{"id":735,"date":"2026-05-18T22:37:33","date_gmt":"2026-05-18T14:37:33","guid":{"rendered":"https:\/\/junai.ai\/blog\/n8n-ai-agent-19\/"},"modified":"2026-05-18T22:38:33","modified_gmt":"2026-05-18T14:38:33","slug":"n8n-ai-agent-19","status":"publish","type":"post","link":"https:\/\/junai.ai\/blog\/n8n-ai-agent-19\/","title":{"rendered":"n8n AI Agent \ub178\ub4dc \uc644\ubcbd \uc815\ub9ac (\uace0\uae09 19\ud3b8)"},"content":{"rendered":"\n<!-- WordPress REST API \ubc1c\ud589\uc6a9 HTML (\uc790\ub3d9 \uc0dd\uc131) -->\n<!-- WP-FEATURED-MEDIA-ID: 724 -->\n<div style=\"max-width:800px;margin:0 auto;\">\n<style>\n:root {\n  --color-primary: #4f46e5;\n  --color-accent: #6366f1;\n  --color-bg: #fafbfc;\n  --color-bg-card: #ffffff;\n  --color-text: #1a202c;\n  --color-text-muted: #64748b;\n  --hero-start: #1e1b4b;\n  --hero-end: #4338ca;\n  --code-bg: #0f172a;\n  --code-fg: #e2e8f0;\n}\n* { box-sizing: border-box; }\n.container { max-width: 760px; margin: 0 auto; padding: 0 24px 80px; }\n.hero {\n  background: linear-gradient(135deg, var(--hero-start) 0%, var(--hero-end) 100%);\n  color: #fff; padding: 80px 24px 60px; text-align: center;\n}\n.hero .eyebrow {\n  display: inline-block; font-size: 14px; color: #a5b4fc;\n  font-weight: 700; letter-spacing: 0.1em; text-transform: uppercase;\n  margin-bottom: 14px;\n}\n.hero h1 { font-size: 36px; margin: 0 0 16px; line-height: 1.3; font-weight: 800; }\n.hero p { color: #cbd5e1; font-size: 18px; max-width: 640px; margin: 0 auto; line-height: 1.6; }\n.hero img {\n  width: 100%; max-width: 640px; height: auto;\n  margin: 32px auto 0; border-radius: 10px; display: block;\n}\narticle { padding-top: 48px; }\narticle h2 {\n  font-size: 26px; margin: 56px 0 20px; padding-left: 14px;\n  border-left: 5px solid var(--color-accent); line-height: 1.4;\n}\narticle h3 { font-size: 19px; margin: 32px 0 12px; color: var(--color-primary); }\narticle p { margin: 16px 0; }\narticle strong { color: var(--color-primary); font-weight: 700; }\narticle code {\n  background: #f1f5f9; padding: 2px 8px; border-radius: 4px;\n  font-family: 'SF Mono', Menlo, Consolas, monospace;\n  font-size: 14px; color: #4338ca;\n}\n.databox {\n  background: #eef2ff; border-left: 4px solid var(--color-accent);\n  padding: 16px 20px; margin: 24px 0; border-radius: 0 8px 8px 0;\n  font-size: 15.5px;\n}\n.databox strong { color: var(--color-primary); }\n.warnbox {\n  background: linear-gradient(135deg, #fef3c7 0%, #fde68a 100%);\n  padding: 16px 20px; margin: 24px 0; border-radius: 8px;\n  font-size: 15.5px;\n}\n.tablewrap { overflow-x: auto; -webkit-overflow-scrolling: touch; margin: 22px 0; }\ntable { width: 100%; border-collapse: collapse; font-size: 15px; background: var(--color-bg-card); }\nth, td { padding: 11px 12px; text-align: left; border-bottom: 1px solid #e2e8f0; vertical-align: top; }\nth { background: #f1f5f9; font-weight: 700; color: #0f172a; }\ntd:first-child, th:first-child { font-weight: 700; }\n@media (max-width: 560px) {\n  .tablewrap table, .tablewrap thead, .tablewrap tbody, .tablewrap tr, .tablewrap th, .tablewrap td { display: block; width: auto; }\n  .tablewrap thead { display: none; }\n  .tablewrap tr { margin: 0 0 14px; border: 1px solid #e2e8f0; border-radius: 10px; overflow: hidden; }\n  .tablewrap td { border: none; border-bottom: 1px solid #f1f5f9; padding: 9px 14px; }\n  .tablewrap td:first-child { background: #f1f5f9; font-weight: 800; font-size: 15.5px; }\n  .tablewrap td:last-child { border-bottom: none; }\n  .tablewrap td[data-label]::before { content: attr(data-label) \" \u2014 \"; font-weight: 700; color: var(--color-primary); }\n}\n.cta {\n  background: linear-gradient(135deg, #4338ca 0%, #6366f1 100%);\n  color: #fff; padding: 28px 24px; border-radius: 12px;\n  margin: 48px 0 0; text-align: center;\n}\n.cta h3 { color: #fff; margin: 0 0 8px; font-size: 20px; }\n.cta p { color: #c7d2fe; margin: 0; font-size: 15.5px; }\n.footer-nav { margin-top: 32px; padding-top: 20px; border-top: 1px solid #e2e8f0; font-size: 14px; color: var(--color-text-muted); }\n.footer-nav a { color: var(--color-primary); text-decoration: none; }\n.footer-nav a:hover { text-decoration: underline; }\n@media (max-width: 480px) {\n  .hero h1 { font-size: 26px; }\n  .hero p { font-size: 16px; }\n  article h2 { font-size: 21px; }\n  article h3 { font-size: 17px; }\n}\n<\/style>\n<section class=\"hero\">\n  <span class=\"eyebrow\">n8n \uad50\uc7ac \u00b7 \uace0\uae09 19\ud3b8<\/span>\n  <h1>n8n AI Agent \ub178\ub4dc \u2014 Claude \u00b7 OpenAI \u00b7 LangChain \ud1b5\ud569<\/h1>\n  <p>\uc6cc\ud06c\ud50c\ub85c \uc548\uc5d0 \uc790\uc728\uc801\uc73c\ub85c \ub3c4\uad6c\ub97c \ud638\ucd9c\ud558\ub294 \uc5d0\uc774\uc804\ud2b8\ub97c \ubc15\ub294\ub2e4. LangChain \ucf54\ub4dc \ud55c \uc904 \uc5c6\uc774 5\ubd84 \ub9cc\uc5d0.<\/p>\n  <img decoding=\"async\" src=\"https:\/\/junai.ai\/blog\/wp-content\/uploads\/2026\/05\/hero-5-15.jpg\" alt=\"\uc911\uc559\uc758 AI Agent \uc624\ucf00\uc2a4\ud2b8\ub808\uc774\ud130 \ub178\ub4dc\uac00 \uac80\uc0c9\u00b7DB\u00b7\ucf54\ub4dc \ub3c4\uad6c\ub85c \uc5f0\uacb0\uc120\uc744 \ubed7\ub294 \uc77c\ub7ec\uc2a4\ud2b8\">\n<\/section>\n\n<div class=\"container\">\n<article>\n\n<p>2026\ub144 \ub4e4\uc5b4 <strong>n8n AI Agent \ub178\ub4dc<\/strong>\uac00 \uc548\uc815 \ub2e8\uacc4\ub85c \ub4e4\uc5b4\uc130\ub2e4. Claude 4 Sonnet\u00b7Opus \uc758 tool use \uac00 \uacac\uace0\ud574\uc9c0\uba74\uc11c &#8220;AI \uac00 \uc9c1\uc811 API \ud638\ucd9c \u2192 \uacb0\uacfc \uad00\ucc30 \u2192 \ub2e4\uc74c \uc561\uc158 \uacb0\uc815&#8221; \ud558\ub294 \uc790\uc728 \uc5d0\uc774\uc804\ud2b8\ub97c \ud504\ub85c\ub355\uc158\uc5d0 \ubc15\ub294 \uc0ac\ub840\uac00 \ub298\uc5c8\ub2e4.<\/p>\n\n<p>19\ud3b8\uc740 \ucc57\ubd07 \uc785\ubb38\uc774 \uc544\ub2c8\ub2e4. <strong>\uc65c LangChain \ucf54\ub4dc \ub300\uc2e0 n8n \uc778\uac00<\/strong>, <strong>Claude vs OpenAI \uc120\ud0dd \uae30\uc900<\/strong>, <strong>4 \ucef4\ud3ec\ub10c\ud2b8 \uad6c\uc870<\/strong>, <strong>\ub3c4\uad6c \ub178\ub4dc \ud568\uc815 3\uac00\uc9c0<\/strong>\ub97c \uc9da\uace0 5 \ub178\ub4dc \uc2e4\uc804 \uc6cc\ud06c\ud50c\ub85c\ub85c \ub05d\ub0b8\ub2e4. 1-14\ud3b8\uc744 \ub9c8\uce5c \ub3c5\uc790\uc758 \ub2e4\uc74c \ub2e8\uacc4.<\/p>\n\n<h2>1. AI Agent \ub178\ub4dc\ub294 \uacb0\uad6d &#8220;LangChain \uc758 \uc2dc\uac01 \ub808\uc774\uc5b4&#8221;<\/h2>\n\n<p>n8n \uacf5\uc2dd \ubb38\uc11c\uc5d0 \uba85\uc2dc \u2014 <strong>AI Agent \ub178\ub4dc\ub294 LangChain JavaScript \ud504\ub808\uc784\uc6cc\ud06c \uc704\uc5d0 \uc5b9\uc740 \uc2dc\uac01 \uc624\ucf00\uc2a4\ud2b8\ub808\uc774\uc158 \ub808\uc774\uc5b4<\/strong>\ub2e4. Python \uc73c\ub85c <code>from langchain.agents import create_tool_calling_agent<\/code> \uc4f0\ub294 \ub300\uc2e0, \uce94\ubc84\uc2a4\uc5d0\uc11c &#8220;\uc5d0\uc774\uc804\ud2b8 + LLM + \ub3c4\uad6c + \uba54\ubaa8\ub9ac&#8221; \ub97c \ub4dc\ub798\uadf8\ub85c \uc870\ub9bd.<\/p>\n\n<p>\ub300\ub2e4\uc218 \ud29c\ud1a0\ub9ac\uc5bc\uc740 &#8220;n8n AI Agent = \ucc57\ubd07 \ub9cc\ub4dc\ub294 \ub3c4\uad6c&#8221; \ub77c\uace0 \ub2e8\uc21c\ud654\ud558\uc9c0\ub9cc \uc785\uad6c\ub9cc \ubcf8 \uac83. \uc9c4\uc9dc \uac00\uce58\ub294 LangChain \uc758 <em>tool calling loop<\/em> (LLM \uc774 \ub3c4\uad6c \uc120\ud0dd \u2192 \ud638\ucd9c \u2192 \uacb0\uacfc \uad00\ucc30 \u2192 \ub2e4\uc74c \ud589\ub3d9 \uacb0\uc815) \uac00 \ud1b5\uc9f8\ub85c \ub4e4\uc5b4\uc788\ub2e4\ub294 \uc810.<\/p>\n\n<div class=\"databox\">\n<strong>\ud55c \uc904 \uc815\ub9ac<\/strong> \u2014 n8n AI Agent \ub178\ub4dc\ub294 LangChain ReAct\u00b7Tools agent \ud328\ud134\uc744 \uc2dc\uac01 \ub178\ub4dc\ub85c \ud480\uc5b4\ub193\uc740 \uac83. \ucf54\ub4dc \uc5c6\uc774 \uac19\uc740 reasoning loop \ub97c \uc6cc\ud06c\ud50c\ub85c\uc5d0 \ubc15\ub294\ub2e4.\n<\/div>\n\n<h2>2. 4\uac1c \ucef4\ud3ec\ub10c\ud2b8 \uad6c\uc870 \u2014 \ube60\ub728\ub9ac\uba74 \uc548 \ub428<\/h2>\n\n<p>\uc2e4\uc81c AI Agent \uc6cc\ud06c\ud50c\ub85c 1\uac74\uc774 \ub3d9\uc791\ud558\ub824\uba74 <strong>\uc815\ud655\ud788 4\uac1c<\/strong>\uc758 \ucef4\ud3ec\ub10c\ud2b8\uac00 \ud544\uc694\ud558\ub2e4. \uc785\ubb38\uc790\uac00 &#8220;AI Agent \ub178\ub4dc 1\uac1c\ub9cc \ubc15\uc73c\uba74 \ub418\ub294 \uac70 \uc544\ub2cc\uac00&#8221; \uc624\ud574\ud558\uae30 \uc26c\uc6b4\ub370, \uadf8 \ub178\ub4dc\ub294 <em>\uc624\ucf00\uc2a4\ud2b8\ub808\uc774\ud130<\/em>\uc77c \ubfd0\uc774\ub2e4.<\/p>\n\n<div class=\"tablewrap\">\n<table>\n<thead><tr><th>\uc5ed\ud560<\/th><th>\uc608\uc2dc \ub178\ub4dc<\/th><th>\uc5c6\uc73c\uba74?<\/th><\/tr><\/thead>\n<tbody>\n<tr><td>\u2460 \ud2b8\ub9ac\uac70<\/td><td data-label=\"\uc608\uc2dc \ub178\ub4dc\">Webhook \u00b7 Schedule \u00b7 Chat Trigger \u00b7 HTTP Request<\/td><td data-label=\"\uc5c6\uc73c\uba74?\">\uc5d0\uc774\uc804\ud2b8\uac00 \uc5b8\uc81c \uae68\uc5b4\ub0a0\uc9c0 \ubaa8\ub984 (\uc2e4\ud589 \uc790\uccb4 \ubd88\uac00)<\/td><\/tr>\n<tr><td>\u2461 AI Agent \ub178\ub4dc<\/td><td data-label=\"\uc608\uc2dc \ub178\ub4dc\">AI Agent (root) \u2014 Tools Agent \/ Conversational Agent \/ ReAct<\/td><td data-label=\"\uc5c6\uc73c\uba74?\">\ub3c4\uad6c \ud638\ucd9c\u00b7\uad00\ucc30\u00b7\ub2e4\uc74c \uc561\uc158 \uacb0\uc815 \ub8e8\ud504 \ubd80\uc7ac<\/td><\/tr>\n<tr><td>\u2462 \uc11c\ube0c\ub178\ub4dc (3\uc885)<\/td><td data-label=\"\uc608\uc2dc \ub178\ub4dc\">Chat Model (Claude\/OpenAI\/Gemini), Memory (Buffer\/Postgres), Tool (HTTP\u00b7Code\u00b7MCP)<\/td><td data-label=\"\uc5c6\uc73c\uba74?\">\uc5b8\uc5b4\ubaa8\ub378\u00b7\uae30\uc5b5\u00b7\uc678\ubd80 \uc561\uc158 \uc911 \ubb34\uc5c7\uc774\ub4e0 \ube60\uc9c0\uba74 \uc5d0\uc774\uc804\ud2b8\uac00 \uc808\ub69d\uac70\ub9bc<\/td><\/tr>\n<tr><td>\u2463 \ucd9c\ub825<\/td><td data-label=\"\uc608\uc2dc \ub178\ub4dc\">Slack \u00b7 Email \u00b7 DB write \u00b7 Respond to Webhook<\/td><td data-label=\"\uc5c6\uc73c\uba74?\">\uacb0\uacfc\uac00 n8n \uc548\uc5d0 \uac07\ud600\uc11c \uc0ac\uc6a9\uc790\uac00 \ubabb \ubd04<\/td><\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n\n<p>\ud575\uc2ec\uc740 \u2462\ubc88. <strong>Chat Model \uc11c\ube0c\ub178\ub4dc\ub97c \uac08\uc544\ub07c\ub294 \uac83\ub9cc\uc73c\ub85c OpenAI \u2194 Claude \u2194 Gemini \uc804\ud658\uc774 \ub05d\ub09c\ub2e4.<\/strong> \ub178\ub4dc connection \ub9cc \ubc14\uafb8\uba74 \ubaa8\ub378 A\/B \ud14c\uc2a4\ud2b8 \ud55c \ubc88 \ud074\ub9ad. Python \uc774\ub77c\uba74 import\u00b7retry \ub85c\uc9c1 \ub2e4 \ub2e4\uc2dc \uc9dc\uc57c \ud560 \uc77c.<\/p>\n\n<h2>3. Claude vs OpenAI \u2014 \uc9c4\uc9dc \uc120\ud0dd \uae30\uc900<\/h2>\n\n<p>&#8220;Claude \uac00 \ub354 \ub611\ub611\ud55c\uac00 OpenAI \uac00 \ub354 \ub611\ub611\ud55c\uac00&#8221; \uc2dd \ube44\uad50\ub294 \ubb34\uc758\ubbf8\ud558\ub2e4. \ub450 \ubaa8\ub378\uc740 <strong>\ub2e4\ub978 \uc77c\uc744 \uc798\ud55c\ub2e4<\/strong>. n8n \uc548\uc5d0\uc11c Chat Model \uc11c\ube0c\ub178\ub4dc\ub97c \uace0\ub97c \ub54c \uc2e4\uc804 \uae30\uc900\uc740 \uc774\ub807\ub2e4.<\/p>\n\n<div class=\"tablewrap\">\n<table>\n<thead><tr><th>\uc0c1\ud669<\/th><th>Claude \uc6b0\uc704<\/th><th>OpenAI \uc6b0\uc704<\/th><\/tr><\/thead>\n<tbody>\n<tr><td>\uae34 \ubb38\uc11c \ucc98\ub9ac<\/td><td data-label=\"Claude \uc6b0\uc704\">\u2605 Sonnet 4 200K \ucee8\ud14d\uc2a4\ud2b8\ub85c PDF\u00b7\ucf54\ub4dc\ubca0\uc774\uc2a4\u00b7\uacc4\uc57d\uc11c 1\ud328\uc2a4 \ucc98\ub9ac<\/td><td data-label=\"OpenAI \uc6b0\uc704\">128K \ud55c\uacc4 \u2014 \ud070 PR \ub4e4\uc5b4\uac00\uba74 \uc798\ub9bc<\/td><\/tr>\n<tr><td>JSON schema \uac15\uc81c<\/td><td data-label=\"Claude \uc6b0\uc704\">tool_use \uc548\uc815\uc801\uc774\uc9c0\ub9cc strict \ubaa8\ub4dc \uc5c6\uc74c<\/td><td data-label=\"OpenAI \uc6b0\uc704\">\u2605 <code>strict: true<\/code> \ub85c schema 100% \ubcf4\uc7a5 (function calling)<\/td><\/tr>\n<tr><td>\ud55c\uad6d\uc5b4 \uc790\uc5f0\ub3c4<\/td><td data-label=\"Claude \uc6b0\uc704\">\u2605 \ud55c\uad6d\uc5b4 \ud45c\ud604\u00b7\uc874\ub313\ub9d0 \ub258\uc559\uc2a4 \uc6b0\uc704 (\uc0ac\uc6a9\uc790 \uc751\ub300)<\/td><td data-label=\"OpenAI \uc6b0\uc704\">\uae30\ub2a5\uc801\uc73c\ub860 \ucda9\ubd84<\/td><\/tr>\n<tr><td>\uc774\ubbf8\uc9c0 \uc0dd\uc131 \uac19\uc774<\/td><td data-label=\"Claude \uc6b0\uc704\">\ubcc4\ub3c4 \ub3c4\uad6c \ud638\ucd9c \ud544\uc694<\/td><td data-label=\"OpenAI \uc6b0\uc704\">\u2605 DALL\u00b7E \uac19\uc740 \uc6cc\ud06c\ud50c\ub85c \uc548\uc5d0\uc11c \ud638\ucd9c<\/td><\/tr>\n<tr><td>\ube44\uc6a9 (1M \ud1a0\ud070 \ucd9c\ub825)<\/td><td data-label=\"Claude \uc6b0\uc704\">Sonnet 4 \u2248 $15 \u00b7 Haiku 4.5 \u2248 $5<\/td><td data-label=\"OpenAI \uc6b0\uc704\">GPT-4o \u2248 $10 \u00b7 4o-mini \u2248 $0.6<\/td><\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n\n<p>\uc2e4\uc804 \ub77c\uc6b0\ud305 \ud328\ud134 \ud55c \uac00\uc9c0 \u2014 <strong>Switch \ub178\ub4dc + \ub450 AI Agent \ub178\ub4dc \ubcd1\ub82c \ubc30\uce58<\/strong>. \uc0ac\uc6a9\uc790 \uc9c8\ubb38\uc744 \uba3c\uc800 \ubd84\ub958 \ub178\ub4dc (Haiku \uac00 \ube60\ub974\uace0 \uc2f8\ub2e4) \uac00 \ubcf4\uace0, &#8220;\uae34 \ubb38\uc11c \ubd84\uc11d&#8221; \ub77c\uc6b0\ud2b8\ub294 Claude Sonnet 4, &#8220;\uad6c\uc870\ud654\ub41c \ub370\uc774\ud130 \ucd94\ucd9c&#8221; \ub77c\uc6b0\ud2b8\ub294 GPT-4o \ub85c \ubcf4\ub0b8\ub2e4. \ud55c \uc6cc\ud06c\ud50c\ub85c\uc5d0\uc11c \ub450 \ubaa8\ub378\uc758 \uc7a5\uc810\ub9cc \ubf51\uc544 \uc4f0\ub294 \ud328\ud134\uc774\ub2e4.<\/p>\n\n<h2>4. \ub3c4\uad6c(Tool) \ub178\ub4dc \ud568\uc815 3\uac00\uc9c0<\/h2>\n\n<p>AI Agent \uac00 \uc790\uc728 \ud589\ub3d9\ud558\ub824\uba74 <strong>\ub3c4\uad6c \uc11c\ube0c\ub178\ub4dc<\/strong>\uac00 \ud544\uc694. n8n \uc740 4\uc885 \uc81c\uacf5: HTTP Request Tool, Code Tool, Calculator Tool, \uadf8\ub9ac\uace0 <strong>\ub2e4\ub978 \uc6cc\ud06c\ud50c\ub85c\ub97c \ub3c4\uad6c\ud654\ud558\ub294 Sub-Workflow Tool<\/strong>. \ub9c8\uc9c0\ub9c9 \uac8c \uac00\uc7a5 \uac15\ub825\ud55c\ub370 \uc6b4\uc601\uc790\ub4e4\uc774 \ud754\ud788 \ube60\ub728\ub9b0\ub2e4.<\/p>\n\n<h3>\ud568\uc815 1 \u2014 \ub3c4\uad6c description \uc744 \ud55c \uc904\ub85c \ub05d\ub0b4\uae30<\/h3>\n\n<p>LLM \uc740 <code>description<\/code> \ub9cc \ubcf4\uace0 \ub3c4\uad6c \ud638\ucd9c \uc5ec\ubd80 \ud310\ub2e8. &#8220;Search the web&#8221; \uac19\uc740 \ucd94\uc0c1 \uc124\uba85\uc774\uba74 \uc5c9\ub6b1\ud55c \ud0c0\uc774\ubc0d\uc5d0 \ud638\ucd9c\ud558\uac70\ub098 \uc544\uc608 \uc548 \ubd80\ub978\ub2e4. <strong>&#8220;2026 \ub144 \ucd5c\uc2e0 \ud1b5\uacc4\u00b7\uc774\ubca4\ud2b8\u00b7\ub274\uc2a4 \uc9c8\ubb38\uc77c \ub54c\ub9cc \uc0ac\uc6a9. \uc77c\ubc18 \uc9c0\uc2dd\uc5d4 \uae08\uc9c0&#8221;<\/strong> \ucc98\ub7fc <em>\ubd80\ub97c \uc870\uac74 + \uc548 \ubd80\ub97c \uc870\uac74<\/em> \ub458 \ub2e4 \uc801\uc5b4\uc57c \uc815\ud655\ub3c4\uac00 \ud655 \uc624\ub978\ub2e4.<\/p>\n\n<h3>\ud568\uc815 2 \u2014 Memory \ub178\ub4dc \uc5c6\uc774 \uba40\ud2f0\ud134<\/h3>\n\n<p>Memory \uc11c\ube0c\ub178\ub4dc \ube7c\uba74 \ub9e4 \ud638\ucd9c\uc774 \uccab \ub300\ud654. &#8220;\ubc29\uae08 \uadf8\uac70 \ud55c\uad6d\uc5b4\ub85c&#8221; \uac00 \ubb54\uc9c0 \ubaa8\ub978\ub2e4. \ub370\ubaa8\ub294 Window Buffer Memory, \uc6b4\uc601\uc740 <strong>Postgres Chat Memory<\/strong> \ub85c \uc138\uc158\ubcc4 \uc601\uc18d \uc800\uc7a5. session ID \ub294 webhook \uc785\ub825\uc5d0\uc11c \uba85\uc2dc \ucd94\ucd9c \u2014 \ub2e4\ub978 \uc0ac\uc6a9\uc790 \ub300\ud654 \uc11e\uc774\ub294 \uac8c \uac00\uc7a5 \ud754\ud55c \uc6b4\uc601 \ubc84\uadf8.<\/p>\n\n<h3>\ud568\uc815 3 \u2014 \ubb34\ud55c \ub8e8\ud504 \ubc29\uc9c0 \uc5c6\uc774 \ubc30\ud3ec<\/h3>\n\n<p>AI Agent \uc5d0 <strong>Max Iterations<\/strong> \uc635\uc158 \uc788\uc74c (\uae30\ubcf8 10). \uc798\ubabb\ud558\uba74 \uac19\uc740 \ub3c4\uad6c 30\ubc88 \ud638\ucd9c\ud558\uba70 \ud1a0\ud070 \ub2e4 \ud0dc\uc6c0. \uc6b4\uc601 \ubc30\ud3ec \uc804 <strong>5 \uc774\ud558\ub85c \ub0ae\ucd94\uace0 \ube44\uc815\uc0c1 \uc885\ub8cc \uc54c\ub9bc<\/strong>\uc744 Error Workflow \ub85c \uac78\uc5b4\uc57c (17\ud3b8\uc5d0\uc11c \uc790\uc138\ud788).<\/p>\n\n<div class=\"warnbox\">\n<strong>\ube44\uc6a9 \uc0ac\uace0<\/strong> \u2014 Max Iterations 10 + Claude Opus 4 + \ud070 \ub3c4\uad6c \uc751\ub2f5\uc774 \uacb9\uce58\uba74 1\ud68c \uc2e4\ud589\uc5d0 $1~3 \ud284 \uc0ac\ub840 \uc788\uc74c. \uc6b4\uc601 \uccab \uc8fc \ub9e4\uc77c cost \ubaa8\ub2c8\ud130\ub9c1 \ud544\uc218. Anthropic \ucf58\uc194 <em>Usage by API Key<\/em> \u00b7 OpenAI <em>Project usage<\/em> \uc77c 1\ud68c \ud655\uc778.\n<\/div>\n\n<h2>5. \uccab AI Agent \u2014 5\ub178\ub4dc Webhook + Claude \uc6cc\ud06c\ud50c\ub85c<\/h2>\n\n<p>\uc774\ub860 \ub05d, \uc2e4\uc804 1\uac74\uc73c\ub85c \ub9c8\ubb34\ub9ac. <strong>Webhook \uc785\ub825 \u2192 Claude Sonnet 4 \ub2f5\ubcc0 + \ud544\uc694\uc2dc HTTP \ub3c4\uad6c \ud638\ucd9c \u2192 Slack \ucd9c\ub825<\/strong>.<\/p>\n\n<ol>\n<li><strong>Webhook<\/strong> \u2014 POST <code>\/ai-assistant<\/code>, body <code>{\"question\":\"...\",\"session_id\":\"...\"}<\/code><\/li>\n<li><strong>AI Agent (Tools Agent)<\/strong> \u2014 User Message <code>={{$json.question}}<\/code>, Session ID <code>={{$json.session_id}}<\/code><\/li>\n<li><strong>Anthropic Chat Model<\/strong> \u2014 Model: \ucd5c\uc2e0 sonnet, Max Tokens 2048, Temperature 0.7<\/li>\n<li><strong>HTTP Request Tool<\/strong> \u2014 Description \uc5d0 &#8220;\ud658\uc728\u00b7\ub0a0\uc528\u00b7\uc8fc\uac00 \ub4f1 \uc2e4\uc2dc\uac04 \ub370\uc774\ud130 \ud544\uc694\ud560 \ub54c\ub9cc&#8221; \uba85\uc2dc<\/li>\n<li><strong>Window Buffer Memory<\/strong> \u2014 Window Length 5, Session Key <code>={{$json.session_id}}<\/code><\/li>\n<li><strong>Slack \ucd9c\ub825<\/strong> \u2014 Channel \uc54c\ub9bc\uc6a9, Text <code>={{$json.output}}<\/code><\/li>\n<\/ol>\n\n<p>\ubb18\ubbf8\ub294 <strong>HTTP Request Tool \uc758 \uc790\uc728\uc131<\/strong>. &#8220;\uc624\ub298 \uc11c\uc6b8 \ub0a0\uc528&#8221; \ubb3c\uc73c\uba74 LLM \uc774 OpenWeatherMap URL \uc744 \uc54c\uc544\uc11c \uc0dd\uc131\u00b7\ud638\ucd9c\u00b7\ub2f5\ubcc0. if-else \ubd88\ud544\uc694.<\/p>\n\n<p>\ub2e4\uc74c\uc740 <strong>Sub-Workflow Tool<\/strong> \u2014 \ub2e4\ub978 \uc6cc\ud06c\ud50c\ub85c\ub97c \ub3c4\uad6c\ud654\ud574 \uba40\ud2f0 \uc5d0\uc774\uc804\ud2b8 \uc313\uae30. 16\u00b717\u00b720\ud3b8\uae4c\uc9c0 \ub05d\ub098\uba74 production \ubc15\uc744 \uc900\ube44 \uc644\ub8cc.<\/p>\n\n<div class=\"cta\">\n<h3>\ub2e4\uc74c \uae00<\/h3>\n<p>n8n \uad50\uc7ac 17\ud3b8 \u2014 Error Workflow + \uc7ac\uc2dc\ub3c4 \uc804\ub7b5. AI Agent \ub3c4\uad6c \ud638\ucd9c \uc2e4\ud328\u00b7timeout \uc2dc \uc790\ub3d9 \ubcf5\uad6c \ud328\ud134.<\/p>\n<\/div>\n\n<div class=\"footer-nav\">\nn8n \uad50\uc7ac \uc2dc\ub9ac\uc988 \u00b7\n<a href=\"https:\/\/junai.ai\/blog\/n8n-intro-1\/\">1\ud3b8 \uc785\ubb38<\/a> \u00b7\n<a href=\"https:\/\/junai.ai\/blog\/n8n-code-node-13\/\">13\ud3b8 Code \ub178\ub4dc<\/a> \u00b7\n<a href=\"https:\/\/junai.ai\/blog\/n8n-loop-batch-14\/\">14\ud3b8 Loop\u00b7Batch<\/a> \u00b7\n<strong>19\ud3b8 AI Agent<\/strong>\n<\/div>\n\n<\/article>\n<\/div>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>n8n AI Agent \ub178\ub4dc \u2014 LangChain \uc2dc\uac01 \ub808\uc774\uc5b4. Claude vs OpenAI \uc120\ud0dd, 4 \ucef4\ud3ec\ub10c\ud2b8, \ub3c4\uad6c \ud568\uc815 3\uac00\uc9c0, 5\ub178\ub4dc \uc2e4\uc804.<\/p>\n","protected":false},"author":1,"featured_media":724,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[21],"tags":[],"class_list":["post-735","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-n8n"],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/posts\/735","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/comments?post=735"}],"version-history":[{"count":1,"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/posts\/735\/revisions"}],"predecessor-version":[{"id":737,"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/posts\/735\/revisions\/737"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/media\/724"}],"wp:attachment":[{"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/media?parent=735"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/categories?post=735"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/junai.ai\/blog\/wp-json\/wp\/v2\/tags?post=735"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}