n<\/div>\n<div class=\"td-platform-card\">\n<p class=\"td-platform-name\">Minimal FSM context<\/p>\n<p><strong>65,1%<\/strong> top-1 accuracy \u03bc\u03b5 \u03c0\u03b9\u03b8\u03b1\u03bd\u03cc\u03c4\u03b7\u03c4\u03b5\u03c2 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1\u03c2 \u03ba\u03b1\u03b9 \u03c3\u03cd\u03bd\u03c4\u03bf\u03bc\u03b5\u03c2 \u03b5\u03c0\u03b9\u03ba\u03c1\u03b1\u03c4\u03ad\u03c3\u03c4\u03b5\u03c1\u03b5\u03c2 \u03c3\u03c5\u03bd\u03ad\u03c7\u03b5\u03b9\u03b5\u03c2.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<h2 id=\"provlepsi-apotychias\">\u0391\u03c0\u03cc \u03c4\u03bf \u00ab\u03c4\u03b9 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03b5\u03af\u00bb \u03c3\u03c4\u03bf \u00ab\u03b8\u03b1 \u03b1\u03c0\u03bf\u03c4\u03cd\u03c7\u03b5\u03b9;\u00bb<\/h2>\n<p>\u0393\u03b9\u03b1 \u03c4\u03b7\u03bd \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03af\u03b1\u03c2 \u03ae \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1\u03c2, \u03bf\u03b9 \u03b5\u03c1\u03b5\u03c5\u03bd\u03b7\u03c4\u03ad\u03c2 \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03ae\u03b3\u03b1\u03b3\u03b1\u03bd \u03ba\u03ac\u03b8\u03b5 trace \u03c0\u03ac\u03bd\u03c9 \u03c3\u03c4\u03bf FSM \u03ba\u03b1\u03b9 \u03b5\u03be\u03ae\u03b3\u03b1\u03b3\u03b1\u03bd \u03c7\u03b1\u03c1\u03b1\u03ba\u03c4\u03b7\u03c1\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac \u03b1\u03bd\u03ac \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7: \u03c3\u03c5\u03c7\u03bd\u03cc\u03c4\u03b7\u03c4\u03b1 \u03b5\u03c0\u03af\u03c3\u03ba\u03b5\u03c8\u03b7\u03c2, \u03bc\u03ae\u03ba\u03bf\u03c2 \u03bc\u03b7\u03bd\u03c5\u03bc\u03ac\u03c4\u03c9\u03bd, error rate \u03ba\u03b1\u03b9 \u03c7\u03c1\u03bf\u03bd\u03b9\u03ba\u03ae \u03b8\u03ad\u03c3\u03b7. \u03a0\u03c1\u03cc\u03c3\u03b8\u03b5\u03c3\u03b1\u03bd \u03b5\u03c0\u03af\u03c3\u03b7\u03c2 \u03c0\u03ad\u03bd\u03c4\u03b5 anomaly features \u03c0\u03bf\u03c5 \u03b2\u03b1\u03c3\u03af\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03c4\u03bf cross-entropy, \u03cc\u03c0\u03c9\u03c2 \u03b7 \u03bc\u03ad\u03b3\u03b9\u03c3\u03c4\u03b7 \u03ad\u03ba\u03c0\u03bb\u03b7\u03be\u03b7, \u03b7 \u03bc\u03b5\u03c4\u03b1\u03b2\u03bf\u03bb\u03ae \u03bc\u03b5\u03c4\u03b1\u03be\u03cd \u03c0\u03c1\u03ce\u03c4\u03bf\u03c5 \u03ba\u03b1\u03b9 \u03b4\u03b5\u03cd\u03c4\u03b5\u03c1\u03bf\u03c5 \u03bc\u03b9\u03c3\u03bf\u03cd \u03ba\u03b1\u03b9 \u03c4\u03bf \u03c0\u03bf\u03c3\u03bf\u03c3\u03c4\u03cc \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03c9\u03bd \u03c5\u03c8\u03b7\u03bb\u03ae\u03c2 \u03ad\u03ba\u03c0\u03bb\u03b7\u03be\u03b7\u03c2.<\/p>\n<p>\u03a4\u03bf \u03b1\u03c0\u03bb\u03cc replay fitness \u03b4\u03b5\u03bd \u03ae\u03c4\u03b1\u03bd \u03b5\u03bd\u03b7\u03bc\u03b5\u03c1\u03c9\u03c4\u03b9\u03ba\u03cc \u03b3\u03b9\u03b1 \u03c4\u03b7\u03bd \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1, \u03bc\u03b5 AUROC \u03c0\u03b5\u03c1\u03af\u03c0\u03bf\u03c5 0,50. \u0391\u03bd\u03c4\u03af\u03b8\u03b5\u03c4\u03b1, \u03c4\u03b1 \u03c7\u03b1\u03c1\u03b1\u03ba\u03c4\u03b7\u03c1\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac \u03c0\u03bf\u03c5 \u03b1\u03c0\u03bf\u03c4\u03cd\u03c0\u03c9\u03bd\u03b1\u03bd \u03c0\u03bf\u03cd \u03ba\u03b1\u03b9 \u03c0\u03ce\u03c2 \u03ba\u03b9\u03bd\u03ae\u03b8\u03b7\u03ba\u03b5 \u03c4\u03bf trace \u03ad\u03c6\u03c4\u03b1\u03c3\u03b1\u03bd held-out AUROC \u03ad\u03c9\u03c2 0,941 \u03c3\u03c4\u03bf tau2-bench telecom. \u03a4\u03b1 \u03b1\u03bd\u03c4\u03af\u03c3\u03c4\u03bf\u03b9\u03c7\u03b1 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03c0\u03bf\u03c5 \u03b1\u03bd\u03b1\u03c6\u03ad\u03c1\u03bf\u03bd\u03c4\u03b1\u03b9 \u03b3\u03b9\u03b1 WebArena, AgentNet \u03ba\u03b1\u03b9 ATBench \u03ae\u03c4\u03b1\u03bd 0,903, 0,890 \u03ba\u03b1\u03b9 0,894. \u03a3\u03c4\u03bf SWE-agent, \u03c4\u03b1 \u03b4\u03bf\u03bc\u03b9\u03ba\u03ac \u03c7\u03b1\u03c1\u03b1\u03ba\u03c4\u03b7\u03c1\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac \u03c0\u03ad\u03c4\u03c5\u03c7\u03b1\u03bd AUROC 0,790 \u03ad\u03bd\u03b1\u03bd\u03c4\u03b9 0,659 \u03cc\u03c4\u03b1\u03bd \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03ae\u03b8\u03b7\u03ba\u03b5 \u03bc\u03cc\u03bd\u03bf \u03c4\u03bf \u03bc\u03ae\u03ba\u03bf\u03c2.<\/p>\n<p>\u0397 \u03b1\u03bd\u03ac\u03bb\u03c5\u03c3\u03b7 \u03c0\u03c1\u03bf\u03c3\u03c6\u03ad\u03c1\u03b5\u03b9 \u03ba\u03b1\u03b9 \u03b5\u03c1\u03bc\u03b7\u03bd\u03b5\u03cd\u03c3\u03b9\u03bc\u03b1 \u03c3\u03ae\u03bc\u03b1\u03c4\u03b1. \u03a3\u03c4\u03bf SWE-agent, \u03b7 \u03b5\u03c0\u03af\u03c4\u03b5\u03c5\u03be\u03b7 \u03c4\u03b7\u03c2 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7\u03c2 submit \u03b5\u03bc\u03c6\u03b1\u03bd\u03af\u03c3\u03c4\u03b7\u03ba\u03b5 \u03c3\u03c4\u03bf 94,8% \u03c4\u03c9\u03bd \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03c9\u03bd traces \u03ba\u03b1\u03b9 \u03c3\u03c4\u03bf 55,7% \u03c4\u03c9\u03bd \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03c9\u03bd. \u03a4\u03b1 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b1 traces \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b1\u03bd 9 \u03b1\u03c0\u03cc \u03c4\u03b9\u03c2 25 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c3\u03b5 \u03bc\u03b9\u03b1 \u03b5\u03c3\u03c4\u03b9\u03b1\u03c3\u03bc\u03ad\u03bd\u03b7 \u03b4\u03b9\u03b1\u03b4\u03c1\u03bf\u03bc\u03ae search\u2013edit\u2013submit, \u03b5\u03bd\u03ce \u03bf\u03b9 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b5\u03c2 \u03b1\u03c0\u03bb\u03ce\u03b8\u03b7\u03ba\u03b1\u03bd \u03ba\u03b1\u03b9 \u03c3\u03c4\u03b9\u03c2 25. \u03a3\u03b5 \u03ac\u03bb\u03bb\u03b1 datasets, \u03c3\u03ae\u03bc\u03b1\u03c4\u03b1 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1\u03c2 \u03ae\u03c4\u03b1\u03bd \u03c4\u03b1 \u03b1\u03c5\u03be\u03b7\u03bc\u03ad\u03bd\u03b1 errors \u03c3\u03b5 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b5\u03c2 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2, \u03bf\u03b9 \u03c0\u03bf\u03bb\u03cd \u03bc\u03b5\u03b3\u03ac\u03bb\u03b5\u03c2 \u03b1\u03c0\u03bf\u03ba\u03c1\u03af\u03c3\u03b5\u03b9\u03c2 \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03c9\u03bd \u03ba\u03b1\u03b9 \u03b7 \u03c7\u03b1\u03bf\u03c4\u03b9\u03ba\u03ae \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac \u03c3\u03c4\u03bf \u03b4\u03b5\u03cd\u03c4\u03b5\u03c1\u03bf \u03bc\u03b9\u03c3\u03cc \u03c4\u03b7\u03c2 \u03b5\u03ba\u03c4\u03ad\u03bb\u03b5\u03c3\u03b7\u03c2.<\/p>\n<p>\u0391\u03c5\u03c4\u03cc \u03b5\u03af\u03bd\u03b1\u03b9 \u03c3\u03b7\u03bc\u03b1\u03bd\u03c4\u03b9\u03ba\u03cc \u03b3\u03b9\u03b1 \u03bc\u03b9\u03b1 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03af\u03b1\u03c2: \u03c4\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03b4\u03b5\u03bd \u03bb\u03ad\u03b5\u03b9 \u03bc\u03cc\u03bd\u03bf \u03cc\u03c4\u03b9 \u03ad\u03bd\u03b1 trace \u00ab\u03c6\u03b1\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03ba\u03b1\u03ba\u03cc\u00bb. \u039c\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03b5\u03af\u03be\u03b5\u03b9 \u03b1\u03bd \u03bf agent \u03b4\u03b5\u03bd \u03c6\u03c4\u03ac\u03bd\u03b5\u03b9 \u03c3\u03b5 terminal state, \u03b1\u03bd \u03c3\u03c5\u03c3\u03c3\u03c9\u03c1\u03b5\u03cd\u03b5\u03b9 errors \u03c3\u03b5 edit state \u03ae \u03b1\u03bd \u03b5\u03c0\u03b1\u03bd\u03ad\u03c1\u03c7\u03b5\u03c4\u03b1\u03b9 \u03c5\u03c0\u03b5\u03c1\u03b2\u03bf\u03bb\u03b9\u03ba\u03ac \u03c3\u03c5\u03c7\u03bd\u03ac \u03c3\u03c4\u03bf\u03bd \u03af\u03b4\u03b9\u03bf \u03ba\u03cd\u03ba\u03bb\u03bf. \u0397 \u03c4\u03bf\u03c0\u03b9\u03ba\u03ae \u03b5\u03be\u03ae\u03b3\u03b7\u03c3\u03b7 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03bf\u03bb\u03cd \u03c0\u03b9\u03bf \u03b1\u03be\u03b9\u03bf\u03c0\u03bf\u03b9\u03ae\u03c3\u03b9\u03bc\u03b7 \u03b3\u03b9\u03b1 debugging \u03b1\u03c0\u03cc \u03ad\u03bd\u03b1\u03bd \u03bc\u03bf\u03bd\u03b1\u03b4\u03b9\u03ba\u03cc \u03c3\u03c5\u03bd\u03bf\u03bb\u03b9\u03ba\u03cc \u03b4\u03b5\u03af\u03ba\u03c4\u03b7.<\/p>\n<h2 id=\"runtime-monitoring-proimo-stamatima\">Runtime monitoring \u03ba\u03b1\u03b9 \u03ad\u03b3\u03ba\u03b1\u03b9\u03c1\u03bf \u03c3\u03c4\u03b1\u03bc\u03ac\u03c4\u03b7\u03bc\u03b1<\/h2>\n<p>\u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b4\u03bf\u03ba\u03af\u03bc\u03b1\u03c3\u03b5 \u03ba\u03b1\u03b9 \u03ad\u03bd\u03b1\u03bd online monitor \u03c7\u03c9\u03c1\u03af\u03c2 \u03b5\u03ba\u03c0\u03b1\u03b9\u03b4\u03b5\u03c5\u03bc\u03ad\u03bd\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf. \u039f monitor \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03ac\u03b3\u03b5\u03b9 \u03c4\u03b9\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b5\u03c2 \u03b2\u03ae\u03bc\u03b1 \u03c0\u03c1\u03bf\u03c2 \u03b2\u03ae\u03bc\u03b1 \u03ba\u03b1\u03b9 \u03c7\u03c1\u03b7\u03c3\u03b9\u03bc\u03bf\u03c0\u03bf\u03b9\u03b5\u03af \u03ba\u03b1\u03bd\u03cc\u03bd\u03b5\u03c2 \u03b3\u03b9\u03b1 \u03c4\u03bf \u03c0\u03bf\u03c3\u03bf\u03c3\u03c4\u03cc \u03ba\u03cd\u03ba\u03bb\u03c9\u03bd \u03ba\u03b1\u03b9 \u03c4\u03bf\u03bd \u03b5\u03bb\u03ac\u03c7\u03b9\u03c3\u03c4\u03bf \u03b1\u03c1\u03b9\u03b8\u03bc\u03cc \u03bc\u03bf\u03bd\u03b1\u03b4\u03b9\u03ba\u03ce\u03bd \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03c9\u03bd, \u03bc\u03b5\u03c4\u03ac \u03b1\u03c0\u03cc \u03bc\u03b9\u03b1 \u03c3\u03cd\u03bd\u03c4\u03bf\u03bc\u03b7 \u03c0\u03b5\u03c1\u03af\u03bf\u03b4\u03bf warm-up. \u0395\u03bd\u03b5\u03c1\u03b3\u03bf\u03c0\u03bf\u03b9\u03b5\u03af\u03c4\u03b1\u03b9 \u03b5\u03c0\u03af\u03c3\u03b7\u03c2 \u03cc\u03c4\u03b1\u03bd \u03bf agent \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03c3\u03c4\u03b7\u03bd \u03af\u03b4\u03b9\u03b1 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 \u03b3\u03b9\u03b1 \u03c4\u03bf\u03c5\u03bb\u03ac\u03c7\u03b9\u03c3\u03c4\u03bf\u03bd \u03c0\u03ad\u03bd\u03c4\u03b5 \u03c3\u03c5\u03bd\u03b5\u03c7\u03cc\u03bc\u03b5\u03bd\u03b1 \u03b2\u03ae\u03bc\u03b1\u03c4\u03b1.<\/p>\n<p>\u03a3\u03c4\u03bf SWE-agent, \u03c4\u03bf \u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1 \u03b5\u03bd\u03b5\u03c1\u03b3\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b5 early stopping \u03ba\u03b1\u03c4\u03ac \u03bc\u03ad\u03c3\u03bf \u03cc\u03c1\u03bf \u03c3\u03c4\u03bf 32% \u03c4\u03b7\u03c2 \u03b5\u03ba\u03c4\u03ad\u03bb\u03b5\u03c3\u03b7\u03c2, \u03bc\u03b5 precision 85,9% \u03ba\u03b1\u03b9 recall 95,5%, \u03b5\u03be\u03bf\u03b9\u03ba\u03bf\u03bd\u03bf\u03bc\u03ce\u03bd\u03c4\u03b1\u03c2 \u03c4\u03bf 68% \u03c4\u03bf\u03c5 \u03c5\u03c0\u03bf\u03bb\u03b5\u03b9\u03c0\u03cc\u03bc\u03b5\u03bd\u03bf\u03c5 compute \u03c3\u03c4\u03b9\u03c2 \u03b4\u03b9\u03b1\u03ba\u03bf\u03c0\u03b5\u03af\u03c3\u03b5\u03c2 \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b5\u03b9\u03c2. \u03a3\u03c4\u03bf tau2-bench airline, \u03b7 \u03c0\u03b1\u03c1\u03ad\u03bc\u03b2\u03b1\u03c3\u03b7 \u03ad\u03b3\u03b9\u03bd\u03b5 \u03c3\u03c4\u03bf 56% \u03bc\u03b5 precision 76,0% \u03ba\u03b1\u03b9 recall 79,2%. \u0397 \u03b5\u03c0\u03b5\u03be\u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 \u03b2\u03b1\u03c3\u03b9\u03b6\u03cc\u03c4\u03b1\u03bd \u03bc\u03cc\u03bd\u03bf \u03c3\u03b5 FSM replay \u03ba\u03b1\u03b9 \u03b1\u03bd\u03b1\u03c6\u03ad\u03c1\u03b5\u03c4\u03b1\u03b9 latency 0,006 ms \u03b1\u03bd\u03ac \u03b2\u03ae\u03bc\u03b1.<\/p>\n<p>\u0394\u03b5\u03bd \u03b5\u03af\u03c7\u03b1\u03bd \u03cc\u03bc\u03c9\u03c2 \u03cc\u03bb\u03b1 \u03c4\u03b1 \u03c0\u03b5\u03b4\u03af\u03b1 \u03c4\u03b7\u03bd \u03af\u03b4\u03b9\u03b1 \u03b1\u03c0\u03cc\u03b4\u03bf\u03c3\u03b7. \u03a3\u03c4\u03bf tau2-bench retail \u03ba\u03b1\u03b9 \u03c3\u03c4\u03bf SWE-smith, \u03c4\u03bf rule-based monitoring \u03c5\u03c0\u03b5\u03c1\u03b5\u03bd\u03b5\u03c1\u03b3\u03bf\u03c0\u03bf\u03b9\u03ae\u03b8\u03b7\u03ba\u03b5, \u03b5\u03c0\u03b5\u03b9\u03b4\u03ae \u03c4\u03b1 \u03bc\u03b9\u03ba\u03c1\u03cc\u03c4\u03b5\u03c1\u03b1 \u03ae \u03c0\u03b9\u03bf \u03c7\u03bf\u03bd\u03b4\u03c1\u03bf\u03b5\u03b9\u03b4\u03ae alphabets \u03b4\u03b5\u03bd \u03be\u03b5\u03c7\u03ce\u03c1\u03b9\u03b6\u03b1\u03bd \u03ba\u03b1\u03bb\u03ac \u03c4\u03b1 failure modes. \u0397 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1 \u03b1\u03bd\u03b1\u03b3\u03bd\u03c9\u03c1\u03af\u03b6\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03c3\u03b5 \u03c4\u03ad\u03c4\u03bf\u03b9\u03b1 domains \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af\u03c4\u03b1\u03b9 \u03c0\u03c1\u03bf\u03c3\u03b1\u03c1\u03bc\u03bf\u03b3\u03ae \u03ba\u03b1\u03bd\u03cc\u03bd\u03c9\u03bd \u03b1\u03bd\u03ac dataset.<\/p>\n<p>\u0397 \u03b5\u03c0\u03b9\u03bb\u03bf\u03b3\u03ae operating point \u03b1\u03bb\u03bb\u03ac\u03b6\u03b5\u03b9 \u03b5\u03c0\u03af\u03c3\u03b7\u03c2 \u03c4\u03b7 \u03c7\u03c1\u03ae\u03c3\u03b7. \u039c\u03b5 \u03c5\u03c8\u03b7\u03bb\u03cc threshold \u03ba\u03cd\u03ba\u03bb\u03c9\u03bd \u03c3\u03c4\u03bf SWE-agent, \u03bf monitor \u03c0\u03ad\u03c4\u03c5\u03c7\u03b5 100% precision \u03b1\u03bb\u03bb\u03ac \u03bc\u03cc\u03bb\u03b9\u03c2 11,3% recall. \u0391\u03c5\u03c4\u03cc \u03c4\u03b1\u03b9\u03c1\u03b9\u03ac\u03b6\u03b5\u03b9 \u03c3\u03b5 \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03bf reset \u03cc\u03c4\u03b1\u03bd \u03ad\u03bd\u03b1 false positive \u03b5\u03af\u03bd\u03b1\u03b9 \u03b1\u03ba\u03c1\u03b9\u03b2\u03cc. \u03a3\u03b5 \u03c0\u03b9\u03bf \u03b9\u03c3\u03bf\u03c1\u03c1\u03bf\u03c0\u03b7\u03bc\u03ad\u03bd\u03bf threshold, \u03b7 precision \u03ae\u03c4\u03b1\u03bd 86,4% \u03ba\u03b1\u03b9 \u03b7 recall 85,2%, \u03bc\u03b9\u03b1 \u03c1\u03cd\u03b8\u03bc\u03b9\u03c3\u03b7 \u03ba\u03b1\u03c4\u03b1\u03bb\u03bb\u03b7\u03bb\u03cc\u03c4\u03b5\u03c1\u03b7 \u03b3\u03b9\u03b1 \u03b5\u03b9\u03b4\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b7 \u03b1\u03bd\u03b8\u03c1\u03ce\u03c0\u03b9\u03bd\u03bf\u03c5 operator. \u0386\u03c1\u03b1, \u03b7 \u03c4\u03b5\u03c7\u03bd\u03b9\u03ba\u03ae \u03b4\u03b5\u03bd \u03b4\u03af\u03bd\u03b5\u03b9 \u03bc\u03af\u03b1 \u03ba\u03b1\u03b8\u03bf\u03bb\u03b9\u03ba\u03ae \u00ab\u03c3\u03c9\u03c3\u03c4\u03ae\u00bb \u03c1\u03cd\u03b8\u03bc\u03b9\u03c3\u03b7\u00b7 \u03b4\u03af\u03bd\u03b5\u03b9 \u03ad\u03bd\u03b1 \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03bf \u03b3\u03b9\u03b1 \u03c3\u03c5\u03bd\u03b5\u03b9\u03b4\u03b7\u03c4\u03ae \u03b1\u03bd\u03c4\u03b1\u03bb\u03bb\u03b1\u03b3\u03ae \u03bc\u03b5\u03c4\u03b1\u03be\u03cd \u03ad\u03b3\u03ba\u03b1\u03b9\u03c1\u03b7\u03c2 \u03c0\u03b1\u03c1\u03ad\u03bc\u03b2\u03b1\u03c3\u03b7\u03c2 \u03ba\u03b1\u03b9 \u03c8\u03b5\u03c5\u03b4\u03ce\u03bd \u03c3\u03c5\u03bd\u03b1\u03b3\u03b5\u03c1\u03bc\u03ce\u03bd.<\/p>\n<div class=\"td-note\">\n<p class=\"td-note-title\">\u03a4\u03bf threshold \u03b5\u03af\u03bd\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03ae \u03b1\u03c0\u03cc\u03c6\u03b1\u03c3\u03b7<\/p>\n<p>\u03a3\u03c4\u03bf SWE-agent, \u03b1\u03c5\u03c3\u03c4\u03b7\u03c1\u03cc\u03c4\u03b5\u03c1\u03bf cycle threshold \u03ad\u03b4\u03c9\u03c3\u03b5 100% precision \u03b1\u03bb\u03bb\u03ac \u03bc\u03cc\u03bb\u03b9\u03c2 11,3% recall. \u0391\u03c5\u03c4\u03cc \u03c4\u03b1\u03b9\u03c1\u03b9\u03ac\u03b6\u03b5\u03b9 \u03c3\u03b5 \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03bf stop \u03bc\u03cc\u03bd\u03bf \u03cc\u03c4\u03b1\u03bd \u03c4\u03bf false positive \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03bf\u03bb\u03cd \u03b1\u03ba\u03c1\u03b9\u03b2\u03cc. \u0393\u03b9\u03b1 \u03b5\u03c5\u03c1\u03cd\u03c4\u03b5\u03c1\u03b7 \u03ba\u03ac\u03bb\u03c5\u03c8\u03b7, \u03c4\u03bf \u03af\u03b4\u03b9\u03bf \u03c3\u03ae\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b1\u03c3\u03c6\u03b1\u03bb\u03ad\u03c3\u03c4\u03b5\u03c1\u03bf \u03c9\u03c2 \u03b5\u03b9\u03b4\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b7 \u03c0\u03c1\u03bf\u03c2 \u03ac\u03bd\u03b8\u03c1\u03c9\u03c0\u03bf, \u03cc\u03c0\u03c9\u03c2 \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03ba\u03b1\u03b9 \u03bf \u03c0\u03c1\u03b1\u03ba\u03c4\u03b9\u03ba\u03cc\u03c2 \u03c3\u03c7\u03b5\u03b4\u03b9\u03b1\u03c3\u03bc\u03cc\u03c2 <a href=\"https:\/\/twodots.gr\/human-in-the-loop-ai-agents-anthropini-epopteia\/\">human-in-the-loop \u03b5\u03c0\u03bf\u03c0\u03c4\u03b5\u03af\u03b1\u03c2<\/a>.<\/p>\n<\/div>\n<h2 id=\"harness-vs-montelo\">\u0397 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac \u03b5\u03c0\u03b7\u03c1\u03b5\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c0\u03b5\u03c1\u03b9\u03c3\u03c3\u03cc\u03c4\u03b5\u03c1\u03bf \u03b1\u03c0\u03cc \u03c4\u03bf harness \u03c0\u03b1\u03c1\u03ac \u03b1\u03c0\u03cc \u03c4\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf<\/h2>\n<p>\u0388\u03bd\u03b1 \u03b1\u03c0\u03cc \u03c4\u03b1 \u03c0\u03b9\u03bf \u03b5\u03bd\u03b4\u03b9\u03b1\u03c6\u03ad\u03c1\u03bf\u03bd\u03c4\u03b1 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03ac\u03c3\u03bc\u03b1\u03c4\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03b7 \u03c4\u03bf\u03c0\u03bf\u03bb\u03bf\u03b3\u03af\u03b1 \u03c6\u03ac\u03bd\u03b7\u03ba\u03b5 \u03c3\u03b5 \u03c3\u03b7\u03bc\u03b1\u03bd\u03c4\u03b9\u03ba\u03cc \u03b2\u03b1\u03b8\u03bc\u03cc \u03b1\u03bd\u03b5\u03be\u03ac\u03c1\u03c4\u03b7\u03c4\u03b7 \u03b1\u03c0\u03cc \u03c4\u03bf \u03b5\u03c0\u03b9\u03bc\u03ad\u03c1\u03bf\u03c5\u03c2 LLM. \u03a3\u03c4\u03b1 \u03c4\u03c1\u03af\u03b1 suites \u03c4\u03bf\u03c5 tau2-bench, \u03ad\u03bd\u03b1 \u03b5\u03bd\u03b9\u03b1\u03af\u03bf FSM \u03c0\u03bf\u03c5 \u03ba\u03b1\u03c4\u03b1\u03c3\u03ba\u03b5\u03c5\u03ac\u03c3\u03c4\u03b7\u03ba\u03b5 \u03b1\u03c0\u03cc traces \u03c4\u03b5\u03c3\u03c3\u03ac\u03c1\u03c9\u03bd \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03c9\u03bd \u03b5\u03af\u03c7\u03b5 fitness 1,000 \u03b3\u03b9\u03b1 \u03c4\u03bf \u03ba\u03b1\u03b8\u03ad\u03bd\u03b1. \u03a4\u03b1 per-model FSM \u03bc\u03bf\u03b9\u03c1\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03bd \u03c3\u03c7\u03b5\u03b4\u03cc\u03bd \u03af\u03b4\u03b9\u03bf \u03bb\u03b5\u03be\u03b9\u03bb\u03cc\u03b3\u03b9\u03bf \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03c9\u03bd \u03ba\u03b1\u03b9 \u03ad\u03bd\u03b1\u03bd \u03ba\u03c5\u03c1\u03af\u03b1\u03c1\u03c7\u03bf \u03ba\u03bf\u03c1\u03bc\u03cc \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03c9\u03bd 80% \u03ad\u03c9\u03c2 92%.<\/p>\n<p>\u0397 \u03c0\u03c1\u03bf\u03c3\u03b5\u03ba\u03c4\u03b9\u03ba\u03ae \u03b5\u03c1\u03bc\u03b7\u03bd\u03b5\u03af\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03c4\u03b1 \u03b4\u03b9\u03b1\u03b8\u03ad\u03c3\u03b9\u03bc\u03b1 \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03b1, \u03c4\u03bf schema, \u03c4\u03bf prompt \u03ba\u03b1\u03b9 \u03bf \u03c3\u03c7\u03b5\u03b4\u03b9\u03b1\u03c3\u03bc\u03cc\u03c2 \u03c4\u03bf\u03c5 \u03c0\u03b5\u03c1\u03b9\u03b2\u03ac\u03bb\u03bb\u03bf\u03bd\u03c4\u03bf\u03c2 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03b9\u03b1\u03bc\u03bf\u03c1\u03c6\u03ce\u03bd\u03bf\u03c5\u03bd \u03c4\u03b7\u03bd \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03bf\u03cd\u03bc\u03b5\u03bd\u03b7 \u03c0\u03bf\u03c1\u03b5\u03af\u03b1 \u03c0\u03b5\u03c1\u03b9\u03c3\u03c3\u03cc\u03c4\u03b5\u03c1\u03bf \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03c4\u03b1\u03c5\u03c4\u03cc\u03c4\u03b7\u03c4\u03b1 \u03c4\u03bf\u03c5 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf\u03c5. \u0391\u03c5\u03c4\u03cc \u03b4\u03b5\u03bd \u03b1\u03c0\u03bf\u03b4\u03b5\u03b9\u03ba\u03bd\u03cd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03cc\u03bb\u03b1 \u03c4\u03b1 LLM \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03ad\u03c1\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c4\u03bf \u03af\u03b4\u03b9\u03bf. \u0397 \u03bc\u03b5\u03c4\u03b1\u03c6\u03bf\u03c1\u03ac \u03c4\u03c9\u03bd failure features \u03bc\u03b5\u03c4\u03b1\u03be\u03cd \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03c9\u03bd \u03ae\u03c4\u03b1\u03bd \u03c7\u03b1\u03bc\u03b7\u03bb\u03cc\u03c4\u03b5\u03c1\u03b7 \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03b1\u03c0\u03cc\u03b4\u03bf\u03c3\u03b7 \u03bc\u03ad\u03c3\u03b1 \u03c3\u03c4\u03bf \u03af\u03b4\u03b9\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf: mean cross-AUROC 0,786 \u03ad\u03bd\u03b1\u03bd\u03c4\u03b9 0,877 self \u03c3\u03c4\u03b1 tau2-bench suites.<\/p>\n<p>\u0393\u03b9\u03b1 \u03bc\u03b9\u03b1 \u03b5\u03c0\u03b9\u03c7\u03b5\u03af\u03c1\u03b7\u03c3\u03b7, \u03b7 \u03b4\u03b9\u03ac\u03ba\u03c1\u03b9\u03c3\u03b7 \u03ad\u03c7\u03b5\u03b9 \u03c0\u03c1\u03b1\u03ba\u03c4\u03b9\u03ba\u03ae \u03b1\u03be\u03af\u03b1. \u0391\u03bd \u03b1\u03bb\u03bb\u03ac\u03be\u03b5\u03b9 \u03bf underlying model provider \u03b1\u03bb\u03bb\u03ac \u03bc\u03b5\u03af\u03bd\u03b5\u03b9 \u03c3\u03c4\u03b1\u03b8\u03b5\u03c1\u03cc \u03c4\u03bf workflow, \u03ad\u03bd\u03b1 \u03bc\u03ad\u03c1\u03bf\u03c2 \u03c4\u03b7\u03c2 \u03b4\u03bf\u03bc\u03b9\u03ba\u03ae\u03c2 \u03c0\u03b1\u03c1\u03b1\u03ba\u03bf\u03bb\u03bf\u03cd\u03b8\u03b7\u03c3\u03b7\u03c2 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03c7\u03c1\u03ae\u03c3\u03b9\u03bc\u03bf. \u0391\u03bd \u03b1\u03bb\u03bb\u03ac\u03be\u03bf\u03c5\u03bd \u03c4\u03b1 \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03b1, \u03bf\u03b9 \u03ac\u03b4\u03b5\u03b9\u03b5\u03c2, \u03c4\u03b1 action schemas \u03ae \u03b7 \u03c3\u03b5\u03b9\u03c1\u03ac \u03c4\u03c9\u03bd \u03b2\u03b7\u03bc\u03ac\u03c4\u03c9\u03bd, \u03c4\u03bf FSM \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03b5\u03c0\u03b1\u03bd\u03b5\u03ba\u03c4\u03af\u03bc\u03b7\u03c3\u03b7, \u03b1\u03ba\u03cc\u03bc\u03b7 \u03ba\u03b1\u03b9 \u03b1\u03bd \u03c4\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03af\u03b4\u03b9\u03bf.<\/p>\n<div class=\"td-decision-band\">\n<p class=\"td-decision-kicker\">\u03a4\u03b9 \u03b1\u03bb\u03bb\u03ac\u03b6\u03b5\u03b9 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ac \u03c4\u03bf monitoring<\/p>\n<p class=\"td-decision-title\">\u039a\u03c1\u03b1\u03c4\u03ae\u03c3\u03c4\u03b5 \u03ad\u03ba\u03b4\u03bf\u03c3\u03b7 \u03c4\u03bf\u03c5 harness, \u03cc\u03c7\u03b9 \u03bc\u03cc\u03bd\u03bf \u03c4\u03bf\u03c5 \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf\u03c5<\/p>\n<p>\u038c\u03c4\u03b1\u03bd \u03b1\u03bb\u03bb\u03ac\u03b6\u03bf\u03c5\u03bd tools, permissions, action schemas \u03ae \u03c3\u03b5\u03b9\u03c1\u03ac \u03b2\u03b7\u03bc\u03ac\u03c4\u03c9\u03bd, \u03b1\u03bb\u03bb\u03ac\u03b6\u03b5\u03b9 \u03ba\u03b1\u03b9 \u03b7 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03b9\u03ba\u03ae \u03c4\u03bf\u03c0\u03bf\u03bb\u03bf\u03b3\u03af\u03b1. \u0393\u03b9\u2019 \u03b1\u03c5\u03c4\u03cc \u03c4\u03bf FSM \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03b5\u03c0\u03b1\u03bd\u03b5\u03ba\u03c4\u03b9\u03bc\u03ac\u03c4\u03b1\u03b9 \u03bc\u03b1\u03b6\u03af \u03bc\u03b5 \u03c4\u03bf workflow. \u0397 \u03b4\u03b9\u03ac\u03ba\u03c1\u03b9\u03c3\u03b7 \u03c3\u03c5\u03bc\u03c0\u03bb\u03b7\u03c1\u03ce\u03bd\u03b5\u03b9 \u03c4\u03b7\u03bd <a href=\"https:\/\/twodots.gr\/ai-agents-poios-ftaiei-otan-aftomatopoiisi-apotygchanei\/\">\u03b1\u03c0\u03cc\u03b4\u03bf\u03c3\u03b7 \u03b5\u03c5\u03b8\u03cd\u03bd\u03b7\u03c2 \u03cc\u03c4\u03b1\u03bd \u03b1\u03c0\u03bf\u03c4\u03c5\u03b3\u03c7\u03ac\u03bd\u03b5\u03b9 \u03ad\u03bd\u03b1\u03c2 AI agent<\/a>: \u03b4\u03b5\u03bd \u03b1\u03c1\u03ba\u03b5\u03af \u03bd\u03b1 \u03b3\u03bd\u03c9\u03c1\u03af\u03b6\u03bf\u03c5\u03bc\u03b5 \u03c0\u03bf\u03b9\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03ad\u03c4\u03c1\u03b5\u03be\u03b5, \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03b3\u03bd\u03c9\u03c1\u03af\u03b6\u03bf\u03c5\u03bc\u03b5 \u03c0\u03bf\u03b9\u03bf \u03c3\u03cd\u03c3\u03c4\u03b7\u03bc\u03b1 \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd \u03c4\u03bf \u03c0\u03b5\u03c1\u03b9\u03ad\u03b2\u03b1\u03bb\u03b5.<\/p>\n<\/div>\n<h2 id=\"ecommerce-epicheirisiakoi-agents\">\u03a4\u03b9 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c3\u03b7\u03bc\u03b1\u03af\u03bd\u03b5\u03b9 \u03b3\u03b9\u03b1 e-commerce \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03bf\u03cd\u03c2 agents<\/h2>\n<p>\u03a3\u03b5 \u03ad\u03bd\u03b1 e-commerce \u03c0\u03b5\u03c1\u03b9\u03b2\u03ac\u03bb\u03bb\u03bf\u03bd, \u03ad\u03bd\u03b1\u03c2 agent \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b5\u03bb\u03ad\u03b3\u03c7\u03b5\u03b9 \u03c0\u03b1\u03c1\u03b1\u03b3\u03b3\u03b5\u03bb\u03af\u03b5\u03c2, \u03bd\u03b1 \u03b1\u03bd\u03b1\u03b6\u03b7\u03c4\u03ac \u03c0\u03bf\u03bb\u03b9\u03c4\u03b9\u03ba\u03ad\u03c2 \u03b5\u03c0\u03b9\u03c3\u03c4\u03c1\u03bf\u03c6\u03ce\u03bd, \u03bd\u03b1 \u03c0\u03c1\u03bf\u03c4\u03b5\u03af\u03bd\u03b5\u03b9 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b5\u03c2 \u03ae \u03bd\u03b1 \u03c0\u03c1\u03bf\u03b5\u03c4\u03bf\u03b9\u03bc\u03ac\u03b6\u03b5\u03b9 \u03b1\u03c0\u03b1\u03bd\u03c4\u03ae\u03c3\u03b5\u03b9\u03c2 \u03b3\u03b9\u03b1 \u03c0\u03b5\u03bb\u03ac\u03c4\u03b5\u03c2. \u0397 \u03b5\u03c6\u03b1\u03c1\u03bc\u03bf\u03b3\u03ae \u03c4\u03b7\u03c2 \u03b9\u03b4\u03ad\u03b1\u03c2 \u03c4\u03bf\u03c5 FSM \u03b4\u03b5\u03bd \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af \u03bd\u03b1 \u03b4\u03b9\u03b1\u03b2\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03ba\u03ac\u03b8\u03b5 chain-of-thought \u03bf\u03cd\u03c4\u03b5 \u03bd\u03b1 \u03b1\u03c0\u03bf\u03b8\u03b7\u03ba\u03b5\u03cd\u03b5\u03c4\u03b1\u03b9 \u03cc\u03bb\u03bf \u03c4\u03bf \u03ba\u03b5\u03af\u03bc\u03b5\u03bd\u03bf \u03c9\u03c2 \u03b5\u03bd\u03b9\u03b1\u03af\u03bf \u03b1\u03b4\u03b9\u03b1\u03c6\u03b1\u03bd\u03ad\u03c2 \u03b1\u03bd\u03c4\u03b9\u03ba\u03b5\u03af\u03bc\u03b5\u03bd\u03bf. \u0397 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c7\u03b1\u03c1\u03c4\u03bf\u03b3\u03c1\u03b1\u03c6\u03ae\u03c3\u03b5\u03b9 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03ad\u03c2 \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b5\u03c2, \u03cc\u03c0\u03c9\u03c2 order_lookup, policy_check, inventory_query, customer_message \u03ba\u03b1\u03b9 escalation, \u03ba\u03b1\u03b9 \u03bd\u03b1 \u03b5\u03be\u03b5\u03c4\u03ac\u03c3\u03b5\u03b9 \u03c4\u03b9\u03c2 \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c4\u03bf\u03c5\u03c2.<\/p>\n<p>\u0391\u03bd \u03c4\u03b1 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b1 \u03c0\u03b5\u03c1\u03b9\u03c3\u03c4\u03b1\u03c4\u03b9\u03ba\u03ac \u03c6\u03c4\u03ac\u03bd\u03bf\u03c5\u03bd \u03c3\u03c5\u03bd\u03ae\u03b8\u03c9\u03c2 \u03c3\u03b5 \u03bc\u03b9\u03b1 \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 verification \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc refund \u03ae shipment_change, \u03b7 \u03c3\u03c5\u03c7\u03bd\u03ae \u03c0\u03b1\u03c1\u03ac\u03bb\u03b5\u03b9\u03c8\u03b7 \u03b1\u03c5\u03c4\u03bf\u03cd \u03c4\u03bf\u03c5 \u03b2\u03ae\u03bc\u03b1\u03c4\u03bf\u03c2 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b3\u03af\u03bd\u03b5\u03b9 \u03b5\u03bb\u03b5\u03b3\u03ba\u03c4\u03b9\u03ba\u03cc \u03c3\u03ae\u03bc\u03b1. \u0391\u03bd \u03ad\u03bd\u03b1 workflow marketing \u03c0\u03b5\u03c1\u03bd\u03ac \u03b5\u03c0\u03b1\u03bd\u03b5\u03b9\u03bb\u03b7\u03bc\u03bc\u03ad\u03bd\u03b1 \u03b1\u03c0\u03cc draft \u03c3\u03b5 rewrite \u03c7\u03c9\u03c1\u03af\u03c2 approval \u03ae publish-ready \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7, \u03bf \u03c5\u03c8\u03b7\u03bb\u03cc\u03c2 cycle rate \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03b5\u03af\u03be\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03b7 \u03c0\u03c1\u03bf\u03c4\u03c1\u03bf\u03c0\u03ae, \u03c4\u03b1 \u03ba\u03c1\u03b9\u03c4\u03ae\u03c1\u03b9\u03b1 \u03c0\u03bf\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1\u03c2 \u03ae \u03c4\u03b1 \u03b4\u03b9\u03b1\u03b8\u03ad\u03c3\u03b9\u03bc\u03b1 \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03b1 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03b5\u03c0\u03b1\u03c1\u03ba\u03ae. \u0391\u03c5\u03c4\u03ac \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03b1\u03c1\u03b1\u03b4\u03b5\u03af\u03b3\u03bc\u03b1\u03c4\u03b1 \u03b5\u03c6\u03b1\u03c1\u03bc\u03bf\u03b3\u03ae\u03c2 \u03c4\u03b7\u03c2 \u03bc\u03b5\u03b8\u03bf\u03b4\u03bf\u03bb\u03bf\u03b3\u03af\u03b1\u03c2, \u03cc\u03c7\u03b9 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03c0\u03bf\u03c5 \u03b4\u03bf\u03ba\u03b9\u03bc\u03ac\u03c3\u03c4\u03b7\u03ba\u03b1\u03bd \u03c3\u03c4\u03b7 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1.<\/p>\n<p>\u0397 \u03c3\u03c9\u03c3\u03c4\u03ae \u03c5\u03bb\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b7 \u03b8\u03b1 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03be\u03b5\u03ba\u03b9\u03bd\u03ae\u03c3\u03b5\u03b9 \u03bc\u03b5 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03b7\u03c3\u03b9\u03bc\u03cc\u03c4\u03b7\u03c4\u03b1 \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03bc\u03b5 \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03bf \u03bc\u03c0\u03bb\u03bf\u03ba\u03ac\u03c1\u03b9\u03c3\u03bc\u03b1. \u03a0\u03c1\u03ce\u03c4\u03b1 \u03ba\u03b1\u03c4\u03b1\u03b3\u03c1\u03ac\u03c6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c4\u03b1 activities \u03bc\u03b5 \u03c3\u03c4\u03b1\u03b8\u03b5\u03c1\u03ac \u03bf\u03bd\u03cc\u03bc\u03b1\u03c4\u03b1, \u03b5\u03c0\u03b9\u03c3\u03b7\u03bc\u03b1\u03af\u03bd\u03bf\u03bd\u03c4\u03b1\u03b9 \u03c3\u03b1\u03c6\u03ce\u03c2 \u03c4\u03b1 outcomes \u03ba\u03b1\u03b9 \u03c3\u03c5\u03b3\u03ba\u03c1\u03af\u03bd\u03b5\u03c4\u03b1\u03b9 \u03c4\u03bf FSM \u03bc\u03b5 \u03c4\u03b9\u03c2 \u03b3\u03bd\u03c9\u03c3\u03c4\u03ad\u03c2 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03ad\u03c2 \u03b1\u03c0\u03b1\u03b9\u03c4\u03ae\u03c3\u03b5\u03b9\u03c2. \u0388\u03c0\u03b5\u03b9\u03c4\u03b1 \u03bc\u03c0\u03bf\u03c1\u03bf\u03cd\u03bd \u03bd\u03b1 \u03c0\u03c1\u03bf\u03c3\u03c4\u03b5\u03b8\u03bf\u03cd\u03bd alerts, human review \u03ba\u03b1\u03b9, \u03bc\u03cc\u03bd\u03bf \u03c3\u03b5 \u03ba\u03b1\u03bb\u03ac \u03b5\u03c0\u03b1\u03bb\u03b7\u03b8\u03b5\u03c5\u03bc\u03ad\u03bd\u03b1 operating points, \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03b5\u03c2 \u03c0\u03b1\u03c1\u03b5\u03bc\u03b2\u03ac\u03c3\u03b5\u03b9\u03c2.<\/p>\n<p>\u0399\u03b4\u03b9\u03b1\u03af\u03c4\u03b5\u03c1\u03b7 \u03c3\u03b7\u03bc\u03b1\u03c3\u03af\u03b1 \u03ad\u03c7\u03b5\u03b9 \u03b7 \u03b5\u03c0\u03b9\u03bb\u03bf\u03b3\u03ae activity granularity. \u0391\u03bd \u03cc\u03bb\u03b5\u03c2 \u03bf\u03b9 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b5\u03c2 \u03c3\u03c5\u03bc\u03c0\u03c4\u03c5\u03c7\u03b8\u03bf\u03cd\u03bd \u03c3\u03b5 \u00abtool call\u00bb, \u03c4\u03bf \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03b8\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bc\u03b9\u03ba\u03c1\u03cc \u03b1\u03bb\u03bb\u03ac \u03c0\u03b9\u03b8\u03b1\u03bd\u03ce\u03c2 \u03ac\u03c7\u03c1\u03b7\u03c3\u03c4\u03bf. \u0391\u03bd \u03ba\u03ac\u03b8\u03b5 \u03c0\u03b1\u03c1\u03ac\u03bc\u03b5\u03c4\u03c1\u03bf\u03c2 \u03ba\u03b1\u03b9 \u03ba\u03ac\u03b8\u03b5 \u03ba\u03b5\u03b9\u03bc\u03b5\u03bd\u03b9\u03ba\u03ae \u03c0\u03b1\u03c1\u03b1\u03bb\u03bb\u03b1\u03b3\u03ae \u03b3\u03af\u03bd\u03b5\u03b9 \u03be\u03b5\u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03ae \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1, \u03c4\u03bf alphabet \u03b8\u03b1 \u03b4\u03b9\u03bf\u03b3\u03ba\u03c9\u03b8\u03b5\u03af \u03ba\u03b1\u03b9 \u03b8\u03b1 \u03c7\u03b1\u03b8\u03b5\u03af \u03b7 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03bf\u03cd\u03bc\u03b5\u03bd\u03b7 \u03c3\u03c5\u03bc\u03c0\u03cd\u03ba\u03bd\u03c9\u03c3\u03b7. \u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03b1\u03bd\u03b8\u03b5\u03ba\u03c4\u03b9\u03ba\u03cc\u03c4\u03b7\u03c4\u03b1 \u03c3\u03b5 \u03b1\u03c1\u03ba\u03b5\u03c4\u03ac \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03b1 \u03b5\u03be\u03b1\u03b3\u03c9\u03b3\u03ae\u03c2, \u03cc\u03c7\u03b9 \u03cc\u03bc\u03c9\u03c2 \u03cc\u03c4\u03b9 \u03ba\u03ac\u03b8\u03b5 \u03b1\u03c5\u03b8\u03b1\u03af\u03c1\u03b5\u03c4\u03b7 \u03c4\u03b1\u03be\u03b9\u03bd\u03cc\u03bc\u03b7\u03c3\u03b7 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b5\u03af \u03b5\u03be\u03af\u03c3\u03bf\u03c5 \u03ba\u03b1\u03bb\u03ac.<\/p>\n<div class=\"td-step-list\">\n<p class=\"td-step-list-title\">\u0395\u03c0\u03c4\u03ac \u03b2\u03ae\u03bc\u03b1\u03c4\u03b1 \u03b3\u03b9\u03b1 FSM monitoring \u03c3\u03b5 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03cc agent workflow<\/p>\n<ol>\n<li><span class=\"td-step-kicker\">Activities<\/span><strong>\u039f\u03c1\u03af\u03c3\u03c4\u03b5 \u03c3\u03c4\u03b1\u03b8\u03b5\u03c1\u03ae \u03c4\u03b1\u03be\u03b9\u03bd\u03bf\u03bc\u03af\u03b1 \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd<\/strong>\n<p>\u0394\u03b9\u03b1\u03c7\u03c9\u03c1\u03af\u03c3\u03c4\u03b5 order lookup, policy check, edit, execute, approval \u03ba\u03b1\u03b9 escalation. \u0388\u03bd\u03b1 \u03b3\u03b5\u03bd\u03b9\u03ba\u03cc \u00abtool call\u00bb \u03ba\u03c1\u03cd\u03b2\u03b5\u03b9 \u03c4\u03b9\u03c2 \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c0\u03bf\u03c5 \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03c0\u03b1\u03c1\u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03ae\u03c3\u03b5\u03c4\u03b5.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Outcomes<\/span><strong>\u03a3\u03c5\u03bd\u03b4\u03ad\u03c3\u03c4\u03b5 \u03ba\u03ac\u03b8\u03b5 trace \u03bc\u03b5 \u03b1\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03bf \u03b1\u03c0\u03bf\u03c4\u03ad\u03bb\u03b5\u03c3\u03bc\u03b1<\/strong>\n<p>\u039a\u03b1\u03c4\u03b1\u03b3\u03c1\u03ac\u03c8\u03c4\u03b5 success, failure, \u03b1\u03bd\u03b8\u03c1\u03ce\u03c0\u03b9\u03bd\u03b7 \u03b4\u03b9\u03cc\u03c1\u03b8\u03c9\u03c3\u03b7 \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03c3\u03b9\u03b1\u03ba\u03cc \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2. \u03a7\u03c9\u03c1\u03af\u03c2 \u03c3\u03c9\u03c3\u03c4\u03ac labels, \u03c4\u03b1 failure features \u03b4\u03b5\u03bd \u03ad\u03c7\u03bf\u03c5\u03bd \u03c3\u03c4\u03b1\u03b8\u03b5\u03c1\u03cc \u03c3\u03b7\u03bc\u03b5\u03af\u03bf \u03b1\u03bd\u03b1\u03c6\u03bf\u03c1\u03ac\u03c2.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Versioning<\/span><strong>\u039a\u03c1\u03b1\u03c4\u03ae\u03c3\u03c4\u03b5 \u03ad\u03ba\u03b4\u03bf\u03c3\u03b7 \u03b3\u03b9\u03b1 prompt, tools \u03ba\u03b1\u03b9 permissions<\/strong>\n<p>\u0397 \u03b1\u03bb\u03bb\u03b1\u03b3\u03ae \u03c4\u03bf\u03c5 harness \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03bb\u03b5\u03b9 \u03c4\u03b7\u03bd \u03c4\u03bf\u03c0\u03bf\u03bb\u03bf\u03b3\u03af\u03b1 \u03b1\u03ba\u03cc\u03bc\u03b7 \u03ba\u03b1\u03b9 \u03cc\u03c4\u03b1\u03bd \u03c4\u03bf \u03b2\u03b1\u03c3\u03b9\u03ba\u03cc LLM \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03af\u03b4\u03b9\u03bf.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Baseline<\/span><strong>\u03a7\u03c4\u03af\u03c3\u03c4\u03b5 \u03c4\u03bf \u03c0\u03c1\u03ce\u03c4\u03bf FSM \u03c3\u03b5 observe-only \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03af\u03b1<\/strong>\n<p>\u039c\u03b5\u03c4\u03c1\u03ae\u03c3\u03c4\u03b5 replay fitness, \u03c3\u03c5\u03c7\u03bd\u03cc\u03c4\u03b7\u03c4\u03b5\u03c2 \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03c9\u03bd, cycles \u03ba\u03b1\u03b9 terminal states \u03c0\u03c1\u03b9\u03bd \u03c4\u03bf monitoring \u03b5\u03c0\u03b7\u03c1\u03b5\u03ac\u03c3\u03b5\u03b9 \u03b6\u03c9\u03bd\u03c4\u03b1\u03bd\u03ad\u03c2 \u03b5\u03ba\u03c4\u03b5\u03bb\u03ad\u03c3\u03b5\u03b9\u03c2.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Thresholds<\/span><strong>\u0392\u03b1\u03b8\u03bc\u03bf\u03bd\u03bf\u03bc\u03ae\u03c3\u03c4\u03b5 alerts \u03ba\u03b1\u03b9 stops \u03be\u03b5\u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03ac<\/strong>\n<p>\u0388\u03bd\u03b1 threshold \u03b3\u03b9\u03b1 \u03b1\u03bd\u03b8\u03c1\u03ce\u03c0\u03b9\u03bd\u03b7 \u03b5\u03b9\u03b4\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b7 \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03af\u03bd\u03b5\u03b9 \u03bc\u03b5\u03b3\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b7 \u03ba\u03ac\u03bb\u03c5\u03c8\u03b7. \u03a4\u03bf \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03bf stop \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c5\u03c8\u03b7\u03bb\u03cc\u03c4\u03b5\u03c1\u03b7 precision \u03ba\u03b1\u03b9 \u03c3\u03b1\u03c6\u03ad\u03c2 \u03ba\u03cc\u03c3\u03c4\u03bf\u03c2 \u03bb\u03ac\u03b8\u03bf\u03c5\u03c2.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">Recovery<\/span><strong>\u03a3\u03c7\u03b5\u03b4\u03b9\u03ac\u03c3\u03c4\u03b5 \u03b5\u03c0\u03b1\u03bd\u03b1\u03c6\u03bf\u03c1\u03ac \u03c0\u03ad\u03c1\u03b1 \u03b1\u03c0\u03cc \u03c4\u03b7 \u03b2\u03ac\u03c3\u03b7 \u03b4\u03b5\u03b4\u03bf\u03bc\u03ad\u03bd\u03c9\u03bd<\/strong>\n<p>\u0391\u03bd \u03c4\u03bf trace \u03ad\u03c7\u03b5\u03b9 \u03ae\u03b4\u03b7 \u03b1\u03bb\u03bb\u03ac\u03be\u03b5\u03b9 \u03bc\u03bd\u03ae\u03bc\u03b7, \u03bc\u03ae\u03bd\u03c5\u03bc\u03b1 \u03ae \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1, \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 <a href=\"https:\/\/twodots.gr\/ai-agents-thymountai-lathos-semantic-rollback\/\">semantic rollback \u03b3\u03b9\u03b1 AI agents<\/a> \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03bc\u03cc\u03bd\u03bf \u03b1\u03ba\u03cd\u03c1\u03c9\u03c3\u03b7 \u03bc\u03b9\u03b1\u03c2 \u03b5\u03b3\u03b3\u03c1\u03b1\u03c6\u03ae\u03c2.<\/p>\n<\/li>\n<li><span class=\"td-step-kicker\">\u041f\u0440\u0435\u0433\u043b\u0435\u0434 \u043d\u0430<\/span><strong>\u0395\u03bb\u03ad\u03b3\u03be\u03c4\u03b5 \u03c4\u03b1 failure states \u03bc\u03b5 \u03c4\u03bf\u03c5\u03c2 \u03b9\u03b4\u03b9\u03bf\u03ba\u03c4\u03ae\u03c4\u03b5\u03c2 \u03c4\u03b7\u03c2 \u03b4\u03b9\u03b1\u03b4\u03b9\u03ba\u03b1\u03c3\u03af\u03b1\u03c2<\/strong>\n<p>\u03a3\u03c5\u03bd\u03b4\u03ad\u03c3\u03c4\u03b5 \u03ba\u03ac\u03b8\u03b5 \u03b1\u03c3\u03c5\u03bd\u03ae\u03b8\u03b9\u03c3\u03c4\u03b7 \u03bc\u03b5\u03c4\u03ac\u03b2\u03b1\u03c3\u03b7 \u03bc\u03b5 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03cc \u03ba\u03af\u03bd\u03b4\u03c5\u03bd\u03bf, \u03cc\u03c0\u03c9\u03c2 \u03bb\u03b1\u03bd\u03b8\u03b1\u03c3\u03bc\u03ad\u03bd\u03b7 \u03b5\u03c0\u03b9\u03c3\u03c4\u03c1\u03bf\u03c6\u03ae, \u03bc\u03b7 \u03b5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7 \u03b4\u03b7\u03bc\u03bf\u03c3\u03af\u03b5\u03c5\u03c3\u03b7 \u03ae \u03b5\u03c0\u03b1\u03bd\u03b1\u03bb\u03b1\u03bc\u03b2\u03b1\u03bd\u03cc\u03bc\u03b5\u03bd\u03b7 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b7 \u03b1\u03bb\u03bb\u03b1\u03b3\u03ae \u03ba\u03ce\u03b4\u03b9\u03ba\u03b1. \u0393\u03b9\u03b1 coding agents, \u03b7 <a href=\"https:\/\/twodots.gr\/step-ktoder-ai-lathos-synartisi\/\">\u03b1\u03bd\u03b1\u03c4\u03c1\u03bf\u03c6\u03bf\u03b4\u03cc\u03c4\u03b7\u03c3\u03b7 \u03c3\u03b5 \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03bf \u03c3\u03c5\u03bd\u03ac\u03c1\u03c4\u03b7\u03c3\u03b7\u03c2<\/a> \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03c0\u03cc\u03c3\u03bf \u03c3\u03b7\u03bc\u03b1\u03bd\u03c4\u03b9\u03ba\u03cc \u03b5\u03af\u03bd\u03b1\u03b9 \u03c4\u03bf \u03c3\u03ae\u03bc\u03b1 \u03bd\u03b1 \u03c6\u03c4\u03ac\u03bd\u03b5\u03b9 \u03c3\u03c4\u03bf \u03c3\u03c9\u03c3\u03c4\u03cc \u03c3\u03b7\u03bc\u03b5\u03af\u03bf \u03c4\u03b7\u03c2 \u03c1\u03bf\u03ae\u03c2.<\/p>\n<\/li>\n<\/ol>\n<\/div>\n<h2 id=\"periorismoi-fsm\">\u039f\u03b9 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03bf\u03af \u03c0\u03bf\u03c5 \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03bc\u03b5\u03af\u03bd\u03bf\u03c5\u03bd \u03c3\u03c4\u03bf \u03c0\u03c1\u03bf\u03c3\u03ba\u03ae\u03bd\u03b9\u03bf<\/h2>\n<p>\u03a4\u03bf FSM \u03b1\u03bd\u03b1\u03ba\u03c4\u03ac \u03c4\u03bf directly-follows closure \u03c4\u03c9\u03bd traces \u03c0\u03bf\u03c5 \u03b5\u03af\u03b4\u03b5\u00b7 \u03b4\u03b5\u03bd \u03b1\u03bd\u03b1\u03ba\u03b1\u03bb\u03cd\u03c0\u03c4\u03b5\u03b9 \u03c4\u03b7\u03bd \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ae \u00ab\u03b3\u03bb\u03ce\u03c3\u03c3\u03b1\u00bb \u03c0\u03bf\u03c5 \u03c0\u03b1\u03c1\u03ac\u03b3\u03b5\u03b9 \u03bf agent. \u0397 \u03b4\u03b9\u03ac\u03ba\u03c1\u03b9\u03c3\u03b7 \u03b5\u03af\u03bd\u03b1\u03b9 \u03ba\u03c1\u03af\u03c3\u03b9\u03bc\u03b7. \u0395\u03c0\u03b5\u03b9\u03b4\u03ae \u03b7 \u03bc\u03ad\u03b8\u03bf\u03b4\u03bf\u03c2 \u03c3\u03c5\u03b3\u03c7\u03c9\u03bd\u03b5\u03cd\u03b5\u03b9 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03bc\u03b5 \u03b2\u03ac\u03c3\u03b7 \u03c4\u03b7\u03bd \u03c4\u03b5\u03bb\u03b5\u03c5\u03c4\u03b1\u03af\u03b1 \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1, \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03b4\u03b5\u03c7\u03c4\u03b5\u03af adversarial traces \u03c0\u03bf\u03c5 \u03b4\u03b9\u03b1\u03c4\u03b7\u03c1\u03bf\u03cd\u03bd \u03c4\u03b1 \u03af\u03b4\u03b9\u03b1 \u03c3\u03c4\u03b1\u03c4\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac \u03b6\u03b5\u03c5\u03b3\u03ce\u03bd \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd. \u0394\u03b5\u03bd \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03b5\u03af \u03b1\u03c0\u03cc \u03bc\u03cc\u03bd\u03bf \u03c4\u03bf\u03c5 \u03b1\u03c0\u03cc\u03b4\u03b5\u03b9\u03be\u03b7 \u03b1\u03c3\u03c6\u03ac\u03bb\u03b5\u03b9\u03b1\u03c2 \u03bf\u03cd\u03c4\u03b5 \u03c0\u03bb\u03ae\u03c1\u03b5\u03c2 policy engine.<\/p>\n<p>\u0397 \u03b5\u03be\u03b1\u03b3\u03c9\u03b3\u03ae \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03bf\u03c4\u03ae\u03c4\u03c9\u03bd \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af \u03b5\u03c0\u03af\u03c3\u03b7\u03c2 \u03bb\u03af\u03b3\u03b7 \u03b3\u03bd\u03ce\u03c3\u03b7 \u03c4\u03bf\u03c5 domain. \u039f\u03b9 \u03ba\u03b1\u03bd\u03cc\u03bd\u03b5\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bd\u03c4\u03b5\u03c4\u03b5\u03c1\u03bc\u03b9\u03bd\u03b9\u03c3\u03c4\u03b9\u03ba\u03bf\u03af \u03ba\u03b1\u03b9 \u03b1\u03c0\u03bb\u03bf\u03af, \u03b1\u03bb\u03bb\u03ac \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03bb\u03ae\u03c1\u03c9\u03c2 \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03bf\u03b9. \u039c\u03b5\u03b3\u03ac\u03bb\u03bf\u03b9 \u03c7\u03ce\u03c1\u03bf\u03b9 \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd \u03ae workflows \u03bc\u03b5 \u03b1\u03c3\u03b8\u03b5\u03bd\u03ae \u03c3\u03b5\u03b9\u03c1\u03b9\u03b1\u03ba\u03ae \u03ba\u03b1\u03bd\u03bf\u03bd\u03b9\u03ba\u03cc\u03c4\u03b7\u03c4\u03b1 \u03bc\u03c0\u03bf\u03c1\u03bf\u03cd\u03bd \u03bd\u03b1 \u03b4\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03ae\u03c3\u03bf\u03c5\u03bd \u03bc\u03b5\u03b3\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b1 FSM, \u03bd\u03b1 \u03bc\u03b5\u03b9\u03ce\u03c3\u03bf\u03c5\u03bd \u03c4\u03b9\u03c2 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03ae\u03c3\u03b5\u03b9\u03c2 \u03b1\u03bd\u03ac \u03ba\u03b1\u03c4\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 \u03ba\u03b1\u03b9 \u03bd\u03b1 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03af\u03c3\u03bf\u03c5\u03bd \u03c4\u03b1 \u03bf\u03c6\u03ad\u03bb\u03b7 \u03c3\u03c4\u03b7\u03bd \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7.<\/p>\n<p>\u0397 cross-model \u03bc\u03b5\u03c4\u03b1\u03c6\u03bf\u03c1\u03ac \u03b5\u03be\u03b5\u03c4\u03ac\u03c3\u03c4\u03b7\u03ba\u03b5 \u03c3\u03b5 \u03c4\u03c1\u03af\u03b1 tau2-bench suites \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03c3\u03b5 \u03ba\u03ac\u03b8\u03b5 \u03b1\u03c1\u03c7\u03b9\u03c4\u03b5\u03ba\u03c4\u03bf\u03bd\u03b9\u03ba\u03ae \u03ae \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03b7\u03bc\u03b1\u03c4\u03b9\u03ba\u03cc \u03c0\u03b5\u03b4\u03af\u03bf. \u0397 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 workflow memory \u03bc\u03b5 LLM context \u03ad\u03b3\u03b9\u03bd\u03b5 \u03b1\u03c0\u03ad\u03bd\u03b1\u03bd\u03c4\u03b9 \u03c3\u03c4\u03bf AWM, \u03b5\u03bd\u03ce \u03ac\u03bb\u03bb\u03b5\u03c2 \u03bd\u03b5\u03cc\u03c4\u03b5\u03c1\u03b5\u03c2 \u03bc\u03ad\u03b8\u03bf\u03b4\u03bf\u03b9 \u03c0\u03bf\u03c5 \u03b1\u03be\u03b9\u03bf\u03c0\u03bf\u03b9\u03bf\u03cd\u03bd \u03c4\u03cc\u03c3\u03bf \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03af\u03b5\u03c2 \u03cc\u03c3\u03bf \u03ba\u03b1\u03b9 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b5\u03c2 \u03ad\u03bc\u03b5\u03b9\u03bd\u03b1\u03bd \u03b3\u03b9\u03b1 \u03bc\u03b5\u03bb\u03bb\u03bf\u03bd\u03c4\u03b9\u03ba\u03ae \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1. \u0395\u03c0\u03b9\u03c0\u03bb\u03ad\u03bf\u03bd, \u03bf\u03c1\u03b9\u03c3\u03bc\u03ad\u03bd\u03b1 \u03b5\u03bd\u03c4\u03c5\u03c0\u03c9\u03c3\u03b9\u03b1\u03ba\u03ac \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03b5\u03be\u03b1\u03c1\u03c4\u03ce\u03bd\u03c4\u03b1\u03b9 \u03b1\u03c0\u03cc \u03c4\u03bf dataset: \u03bf runtime monitor \u03ae\u03c4\u03b1\u03bd \u03b4\u03c5\u03bd\u03b1\u03c4\u03cc\u03c2 \u03c3\u03c4\u03bf SWE-agent \u03b1\u03bb\u03bb\u03ac \u03c3\u03b1\u03c6\u03ce\u03c2 \u03b1\u03c3\u03b8\u03b5\u03bd\u03ad\u03c3\u03c4\u03b5\u03c1\u03bf\u03c2 \u03c3\u03b5 \u03bc\u03b9\u03ba\u03c1\u03cc\u03c4\u03b5\u03c1\u03b1 alphabets.<\/p>\n<p>\u03a4\u03ad\u03bb\u03bf\u03c2, \u03bf\u03b9 \u03b1\u03c1\u03b9\u03b8\u03bc\u03bf\u03af \u03b5\u03af\u03bd\u03b1\u03b9 \u03b5\u03c1\u03b5\u03c5\u03bd\u03b7\u03c4\u03b9\u03ba\u03ac \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03c0\u03ac\u03bd\u03c9 \u03c3\u03b5 \u03b4\u03b7\u03bc\u03cc\u03c3\u03b9\u03b1 benchmarks \u03ba\u03b1\u03b9 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b1 splits, \u03cc\u03c7\u03b9 \u03b5\u03b3\u03b3\u03cd\u03b7\u03c3\u03b7 \u03c0\u03b1\u03c1\u03b1\u03b3\u03c9\u03b3\u03b9\u03ba\u03ae\u03c2 \u03b1\u03c0\u03cc\u03b4\u03bf\u03c3\u03b7\u03c2. \u039a\u03ac\u03b8\u03b5 \u03bf\u03c1\u03b3\u03b1\u03bd\u03b9\u03c3\u03bc\u03cc\u03c2 \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03b4\u03b9\u03ba\u03ae \u03c4\u03bf\u03c5 validation \u03b4\u03b9\u03b1\u03b4\u03b9\u03ba\u03b1\u03c3\u03af\u03b1, \u03c0\u03b1\u03c1\u03b1\u03ba\u03bf\u03bb\u03bf\u03cd\u03b8\u03b7\u03c3\u03b7 drift, logging \u03bc\u03b5 \u03c3\u03b5\u03b2\u03b1\u03c3\u03bc\u03cc \u03c3\u03c4\u03b7\u03bd \u03b9\u03b4\u03b9\u03c9\u03c4\u03b9\u03ba\u03cc\u03c4\u03b7\u03c4\u03b1 \u03ba\u03b1\u03b9 \u03c3\u03b1\u03c6\u03ae \u03cc\u03c1\u03b9\u03b1 \u03b3\u03b9\u03b1 \u03c4\u03bf \u03c0\u03cc\u03c4\u03b5 \u03ad\u03bd\u03b1\u03c2 agent \u03b5\u03c0\u03b9\u03c4\u03c1\u03ad\u03c0\u03b5\u03c4\u03b1\u03b9 \u03bd\u03b1 \u03c3\u03c5\u03bd\u03b5\u03c7\u03af\u03c3\u03b5\u03b9, \u03bd\u03b1 \u03b6\u03b7\u03c4\u03ae\u03c3\u03b5\u03b9 \u03ad\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03ae \u03bd\u03b1 \u03c3\u03c4\u03b1\u03bc\u03b1\u03c4\u03ae\u03c3\u03b5\u03b9.<\/p>\n<h2 id=\"diakyvernisi-agentic-ai\">\u0388\u03bd\u03b1 \u03c0\u03b9\u03bf \u03c0\u03c1\u03b1\u03ba\u03c4\u03b9\u03ba\u03cc \u03bc\u03bf\u03bd\u03c4\u03ad\u03bb\u03bf \u03b4\u03b9\u03b1\u03ba\u03c5\u03b2\u03ad\u03c1\u03bd\u03b7\u03c3\u03b7\u03c2 agentic AI<\/h2>\n<p>\u0397 \u03bc\u03b5\u03b3\u03b1\u03bb\u03cd\u03c4\u03b5\u03c1\u03b7 \u03b1\u03be\u03af\u03b1 \u03c4\u03b7\u03c2 \u03b5\u03c1\u03b3\u03b1\u03c3\u03af\u03b1\u03c2 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03ad\u03bd\u03b1 \u03ba\u03bb\u03b1\u03c3\u03b9\u03ba\u03cc \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03bf \u03c4\u03c9\u03bd computer science \u03bb\u03cd\u03bd\u03b5\u03b9 \u03be\u03b1\u03c6\u03bd\u03b9\u03ba\u03ac \u03cc\u03bb\u03b1 \u03c4\u03b1 \u03c0\u03c1\u03bf\u03b2\u03bb\u03ae\u03bc\u03b1\u03c4\u03b1 \u03c4\u03c9\u03bd agents. \u0395\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03bc\u03b5\u03c4\u03b1\u03c6\u03ad\u03c1\u03b5\u03b9 \u03c4\u03b7 \u03c3\u03c5\u03b6\u03ae\u03c4\u03b7\u03c3\u03b7 \u03b1\u03c0\u03cc \u03c4\u03b7\u03bd \u03b1\u03c3\u03b1\u03c6\u03ae \u00ab\u03bd\u03bf\u03b7\u03bc\u03bf\u03c3\u03cd\u03bd\u03b7\u00bb \u03c3\u03c4\u03b7 \u03bc\u03b5\u03c4\u03c1\u03ae\u03c3\u03b9\u03bc\u03b7 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac. \u0388\u03bd\u03b1\u03c2 agent \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c0\u03b1\u03c1\u03b1\u03bc\u03ad\u03bd\u03b5\u03b9 \u03c0\u03b9\u03b8\u03b1\u03bd\u03bf\u03c4\u03b9\u03ba\u03cc\u03c2 \u03c3\u03c4\u03bf \u03ba\u03b5\u03af\u03bc\u03b5\u03bd\u03bf \u03c0\u03bf\u03c5 \u03c0\u03b1\u03c1\u03ac\u03b3\u03b5\u03b9, \u03b5\u03bd\u03ce \u03b7 \u03b1\u03bb\u03bb\u03b7\u03bb\u03bf\u03c5\u03c7\u03af\u03b1 \u03c4\u03c9\u03bd \u03b5\u03c1\u03b3\u03b1\u03bb\u03b5\u03af\u03c9\u03bd \u03c4\u03bf\u03c5 \u03b5\u03bc\u03c6\u03b1\u03bd\u03af\u03b6\u03b5\u03b9 \u03c3\u03c4\u03b1\u03b8\u03b5\u03c1\u03ac, \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03ae\u03c3\u03b9\u03bc\u03b1 \u03bc\u03bf\u03c4\u03af\u03b2\u03b1.<\/p>\n<p>\u0391\u03c5\u03c4\u03cc \u03b5\u03c0\u03b9\u03c4\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bc\u03b9\u03b1 \u03c0\u03bf\u03bb\u03c5\u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03b7 \u03ac\u03bc\u03c5\u03bd\u03b1. \u039f\u03b9 \u03c0\u03bf\u03bb\u03b9\u03c4\u03b9\u03ba\u03ad\u03c2 \u03ba\u03b1\u03b9 \u03c4\u03b1 permissions \u03bf\u03c1\u03af\u03b6\u03bf\u03c5\u03bd \u03c4\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c1\u03ad\u03c0\u03b5\u03c4\u03b1\u03b9. \u03a4\u03bf FSM \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03c4\u03b9 \u03c3\u03c5\u03bc\u03b2\u03b1\u03af\u03bd\u03b5\u03b9 \u03c3\u03c4\u03b7\u03bd \u03c0\u03c1\u03ac\u03be\u03b7. \u03a4\u03b1 per-state \u03c7\u03b1\u03c1\u03b1\u03ba\u03c4\u03b7\u03c1\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac \u03b5\u03bd\u03c4\u03bf\u03c0\u03af\u03b6\u03bf\u03c5\u03bd \u03c0\u03bf\u03cd \u03b1\u03c5\u03be\u03ac\u03bd\u03b5\u03c4\u03b1\u03b9 \u03bf \u03ba\u03af\u03bd\u03b4\u03c5\u03bd\u03bf\u03c2. \u0388\u03bd\u03b1\u03c2 human operator \u03ae \u03ad\u03bd\u03b1 \u03be\u03b5\u03c7\u03c9\u03c1\u03b9\u03c3\u03c4\u03cc control layer \u03b1\u03c0\u03bf\u03c6\u03b1\u03c3\u03af\u03b6\u03b5\u03b9 \u03c4\u03b7\u03bd \u03c0\u03b1\u03c1\u03ad\u03bc\u03b2\u03b1\u03c3\u03b7. \u039a\u03b1\u03bd\u03ad\u03bd\u03b1 \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03bf \u03b4\u03b5\u03bd \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03b8\u03b9\u03c3\u03c4\u03ac \u03c4\u03b1 \u03c5\u03c0\u03cc\u03bb\u03bf\u03b9\u03c0\u03b1.<\/p>\n<p>\u0393\u03b9\u03b1 \u03c4\u03b9\u03c2 \u03b5\u03c0\u03b9\u03c7\u03b5\u03b9\u03c1\u03ae\u03c3\u03b5\u03b9\u03c2 \u03c0\u03bf\u03c5 \u03b5\u03be\u03b5\u03c4\u03ac\u03b6\u03bf\u03c5\u03bd agents \u03c3\u03b5 customer service, content operations, software workflows \u03ae \u03b5\u03c3\u03c9\u03c4\u03b5\u03c1\u03b9\u03ba\u03ad\u03c2 \u03b4\u03b9\u03b1\u03b4\u03b9\u03ba\u03b1\u03c3\u03af\u03b5\u03c2, \u03c4\u03bf \u03bc\u03ae\u03bd\u03c5\u03bc\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03bf: \u03b7 observability \u03c0\u03c1\u03ad\u03c0\u03b5\u03b9 \u03bd\u03b1 \u03c3\u03c7\u03b5\u03b4\u03b9\u03b1\u03c3\u03c4\u03b5\u03af \u03bc\u03b1\u03b6\u03af \u03bc\u03b5 \u03c4\u03bf\u03bd agent, \u03cc\u03c7\u03b9 \u03bc\u03b5\u03c4\u03ac \u03c4\u03b7\u03bd \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1. \u03a3\u03c4\u03b1\u03b8\u03b5\u03c1\u03ac activity labels, outcome data, versioning \u03c4\u03bf\u03c5 harness \u03ba\u03b1\u03b9 \u03b4\u03b9\u03b1\u03c7\u03c9\u03c1\u03b9\u03c3\u03bc\u03cc\u03c2 \u03bc\u03b5\u03c4\u03b1\u03be\u03cd alert \u03ba\u03b1\u03b9 automatic stop \u03b4\u03b7\u03bc\u03b9\u03bf\u03c5\u03c1\u03b3\u03bf\u03cd\u03bd \u03c4\u03b7 \u03b2\u03ac\u03c3\u03b7 \u03b3\u03b9\u03b1 \u03b1\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03b7 \u03b1\u03be\u03b9\u03bf\u03c0\u03bf\u03af\u03b7\u03c3\u03b7.<\/p>\n<p>\u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03cc\u03c4\u03b9 \u03b1\u03ba\u03cc\u03bc\u03b7 \u03ba\u03b1\u03b9 \u03c0\u03bf\u03bb\u03cd\u03c0\u03bb\u03bf\u03ba\u03b1 agent traces \u03bc\u03c0\u03bf\u03c1\u03bf\u03cd\u03bd \u03bd\u03b1 \u03b1\u03c0\u03bf\u03ba\u03c4\u03ae\u03c3\u03bf\u03c5\u03bd \u03ad\u03bd\u03b1\u03bd \u03bc\u03b9\u03ba\u03c1\u03cc \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b9\u03ba\u03cc \u03c7\u03ac\u03c1\u03c4\u03b7. \u039f \u03c7\u03ac\u03c1\u03c4\u03b7\u03c2 \u03b4\u03b5\u03bd \u03b5\u03af\u03bd\u03b1\u03b9 \u03b7 \u03b5\u03c0\u03b9\u03ba\u03c1\u03ac\u03c4\u03b5\u03b9\u03b1 \u03ba\u03b1\u03b9 \u03b4\u03b5\u03bd \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03b8\u03b9\u03c3\u03c4\u03ac \u03c4\u03b7\u03bd \u03b1\u03c3\u03c6\u03ac\u03bb\u03b5\u03b9\u03b1, \u03b1\u03bb\u03bb\u03ac \u03bc\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03ba\u03ac\u03bd\u03b5\u03b9 \u03c4\u03b9\u03c2 \u03b1\u03c0\u03bf\u03ba\u03bb\u03af\u03c3\u03b5\u03b9\u03c2, \u03c4\u03bf\u03c5\u03c2 \u03b2\u03c1\u03cc\u03c7\u03bf\u03c5\u03c2 \u03ba\u03b1\u03b9 \u03c4\u03b1 \u03c3\u03b7\u03bc\u03b5\u03af\u03b1 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1\u03c2 \u03bf\u03c1\u03b1\u03c4\u03ac \u03b1\u03c1\u03ba\u03b5\u03c4\u03ac \u03bd\u03c9\u03c1\u03af\u03c2 \u03ce\u03c3\u03c4\u03b5 \u03b7 \u03bf\u03bc\u03ac\u03b4\u03b1 \u03bd\u03b1 \u03b4\u03c1\u03ac\u03c3\u03b5\u03b9.<\/p>\n<section class=\"td-service-cta\">\n<div class=\"td-service-cta-content\">\n<p class=\"td-service-cta-eyebrow\">Observability \u03b3\u03b9\u03b1 agentic workflows<\/p>\n<p class=\"td-service-cta-title\">\u03a7\u03b1\u03c1\u03c4\u03bf\u03b3\u03c1\u03b1\u03c6\u03ae\u03c3\u03c4\u03b5 \u03c4\u03b7 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac \u03c4\u03bf\u03c5 agent \u03c0\u03c1\u03b9\u03bd \u03b1\u03c5\u03c4\u03bf\u03bc\u03b1\u03c4\u03bf\u03c0\u03bf\u03b9\u03ae\u03c3\u03b5\u03c4\u03b5 \u03c4\u03b7\u03bd \u03c0\u03b1\u03c1\u03ad\u03bc\u03b2\u03b1\u03c3\u03b7<\/p>\n<p>\u0397 TWO DOTS \u03c3\u03c7\u03b5\u03b4\u03b9\u03ac\u03b6\u03b5\u03b9 AI \u03b1\u03c5\u03c4\u03bf\u03bc\u03b1\u03c4\u03b9\u03c3\u03bc\u03bf\u03cd\u03c2 \u03bc\u03b5 \u03c3\u03c4\u03b1\u03b8\u03b5\u03c1\u03ac activity labels, trace monitoring, human review, \u03b1\u03c3\u03c6\u03b1\u03bb\u03ae thresholds \u03ba\u03b1\u03b9 \u03c3\u03c7\u03ad\u03b4\u03b9\u03bf \u03b1\u03bd\u03ac\u03ba\u03b1\u03bc\u03c8\u03b7\u03c2 \u03b3\u03b9\u03b1 e-commerce, support \u03ba\u03b1\u03b9 \u03b5\u03c3\u03c9\u03c4\u03b5\u03c1\u03b9\u03ba\u03ad\u03c2 \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03af\u03b5\u03c2.<\/p>\n<div class=\"td-service-cta-actions\"><a class=\"td-service-cta-button\" href=\"https:\/\/twodots.gr\/aftomatismoi-epicheiriseon-ai\/\">\u0412\u0438\u0436\u0442\u0435 \u0430\u0432\u0442\u043e\u043c\u0430\u0442\u0438\u0437\u0430\u0446\u0438\u044f\u0442\u0430 \u043d\u0430 \u0431\u0438\u0437\u043d\u0435\u0441 \u043f\u0440\u043e\u0446\u0435\u0441\u0438\u0442\u0435 \u0438 \u0438\u0437\u043a\u0443\u0441\u0442\u0432\u0435\u043d\u0438\u044f \u0438\u043d\u0442\u0435\u043b\u0435\u043a\u0442<\/a><\/div>\n<\/div>\n<\/section>\n<section id=\"sychnes-erotiseis\" class=\"td-faq-section\">\n<div class=\"td-faq\">\n<p class=\"td-faq-heading\">\u0427\u0435\u0441\u0442\u043e \u0437\u0430\u0434\u0430\u0432\u0430\u043d\u0438 \u0432\u044a\u043f\u0440\u043e\u0441\u0438<\/p>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a4\u03b9 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bc\u03b9\u03b1 finite-state machine \u03b3\u03b9\u03b1 AI agents;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u0395\u03af\u03bd\u03b1\u03b9 \u03bc\u03b9\u03b1 \u03c3\u03c5\u03bc\u03c0\u03b1\u03b3\u03ae\u03c2 \u03b1\u03bd\u03b1\u03c0\u03b1\u03c1\u03ac\u03c3\u03c4\u03b1\u03c3\u03b7 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03c9\u03bd \u03ba\u03b1\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c1\u03b5\u03c0\u03c4\u03ce\u03bd \u03bc\u03b5\u03c4\u03b1\u03b2\u03ac\u03c3\u03b5\u03c9\u03bd \u03c0\u03bf\u03c5 \u03c0\u03c1\u03bf\u03ba\u03cd\u03c0\u03c4\u03b5\u03b9 \u03b1\u03c0\u03cc \u03c4\u03b9\u03c2 \u03b1\u03ba\u03bf\u03bb\u03bf\u03c5\u03b8\u03af\u03b5\u03c2 \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd \u03c3\u03c4\u03b1 agent traces. \u03a3\u03c4\u03b7 \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03b7 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7, \u03bf\u03b9 \u03ba\u03b1\u03c4\u03b1\u03c3\u03c4\u03ac\u03c3\u03b5\u03b9\u03c2 \u03c3\u03c5\u03bd\u03b4\u03ad\u03bf\u03bd\u03c4\u03b1\u03b9 \u03ba\u03c5\u03c1\u03af\u03c9\u03c2 \u03bc\u03b5 \u03c4\u03b7\u03bd \u03c4\u03b5\u03bb\u03b5\u03c5\u03c4\u03b1\u03af\u03b1 \u03b4\u03c1\u03b1\u03c3\u03c4\u03b7\u03c1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a4\u03bf FSM \u03b1\u03bd\u03c4\u03b9\u03ba\u03b1\u03b8\u03b9\u03c3\u03c4\u03ac \u03c4\u03bf \u03af\u03b4\u03b9\u03bf \u03c4\u03bf LLM;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u038c\u03c7\u03b9. \u039b\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03b5\u03af \u03c9\u03c2 \u03b4\u03bf\u03bc\u03b9\u03ba\u03cc \u03b5\u03c0\u03af\u03c0\u03b5\u03b4\u03bf \u03c0\u03b1\u03c1\u03b1\u03c4\u03ae\u03c1\u03b7\u03c3\u03b7\u03c2 \u03ba\u03b1\u03b9 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7\u03c2 \u03b3\u03cd\u03c1\u03c9 \u03b1\u03c0\u03cc \u03c4\u03bf\u03bd agent. \u03a4\u03bf LLM \u03c3\u03c5\u03bd\u03b5\u03c7\u03af\u03b6\u03b5\u03b9 \u03bd\u03b1 \u03b5\u03c0\u03b9\u03bb\u03ad\u03b3\u03b5\u03b9 \u03ae \u03bd\u03b1 \u03c0\u03b1\u03c1\u03ac\u03b3\u03b5\u03b9 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b5\u03c2, \u03b5\u03bd\u03ce \u03c4\u03bf FSM \u03ba\u03b1\u03c4\u03b1\u03b3\u03c1\u03ac\u03c6\u03b5\u03b9 \u03c4\u03b7\u03bd \u03c4\u03bf\u03c0\u03bf\u03bb\u03bf\u03b3\u03af\u03b1 \u03c4\u03b7\u03c2 \u03c3\u03c5\u03bc\u03c0\u03b5\u03c1\u03b9\u03c6\u03bf\u03c1\u03ac\u03c2.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u039c\u03c0\u03bf\u03c1\u03b5\u03af \u03bd\u03b1 \u03c0\u03c1\u03bf\u03b2\u03bb\u03ad\u03c8\u03b5\u03b9 \u03bc\u03b5 \u03b2\u03b5\u03b2\u03b1\u03b9\u03cc\u03c4\u03b7\u03c4\u03b1 \u03bc\u03b9\u03b1 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u038c\u03c7\u03b9. \u03a4\u03b1 \u03b1\u03c0\u03bf\u03c4\u03b5\u03bb\u03ad\u03c3\u03bc\u03b1\u03c4\u03b1 \u03b5\u03ba\u03c6\u03c1\u03ac\u03b6\u03bf\u03bd\u03c4\u03b1\u03b9 \u03bc\u03b5 \u03bc\u03b5\u03c4\u03c1\u03b9\u03ba\u03ad\u03c2 \u03cc\u03c0\u03c9\u03c2 AUROC, precision \u03ba\u03b1\u03b9 recall. \u0397 \u03b1\u03c0\u03cc\u03b4\u03bf\u03c3\u03b7 \u03b4\u03b9\u03b1\u03c6\u03ad\u03c1\u03b5\u03b9 \u03b1\u03bd\u03ac dataset \u03ba\u03b1\u03b9 operating point, \u03b5\u03c0\u03bf\u03bc\u03ad\u03bd\u03c9\u03c2 \u03b1\u03c0\u03b1\u03b9\u03c4\u03b5\u03af\u03c4\u03b1\u03b9 \u03c4\u03bf\u03c0\u03b9\u03ba\u03ae \u03b5\u03c0\u03b9\u03ba\u03cd\u03c1\u03c9\u03c3\u03b7 \u03c0\u03c1\u03b9\u03bd \u03b1\u03c0\u03cc \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03b5\u03c2 \u03c0\u03b1\u03c1\u03b5\u03bc\u03b2\u03ac\u03c3\u03b5\u03b9\u03c2.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u0393\u03b9\u03b1\u03c4\u03af \u03b4\u03b5\u03bd \u03b1\u03c1\u03ba\u03b5\u03af \u03bd\u03b1 \u03bc\u03b5\u03c4\u03c1\u03ac\u03bc\u03b5 \u03c4\u03bf \u03bc\u03ae\u03ba\u03bf\u03c2 \u03b5\u03bd\u03cc\u03c2 trace;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u03a4\u03bf \u03bc\u03ae\u03ba\u03bf\u03c2 \u03b4\u03b5\u03bd \u03b4\u03b5\u03af\u03c7\u03bd\u03b5\u03b9 \u03c4\u03b7 \u03c3\u03b5\u03b9\u03c1\u03ac \u03ba\u03b1\u03b9 \u03c4\u03bf \u03c3\u03b7\u03bc\u03b5\u03af\u03bf \u03c4\u03c9\u03bd \u03b5\u03bd\u03b5\u03c1\u03b3\u03b5\u03b9\u03ce\u03bd. \u03a3\u03c4\u03bf SWE-agent, \u03c4\u03b1 \u03b4\u03bf\u03bc\u03b9\u03ba\u03ac \u03c7\u03b1\u03c1\u03b1\u03ba\u03c4\u03b7\u03c1\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac \u03c4\u03b7\u03c2 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7\u03c2 \u03be\u03b5\u03c0\u03ad\u03c1\u03b1\u03c3\u03b1\u03bd \u03b1\u03b9\u03c3\u03b8\u03b7\u03c4\u03ac \u03c4\u03bf \u03bc\u03ae\u03ba\u03bf\u03c2 \u03bc\u03cc\u03bd\u03bf \u03c4\u03bf\u03c5 \u03c3\u03c4\u03b7\u03bd \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03b1\u03c0\u03bf\u03c4\u03c5\u03c7\u03af\u03b1\u03c2.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a4\u03b9 \u03ad\u03b4\u03b5\u03b9\u03be\u03b5 \u03b7 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03bc\u03b5 \u03c4\u03bf Agent Workflow Memory;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u03a4\u03bf minimal FSM context \u03c0\u03ad\u03c4\u03c5\u03c7\u03b5 \u03c5\u03c8\u03b7\u03bb\u03cc\u03c4\u03b5\u03c1\u03b7 top-1 \u03c0\u03c1\u03cc\u03b2\u03bb\u03b5\u03c8\u03b7 \u03b5\u03c0\u03cc\u03bc\u03b5\u03bd\u03b7\u03c2 \u03b5\u03bd\u03ad\u03c1\u03b3\u03b5\u03b9\u03b1\u03c2 \u03ba\u03b1\u03b9 \u03c3\u03c4\u03b1 \u03bf\u03ba\u03c4\u03ce datasets \u03c4\u03b7\u03c2 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7\u03c2. \u0388\u03bd\u03b1\u03c2 \u03bb\u03cc\u03b3\u03bf\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03cc\u03c4\u03b9 \u03c4\u03bf AWM \u03b2\u03b1\u03c3\u03af\u03b6\u03b5\u03c4\u03b1\u03b9 \u03c3\u03b5 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03b7\u03bc\u03ad\u03bd\u03b1 workflows \u03ba\u03b1\u03b9 \u03c7\u03ac\u03bd\u03b5\u03b9 \u03ba\u03ac\u03bb\u03c5\u03c8\u03b7 \u03cc\u03c4\u03b1\u03bd \u03bf\u03b9 \u03b5\u03c0\u03b9\u03c4\u03c5\u03c7\u03af\u03b5\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bb\u03af\u03b3\u03b5\u03c2.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u0395\u03af\u03bd\u03b1\u03b9 \u03ba\u03b1\u03c4\u03ac\u03bb\u03bb\u03b7\u03bb\u03bf \u03b3\u03b9\u03b1 real-time monitoring;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u0397 \u03bc\u03b5\u03bb\u03ad\u03c4\u03b7 \u03b1\u03bd\u03b1\u03c6\u03ad\u03c1\u03b5\u03b9 0,006 ms \u03b1\u03bd\u03ac \u03b2\u03ae\u03bc\u03b1 \u03b3\u03b9\u03b1 \u03c4\u03bf\u03bd runtime monitor \u03c3\u03c4\u03bf \u03c3\u03c5\u03b3\u03ba\u03b5\u03ba\u03c1\u03b9\u03bc\u03ad\u03bd\u03bf \u03c0\u03b5\u03af\u03c1\u03b1\u03bc\u03b1. \u0397 \u03bc\u03b9\u03ba\u03c1\u03ae \u03b5\u03c0\u03b9\u03b2\u03ac\u03c1\u03c5\u03bd\u03c3\u03b7 \u03b5\u03af\u03bd\u03b1\u03b9 \u03b5\u03bd\u03b8\u03b1\u03c1\u03c1\u03c5\u03bd\u03c4\u03b9\u03ba\u03ae, \u03b1\u03bb\u03bb\u03ac \u03ba\u03ac\u03b8\u03b5 production workflow \u03c7\u03c1\u03b5\u03b9\u03ac\u03b6\u03b5\u03c4\u03b1\u03b9 \u03b4\u03b9\u03ba\u03ae \u03c4\u03bf\u03c5 \u03bc\u03ad\u03c4\u03c1\u03b7\u03c3\u03b7 \u03ba\u03b1\u03b9 \u03b2\u03b1\u03b8\u03bc\u03bf\u03bd\u03cc\u03bc\u03b7\u03c3\u03b7.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a0\u03bf\u03b9\u03bf\u03c2 \u03b5\u03af\u03bd\u03b1\u03b9 \u03bf \u03b2\u03b1\u03c3\u03b9\u03ba\u03cc\u03c4\u03b5\u03c1\u03bf\u03c2 \u03c4\u03b5\u03c7\u03bd\u03b9\u03ba\u03cc\u03c2 \u03c0\u03b5\u03c1\u03b9\u03bf\u03c1\u03b9\u03c3\u03bc\u03cc\u03c2;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u03a4\u03bf FSM \u03bc\u03bf\u03bd\u03c4\u03b5\u03bb\u03bf\u03c0\u03bf\u03b9\u03b5\u03af \u03c4\u03b9\u03c2 \u03c0\u03b1\u03c1\u03b1\u03c4\u03b7\u03c1\u03b7\u03bc\u03ad\u03bd\u03b5\u03c2 directly-follows \u03c3\u03c7\u03ad\u03c3\u03b5\u03b9\u03c2 \u03ba\u03b1\u03b9 \u03cc\u03c7\u03b9 \u03c4\u03b7\u03bd \u03c0\u03bb\u03ae\u03c1\u03b7 \u03b3\u03b5\u03bd\u03b5\u03c4\u03b9\u03ba\u03ae \u03b3\u03bb\u03ce\u03c3\u03c3\u03b1 \u03c4\u03bf\u03c5 agent. \u039c\u03c0\u03bf\u03c1\u03b5\u03af \u03c3\u03c5\u03bd\u03b5\u03c0\u03ce\u03c2 \u03bd\u03b1 \u03b1\u03c0\u03bf\u03b4\u03b5\u03c7\u03c4\u03b5\u03af \u03c4\u03b5\u03c7\u03bd\u03b7\u03c4\u03ac traces \u03c0\u03bf\u03c5 \u03b4\u03b9\u03b1\u03c4\u03b7\u03c1\u03bf\u03cd\u03bd \u03c4\u03bf\u03c0\u03b9\u03ba\u03ac \u03c3\u03c4\u03b1\u03c4\u03b9\u03c3\u03c4\u03b9\u03ba\u03ac \u03c7\u03c9\u03c1\u03af\u03c2 \u03bd\u03b1 \u03b5\u03af\u03bd\u03b1\u03b9 \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03ac \u03ad\u03b3\u03ba\u03c5\u03c1\u03b1.<\/p>\n<\/div>\n<\/details>\n<details class=\"td-faq-item\">\n<summary class=\"td-faq-title\">\u03a0\u03ce\u03c2 \u03be\u03b5\u03ba\u03b9\u03bd\u03ac \u03bc\u03b9\u03b1 \u03b5\u03c0\u03b9\u03c7\u03b5\u03af\u03c1\u03b7\u03c3\u03b7;<\/summary>\n<div class=\"td-faq-content\">\n<p>\u039c\u03b5 \u03c3\u03c4\u03b1\u03b8\u03b5\u03c1\u03ae \u03c4\u03b1\u03be\u03b9\u03bd\u03bf\u03bc\u03af\u03b1 activities, \u03b1\u03be\u03b9\u03cc\u03c0\u03b9\u03c3\u03c4\u03b1 outcome labels, \u03b1\u03c1\u03c7\u03b9\u03ba\u03ae observe-only \u03bb\u03b5\u03b9\u03c4\u03bf\u03c5\u03c1\u03b3\u03af\u03b1 \u03ba\u03b1\u03b9 \u03c3\u03cd\u03b3\u03ba\u03c1\u03b9\u03c3\u03b7 \u03c4\u03c9\u03bd \u03c3\u03b7\u03bc\u03ac\u03c4\u03c9\u03bd \u03bc\u03b5 \u03b1\u03bd\u03b8\u03c1\u03ce\u03c0\u03b9\u03bd\u03bf review. Alerts \u03ae \u03b1\u03c5\u03c4\u03cc\u03bc\u03b1\u03c4\u03b1 stops \u03c0\u03c1\u03bf\u03c3\u03c4\u03af\u03b8\u03b5\u03bd\u03c4\u03b1\u03b9 \u03bc\u03cc\u03bd\u03bf \u03bc\u03b5\u03c4\u03ac \u03b1\u03c0\u03cc validation \u03c3\u03c4\u03bf \u03c0\u03c1\u03b1\u03b3\u03bc\u03b1\u03c4\u03b9\u03ba\u03cc workflow.<\/p>\n<\/div>\n<\/details>\n<\/div>\n<\/section>\n<div class=\"td-source-list\">\n<p id=\"piges\" class=\"td-source-list-title\">\u0418\u0437\u0442\u043e\u0447\u043d\u0438\u0446\u0438<\/p>\n<ul>\n<li><a href=\"https:\/\/arxiv.org\/abs\/2608.23670\" target=\"_blank\" rel=\"noopener\">Cho et al.: Automata from Agent Traces, Failure and Next-Step Prediction<\/a><\/li>\n<li><a href=\"https:\/\/arxiv.org\/abs\/2409.07429\" target=\"_blank\" rel=\"noopener\">Wang et al.: Agent Workflow Memory<\/a><\/li>\n<li><a href=\"https:\/\/airc.nist.gov\/airmf-resources\/airmf\/5-sec-core\/\" target=\"_blank\" rel=\"noopener\">NIST: AI Risk Management Framework Core<\/a><\/li>\n<li><a href=\"https:\/\/openai.github.io\/openai-agents-python\/tracing\/\" target=\"_blank\" rel=\"noopener\">OpenAI Agents SDK: Tracing<\/a><\/li>\n<\/ul>\n<\/div>","protected":false},"excerpt":{"rendered":"<p>\u041a\u0430\u043a \u043c\u0430\u0448\u0438\u043d\u0438\u0442\u0435 \u0441 \u043a\u0440\u0430\u0435\u043d \u0431\u0440\u043e\u0439 \u0441\u044a\u0441\u0442\u043e\u044f\u043d\u0438\u044f \u043f\u0440\u0435\u043e\u0431\u0440\u0430\u0437\u0443\u0432\u0430\u0442 \u0441\u043b\u0435\u0434\u0438\u0442\u0435 \u043d\u0430 \u0430\u0433\u0435\u043d\u0442\u0438\u0442\u0435 \u0432\u044a\u0432 \u0444\u0443\u043d\u043a\u0446\u0438\u043e\u043d\u0430\u043b\u043d\u0430 \u043a\u0430\u0440\u0442\u0430 \u0437\u0430 \u043f\u0440\u043e\u0433\u043d\u043e\u0437\u0438\u0440\u0430\u043d\u0435 \u043d\u0430 \u0441\u043b\u0435\u0434\u0432\u0430\u0449\u0430\u0442\u0430 \u0441\u0442\u044a\u043f\u043a\u0430, \u043d\u0430\u0432\u0440\u0435\u043c\u0435\u043d\u043d\u043e \u043e\u0442\u043a\u0440\u0438\u0432\u0430\u043d\u0435 \u043d\u0430 \u0446\u0438\u043a\u043b\u0438 \u0438 \u043f\u043e-\u0431\u0435\u0437\u043e\u043f\u0430\u0441\u043d\u0430 \u0440\u0430\u0431\u043e\u0442\u0430 \u043d\u0430 AI \u0430\u0433\u0435\u043d\u0442\u0438\u0442\u0435.<\/p>","protected":false},"author":1,"featured_media":111960,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_gspb_post_css":"","footnotes":""},"categories":[199],"tags":[7231,6506,7477,18900,18449],"class_list":["post-104448","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-techniti-noimosyni","tag-agentic-ai","tag-ai-agents","tag-machine-learning","tag-asfaleia-ai","tag-aftomatismoi"],"_links":{"self":[{"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/posts\/104448","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/comments?post=104448"}],"version-history":[{"count":0,"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/posts\/104448\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/media\/111960"}],"wp:attachment":[{"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/media?parent=104448"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/categories?post=104448"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/twodots.gr\/bg\/wp-json\/wp\/v2\/tags?post=104448"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}