Guillaume Duvernay This workflow corresponds to n8n.io template #12521 — we link there as the canonical source.
This workflow follows the Chainllm → Chat Trigger recipe pattern — see all workflows that pair these two integrations.
The workflow JSON
Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →
{
"nodes": [
{
"id": "60c2f847-f213-4ca2-8b93-85435a34613c",
"name": "When chat message received",
"type": "@n8n/n8n-nodes-langchain.chatTrigger",
"position": [
-176,
496
],
"parameters": {
"options": {
"responseMode": "responseNodes"
}
},
"typeVersion": 1.3
},
{
"id": "b99d2ceb-4a45-48b9-85ff-50d8b9e4c6fa",
"name": "Very small model",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
480,
944
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4.1-nano",
"cachedResultName": "gpt-4.1-nano"
},
"options": {}
},
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
},
"typeVersion": 1.2
},
{
"id": "798c5cdd-38b7-4883-979b-caf044011598",
"name": "Simple response",
"type": "@n8n/n8n-nodes-langchain.chainLlm",
"position": [
960,
304
],
"parameters": {
"text": "={{ $('When chat message received').item.json.chatInput }}",
"batching": {},
"messages": {
"messageValues": [
{
"message": "=You are a helpful assistant that assists the user.\n\nNB: Here are the previous messages from the conversation:\n\n{{ $('Find past messages').item.json.messages.toJsonString() }}"
}
]
},
"promptType": "define"
},
"typeVersion": 1.7
},
{
"id": "bd15aa9e-f288-4033-ba56-ae740f41de1f",
"name": "Prepare retrieval query",
"type": "@n8n/n8n-nodes-langchain.chainLlm",
"position": [
784,
704
],
"parameters": {
"text": "={{ $('When chat message received').item.json.chatInput }}",
"batching": {},
"messages": {
"messageValues": [
{
"message": "=Based on the user message, you will formulate the short and concise query to send through a knowledge retrieval tool that will enable you to retrieve all the information needed to actually answer that user message later on.\nDirectly output the query as a question.\n\nNB: Here are the previous messages from the conversation:\n\n{{ $('Find past messages').item.json.messages.toJsonString() }}"
}
]
},
"promptType": "define"
},
"typeVersion": 1.7
},
{
"id": "c07a6c01-9c13-47aa-896d-60e3f0f9992d",
"name": "Write the final response",
"type": "@n8n/n8n-nodes-langchain.chainLlm",
"position": [
1520,
704
],
"parameters": {
"text": "=Initial query:\n\n\"{{ $('When chat message received').item.json.chatInput }}\"\n\nKnowledge retrieval output:\n\"{{ $json.Output }}\"",
"batching": {},
"messages": {
"messageValues": [
{
"message": "=The user message contains a query and insights from a knowledge retrieval step to prepare you to actually write the final answer to that user.\n\nNB: Here are the previous messages from the conversation:\n\n{{ $('Find past messages').item.json.messages.toJsonString() }}"
}
]
},
"promptType": "define"
},
"typeVersion": 1.7
},
{
"id": "aa532ba1-b606-45c0-8bb2-baeb63daf714",
"name": "Mini model",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
848,
896
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4.1-mini"
},
"options": {}
},
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
},
"typeVersion": 1.2
},
{
"id": "834eb57c-ded3-46f9-8912-2a323f1f091f",
"name": "Large model",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
1584,
896
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4.1",
"cachedResultName": "gpt-4.1"
},
"options": {}
},
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
},
"typeVersion": 1.2
},
{
"id": "4e006500-9e61-45e3-bb20-28cf342667a8",
"name": "RAG via Lookio",
"type": "n8n-nodes-base.httpRequest",
"position": [
1216,
704
],
"parameters": {
"url": "=https://api.lookio.app/webhook/query",
"method": "POST",
"options": {},
"sendBody": true,
"sendHeaders": true,
"bodyParameters": {
"parameters": [
{
"name": "query",
"value": "={{ $json.text }}"
},
{
"name": "assistant_id",
"value": "<YOUR-ASSISTANT-ID>"
},
{
"name": "query_mode",
"value": "flash"
}
]
},
"headerParameters": {
"parameters": [
{
"name": "api_key",
"value": "<YOUR-API-KEY>"
}
]
}
},
"typeVersion": 4.2
},
{
"id": "d90f9ab0-5dd5-46ef-a87a-83f3e077fab9",
"name": "Intent router",
"type": "@n8n/n8n-nodes-langchain.textClassifier",
"position": [
400,
496
],
"parameters": {
"options": {
"systemPromptTemplate": "=Please classify the text provided by the user into one of the following categories: {categories}, and use the provided formatting instructions below. Don't explain, and only output the json.\n\n\nNB: Here are the previous messages from the conversation that came before this new text to categorize - focus on the new text to categorize though, this is just some context:\n\n{{ \n $json.messages\n .map(m => `human: ${m.human}\\nai: ${m.ai}`)\n .join('\\n')\n}}"
},
"inputText": "={{ $('When chat message received').item.json.chatInput }}",
"categories": {
"categories": [
{
"category": "=No knowledge retrieval needed to answer this query",
"description": "=Use this category when the message can be answered directly without consulting any external knowledge or documentation. This includes greetings, confirmations, small talk (e.g., \"hi\", \"hello\", \"thanks\", \"ok\")."
},
{
"category": "=Knowledge retrieval is needed to answer this query",
"description": "=Use this category when the message requires knowledge to be answered. Good use cases are when the user asks questions."
}
]
}
},
"typeVersion": 1.1
},
{
"id": "f83304af-f74a-4aed-8a64-d7b657a9abaa",
"name": "Respond to Chat",
"type": "@n8n/n8n-nodes-langchain.chat",
"position": [
1904,
496
],
"parameters": {
"message": "={{ $json.text }}",
"options": {
"memoryConnection": false
},
"waitUserReply": false
},
"typeVersion": 1
},
{
"id": "e89d675e-eac6-4e95-9238-d7b3501e1b2d",
"name": "Find past messages",
"type": "@n8n/n8n-nodes-langchain.memoryManager",
"position": [
48,
496
],
"parameters": {
"options": {}
},
"typeVersion": 1.1
},
{
"id": "735ad901-9c58-4851-bf1b-d93aaf46d557",
"name": "Simple Memory",
"type": "@n8n/n8n-nodes-langchain.memoryBufferWindow",
"position": [
208,
720
],
"parameters": {},
"typeVersion": 1.3
},
{
"id": "e1f44ef4-86c7-428b-8e6c-cc35d3181123",
"name": "Store messages",
"type": "@n8n/n8n-nodes-langchain.memoryManager",
"position": [
2128,
496
],
"parameters": {
"mode": "insert",
"messages": {
"messageValues": [
{
"type": "user",
"message": "={{ $('When chat message received').item.json.chatInput }}"
},
{
"type": "ai",
"message": "={{ $json.text }}"
}
]
}
},
"typeVersion": 1.1
},
{
"id": "61bf6661-fb9b-4179-b2b7-52f66c13cd39",
"name": "Sticky Note1",
"type": "n8n-nodes-base.stickyNote",
"position": [
1136,
672
],
"parameters": {
"color": 3,
"width": 256,
"height": 320,
"content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n## Action required\n\nMake sure to set your Lookio API key and workspace ID in here."
},
"typeVersion": 1
},
{
"id": "5a21675c-622b-48d5-b076-3c30eb861811",
"name": "Sticky Note",
"type": "n8n-nodes-base.stickyNote",
"position": [
-736,
224
],
"parameters": {
"width": 432,
"height": 576,
"content": "# **Smart & Efficient RAG Chatbot**\n\nThis workflow optimizes AI costs and speed by using an **Intent Router** pattern. \n\nInstead of one big model doing everything, it routes simple chats to small models and saves the heavy-duty LLMs for complex knowledge retrieval via **Lookio**.\n\n## **How to use**\n1. **Connect AI:** Add credentials to the three **OpenAI Language Model** nodes.\n2. **Configure Lookio:** In the **RAG via Lookio** node, replace the placeholders with your **API Key** and **Assistant ID** from [Lookio.app](https://www.lookio.app/).\n3. **Test:** Use the chat window to ask questions about your uploaded documents.\n\n**Note:** For more logic on how to break down agents into modular steps, watch [this video guide](https://www.youtube.com/watch?v=BHdJFnx2wrc).\n\n*A template created by Guillaume Duvernay*"
},
"typeVersion": 1
},
{
"id": "d7dde290-e168-4dd2-b6fd-5b1ee6a592bd",
"name": "Sticky Note2",
"type": "n8n-nodes-base.stickyNote",
"position": [
-240,
368
],
"parameters": {
"color": 7,
"width": 582,
"height": 320,
"content": "## 1. Receive user message & load the conversation"
},
"typeVersion": 1
},
{
"id": "744c2765-8163-4e99-bc8c-36d83cf659ab",
"name": "Sticky Note3",
"type": "n8n-nodes-base.stickyNote",
"position": [
352,
368
],
"parameters": {
"color": 7,
"width": 374,
"height": 320,
"content": "## 2. AI confirms if powerful RAG is needed"
},
"typeVersion": 1
},
{
"id": "9275c57e-f39a-4a29-bf23-0bf5082df61c",
"name": "Sticky Note4",
"type": "n8n-nodes-base.stickyNote",
"position": [
896,
208
],
"parameters": {
"color": 7,
"width": 374,
"height": 256,
"content": "## 3.1. writing a simple response"
},
"typeVersion": 1
},
{
"id": "37947331-e585-4f8e-ad43-90a2449a4458",
"name": "Sticky Note5",
"type": "n8n-nodes-base.stickyNote",
"position": [
752,
608
],
"parameters": {
"color": 7,
"width": 1046,
"height": 432,
"content": "## 3.2. Writing an AI knowledge retrieval based response"
},
"typeVersion": 1
},
{
"id": "16863033-491b-47df-9da3-ef4f190569d1",
"name": "Sticky Note6",
"type": "n8n-nodes-base.stickyNote",
"position": [
1840,
400
],
"parameters": {
"color": 7,
"width": 598,
"height": 272,
"content": "## 4. Response handling"
},
"typeVersion": 1
}
],
"connections": {
"Mini model": {
"ai_languageModel": [
[
{
"node": "Prepare retrieval query",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Large model": {
"ai_languageModel": [
[
{
"node": "Write the final response",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Intent router": {
"main": [
[
{
"node": "Simple response",
"type": "main",
"index": 0
}
],
[
{
"node": "Prepare retrieval query",
"type": "main",
"index": 0
}
]
]
},
"Simple Memory": {
"ai_memory": [
[
{
"node": "Find past messages",
"type": "ai_memory",
"index": 0
},
{
"node": "Store messages",
"type": "ai_memory",
"index": 0
}
]
]
},
"RAG via Lookio": {
"main": [
[
{
"node": "Write the final response",
"type": "main",
"index": 0
}
]
]
},
"Store messages": {
"main": [
[]
]
},
"Respond to Chat": {
"main": [
[
{
"node": "Store messages",
"type": "main",
"index": 0
}
]
]
},
"Simple response": {
"main": [
[
{
"node": "Respond to Chat",
"type": "main",
"index": 0
}
]
]
},
"Very small model": {
"ai_languageModel": [
[
{
"node": "Simple response",
"type": "ai_languageModel",
"index": 0
},
{
"node": "Intent router",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Find past messages": {
"main": [
[
{
"node": "Intent router",
"type": "main",
"index": 0
}
]
]
},
"Prepare retrieval query": {
"main": [
[
{
"node": "RAG via Lookio",
"type": "main",
"index": 0
}
]
]
},
"Write the final response": {
"main": [
[
{
"node": "Respond to Chat",
"type": "main",
"index": 0
}
]
]
},
"When chat message received": {
"main": [
[
{
"node": "Find past messages",
"type": "main",
"index": 0
}
]
]
}
}
}
Credentials you'll need
Each integration node will prompt for credentials when you import. We strip credential IDs before publishing — you'll add your own.
openAiApi
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
About this workflow
This template provides a high-performance, cost-optimized alternative to standard AI Agents for building RAG (Retrieval-Augmented Generation) chatbots.
Source: https://n8n.io/workflows/12521/ — original creator credit. Request a take-down →
Related workflows
Workflows that share integrations, category, or trigger type with this one. All free to copy and import.
This Chatbot automates the process of discovering job openings and generating tailored job application emails.
Chat Trigger, OpenAI Chat, Mcp Client Tool +12
This template attempts to create an AI-powered content assistant for WordPress sites using Mistral AI, enabling article recommendations, content summarization, and contextual Q&A capabilities.
Chat Trigger, Output Parser Structured, Agent +10
Who is this workflow for? This workflow is designed for SEO analysts, content creators, marketing agencies, and developers who need to index a website and then interact with its content as if it were
Agent, OpenAI Chat, Memory Buffer Window +10
This workflow is designed for IT teams, service desk agents, and operations managers who use ServiceNow. It reduces manual effort by automatically classifying chat messages as Incidents or Requests, c
Chat Trigger, OpenAI Chat, Text Classifier +7
This n8n template lets you build a smart AI chat assistant that can handle text, images, and PDFs — using OpenAI's GPT-4o multimodal model. It supports dynamic conversations and file analysis, making
OpenAI, Chain Llm, OpenAI Chat +4