Supabaseを使ってあなたのWebサイト向けのRAGチャットボットを構築
上級
これはSupport Chatbot, AI RAG分野の自動化ワークフローで、18個のノードを含みます。主にFormTrigger, HtmlExtract, HttpRequest, ConvertToFile, Agentなどのノードを使用。 サイトのQ&Aチャットボットを構築：RAG、OpenAI GPT-4o-mini、Supabase ベクター データベース
前提条件
- •ターゲットAPIの認証情報が必要な場合あり
- •OpenAI API Key
- •PostgreSQLデータベース接続情報
- •Supabase URL と API Key
使用ノード (18)
ワークフロープレビュー
ノード接続関係を可視化、ズームとパンをサポート
ワークフローをエクスポート
以下のJSON設定をn8nにインポートして、このワークフローを使用できます
{
"id": "klsAKNm0Q07J8RXY",
"meta": {
"instanceId": "fcb622cbd5234c565b9b78babeaf0ea2455df9d2aad0d1adff7c1ef99583e685",
"templateCredsSetupCompleted": true
},
"name": "Build RAG Chatbot for Your Website Using Supabase",
"tags": [],
"nodes": [
{
"id": "c2c34da8-27e8-4881-a3fd-3b5f6987e0bf",
"name": "チャットメッセージ受信時",
"type": "@n8n/n8n-nodes-langchain.chatTrigger",
"position": [
500,
-580
],
"webhookId": "dec328cc-f47e-4727-b1c5-7370be86a958",
"parameters": {
"mode": "webhook",
"public": true,
"options": {}
},
"typeVersion": 1.1
},
{
"id": "31f5d600-1694-4d50-a581-0880f65a7755",
"name": "HTML抽出",
"type": "n8n-nodes-base.htmlExtract",
"position": [
-460,
-540
],
"parameters": {
"options": {},
"extractionValues": {
"values": [
{
"key": "title",
"cssSelector": "title, h1"
},
{
"key": "content",
"cssSelector": "p, article, .content, .post-content, main"
},
{
"key": "meta_description",
"attribute": "content",
"cssSelector": "meta[name='description']",
"returnValue": "attribute"
},
{
"key": "links",
"attribute": "href",
"cssSelector": "a[href]",
"returnValue": "attribute"
}
]
}
},
"typeVersion": 1
},
{
"id": "d48d0d67-cfb6-4058-96e2-4198b40650bc",
"name": "デフォルトデータローダー",
"type": "@n8n/n8n-nodes-langchain.documentDefaultDataLoader",
"position": [
20,
-360
],
"parameters": {
"loader": "jsonLoader",
"options": {
"metadata": {
"metadataValues": [
{
"name": "website_url",
"value": "={{ $('Enter Website Url').item.json['Website Url'] }}"
}
]
}
},
"dataType": "binary"
},
"typeVersion": 1
},
{
"id": "32b79786-e8d1-4083-80e9-4d2eada2c46b",
"name": "再帰的文字テキスト分割",
"type": "@n8n/n8n-nodes-langchain.textSplitterRecursiveCharacterTextSplitter",
"position": [
120,
-160
],
"parameters": {
"options": {},
"chunkSize": 3000,
"chunkOverlap": 500
},
"typeVersion": 1
},
{
"id": "b7541584-131f-4868-b8dd-053ab509ab1b",
"name": "Supabaseベクトルストア",
"type": "@n8n/n8n-nodes-langchain.vectorStoreSupabase",
"position": [
-40,
-540
],
"parameters": {
"mode": "insert",
"options": {},
"tableName": {
"__rl": true,
"mode": "list",
"value": "documents",
"cachedResultName": "documents"
},
"embeddingBatchSize": 100
},
"credentials": {
"supabaseApi": {
"id": "E3H0gensGX1OYVwU",
"name": "Supabase account"
}
},
"typeVersion": 1.1
},
{
"id": "7b923d0f-c803-4b3c-afd8-ee680793c955",
"name": "ウェブサイトURL入力",
"type": "n8n-nodes-base.formTrigger",
"position": [
-820,
-540
],
"webhookId": "b02a5096-11b0-4677-acfa-fcac5722498c",
"parameters": {
"options": {},
"formTitle": "Form",
"formFields": {
"values": [
{
"fieldLabel": "Website Url",
"placeholder": "Enter Website Url"
}
]
},
"formDescription": "Enter Website Url"
},
"typeVersion": 2.2
},
{
"id": "9e961fc5-b937-4f0e-b9f5-8ada4f3598f5",
"name": "付箋1",
"type": "n8n-nodes-base.stickyNote",
"position": [
-180,
-680
],
"parameters": {
"color": 3,
"width": 520,
"height": 100,
"content": "## Generating Embeddings from Website Content"
},
"typeVersion": 1
},
{
"id": "7f4feec8-62a8-47a0-a8d6-9bce01261b1a",
"name": "付箋2",
"type": "n8n-nodes-base.stickyNote",
"position": [
380,
-680
],
"parameters": {
"width": 300,
"height": 80,
"content": "## User-Initiated Question"
},
"typeVersion": 1
},
{
"id": "1c11199e-d217-4f2d-8afe-913f902ef47c",
"name": "付箋3",
"type": "n8n-nodes-base.stickyNote",
"position": [
700,
-680
],
"parameters": {
"color": 3,
"width": 540,
"height": 80,
"content": "## Vector-Based Answer Retrieval"
},
"typeVersion": 1
},
{
"id": "ddb37d99-4e8d-4eef-83da-6b3088360062",
"name": "付箋",
"type": "n8n-nodes-base.stickyNote",
"position": [
-840,
-680
],
"parameters": {
"width": 640,
"height": 100,
"content": "## Website Data Extraction\n\n"
},
"typeVersion": 1
},
{
"id": "eeffcae9-2e9c-48f2-ad13-2b75b7218fe2",
"name": "ウェブサイトデータスクレイピング",
"type": "n8n-nodes-base.httpRequest",
"maxTries": 3,
"position": [
-640,
-540
],
"parameters": {
"url": "={{ $json['Website Url'] }}",
"options": {
"timeout": 30000,
"redirect": {
"redirect": {
"maxRedirects": 5
}
}
}
},
"retryOnFail": true,
"typeVersion": 4.1
},
{
"id": "df8e6032-41f2-4f95-8906-7150ce565d8a",
"name": "ファイル変換",
"type": "n8n-nodes-base.convertToFile",
"position": [
-320,
-540
],
"parameters": {
"options": {},
"operation": "toJson"
},
"typeVersion": 1.1
},
{
"id": "697c29c6-6bcd-4748-a874-3c754f9b034e",
"name": "Cohere埋め込み",
"type": "@n8n/n8n-nodes-langchain.embeddingsCohere",
"position": [
-100,
-360
],
"parameters": {},
"credentials": {
"cohereApi": {
"id": "5VhbvbDzLliD5nO9",
"name": "CohereApi account"
}
},
"typeVersion": 1
},
{
"id": "d8d3a79a-1ac2-42dd-b383-e39bbcbded75",
"name": "質問＆回答検索",
"type": "@n8n/n8n-nodes-langchain.agent",
"position": [
820,
-580
],
"parameters": {
"options": {
"systemMessage": "=# Overview \nYou are an AI agent that responds strictly using information from stored document embeddings in a supabase vector store. \n\n## Context \n- The agent is designed to provide accurate, document-based answers. \n- It must avoid using external knowledge or assumptions. \n\n## Instructions \n1. Analyze the user's question. \n2. Retrieve relevant content from the document embeddings using supabase. \n3. Construct a response only from the matching document chunks.\n## Tools \n- Vector store containing embedded documents\n- Embeddings Cohere \n\n## SOP (Standard Operating Procedure) \n1. Receive user query. \n2. Search the vector store for relevant document chunks. \n3. Extract and synthesize an answer using only that content. \n4. If no matches, issue fallback message. \n\n## Final Notes \n- Do not rely on prior training or external facts. \n- The response must remain strictly grounded in the documents. "
}
},
"typeVersion": 1.8
},
{
"id": "65d51d86-cae3-4cdb-9d95-025962195953",
"name": "Supabaseベクトルストアからのデータ",
"type": "@n8n/n8n-nodes-langchain.vectorStoreSupabase",
"position": [
1000,
-360
],
"parameters": {
"mode": "retrieve-as-tool",
"topK": null,
"options": {},
"toolName": "documents_knowledge_base",
"tableName": {
"__rl": true,
"mode": "list",
"value": "documents",
"cachedResultName": "documents"
},
"toolDescription": "work with documents data in Supabase vector store",
"includeDocumentMetadata": false
},
"credentials": {
"supabaseApi": {
"id": "E3H0gensGX1OYVwU",
"name": "Supabase account"
}
},
"typeVersion": 1.1
},
{
"id": "6ef28803-e21f-4733-a6a9-d894b97dbb35",
"name": "Cohereによる埋め込み",
"type": "@n8n/n8n-nodes-langchain.embeddingsCohere",
"position": [
1020,
-140
],
"parameters": {},
"credentials": {
"cohereApi": {
"id": "5VhbvbDzLliD5nO9",
"name": "CohereApi account"
}
},
"typeVersion": 1
},
{
"id": "4796d1f3-ab76-4ed4-86b3-96bca2aae4e9",
"name": "チャットメモリ",
"type": "@n8n/n8n-nodes-langchain.memoryPostgresChat",
"position": [
900,
-320
],
"parameters": {
"tableName": "chat_memory",
"contextWindowLength": 3
},
"credentials": {
"postgres": {
"id": "lf7D6rk7s95k7I12",
"name": "Postgres account 2"
}
},
"typeVersion": 1.3
},
{
"id": "2ce21a19-e586-4cc6-88ed-b1957b04a06a",
"name": "OpenAIチャットモデル",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
760,
-340
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4o-mini"
},
"options": {}
},
"credentials": {
"openAiApi": {
"id": "FWzCPYhm2kVkOjE5",
"name": "OpenAi account 7"
}
},
"typeVersion": 1.2
}
],
"active": false,
"pinData": {},
"settings": {
"executionOrder": "v1"
},
"versionId": "1fe65722-2127-4e4c-8af6-d8c967259653",
"connections": {
"4796d1f3-ab76-4ed4-86b3-96bca2aae4e9": {
"ai_memory": [
[
{
"node": "d8d3a79a-1ac2-42dd-b383-e39bbcbded75",
"type": "ai_memory",
"index": 0
}
]
]
},
"31f5d600-1694-4d50-a581-0880f65a7755": {
"main": [
[
{
"node": "df8e6032-41f2-4f95-8906-7150ce565d8a",
"type": "main",
"index": 0
}
]
]
},
"df8e6032-41f2-4f95-8906-7150ce565d8a": {
"main": [
[
{
"node": "b7541584-131f-4868-b8dd-053ab509ab1b",
"type": "main",
"index": 0
}
]
]
},
"697c29c6-6bcd-4748-a874-3c754f9b034e": {
"ai_embedding": [
[
{
"node": "b7541584-131f-4868-b8dd-053ab509ab1b",
"type": "ai_embedding",
"index": 0
}
]
]
},
"7b923d0f-c803-4b3c-afd8-ee680793c955": {
"main": [
[
{
"node": "eeffcae9-2e9c-48f2-ad13-2b75b7218fe2",
"type": "main",
"index": 0
}
]
]
},
"2ce21a19-e586-4cc6-88ed-b1957b04a06a": {
"ai_languageModel": [
[
{
"node": "d8d3a79a-1ac2-42dd-b383-e39bbcbded75",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"d48d0d67-cfb6-4058-96e2-4198b40650bc": {
"ai_document": [
[
{
"node": "b7541584-131f-4868-b8dd-053ab509ab1b",
"type": "ai_document",
"index": 0
}
]
]
},
"6ef28803-e21f-4733-a6a9-d894b97dbb35": {
"ai_embedding": [
[
{
"node": "65d51d86-cae3-4cdb-9d95-025962195953",
"type": "ai_embedding",
"index": 0
}
]
]
},
"eeffcae9-2e9c-48f2-ad13-2b75b7218fe2": {
"main": [
[
{
"node": "31f5d600-1694-4d50-a581-0880f65a7755",
"type": "main",
"index": 0
}
]
]
},
"c2c34da8-27e8-4881-a3fd-3b5f6987e0bf": {
"main": [
[
{
"node": "d8d3a79a-1ac2-42dd-b383-e39bbcbded75",
"type": "main",
"index": 0
}
]
]
},
"65d51d86-cae3-4cdb-9d95-025962195953": {
"ai_tool": [
[
{
"node": "d8d3a79a-1ac2-42dd-b383-e39bbcbded75",
"type": "ai_tool",
"index": 0
}
]
]
},
"32b79786-e8d1-4083-80e9-4d2eada2c46b": {
"ai_textSplitter": [
[
{
"node": "d48d0d67-cfb6-4058-96e2-4198b40650bc",
"type": "ai_textSplitter",
"index": 0
}
]
]
}
}
}よくある質問
このワークフローの使い方は？
上記のJSON設定コードをコピーし、n8nインスタンスで新しいワークフローを作成して「JSONからインポート」を選択、設定を貼り付けて認証情報を必要に応じて変更してください。
このワークフローはどんな場面に適していますか？
上級 - サポートチャットボット, AI RAG検索拡張
有料ですか？
このワークフローは完全無料です。ただし、ワークフローで使用するサードパーティサービス（OpenAI APIなど）は別途料金が発生する場合があります。
関連ワークフロー
Telegram AIサポートチャットボット（マルチモーダル入力）
GPT-4とSupabase RAGを使って多言語Telegramサポートボットを作成
If
Set
Code
+
If
Set
Code
51 ノードEzema Kingsley Chibuzo
サポートチャットボット
リオーダー #1
Apify、GPT-4o、WhatsAppを使った営業のな冷水電話パイプラインの自動化
Set
Code
Webhook
+
Set
Code
Webhook
48 ノードKhairul Muhtadin
リードナーチャリング
n8nノードの探索（可視化リファレンスライブラリ内）
n8nノードを可視化リファレンスライブラリで探索
If
Ftp
Set
+
If
Ftp
Set
113 ノードI versus AI
その他
dian-memo_03のエクスポート
AI音声とテキストノート - LINEメッセージ、Supabaseベクトルデータベース、Gmailの統合
If
Set
Code
+
If
Set
Code
30 ノードkote2
AI RAG検索拡張
Rag を使って自動で再掲載する
RAG、Jina AI、OpenAI を使用したWordPressへの自動職缺抽出と掲載
If
Set
Code
+
If
Set
Code
56 ノードKhairul Muhtadin
人事
Google Drive を使用した RAG チャットボット
OpenAI、Google Drive、Supabaseを使用してRAG型知識チャットボットを構築する
Set
Supabase
Google Drive
+
Set
Supabase
Google Drive
20 ノードBabish Shrestha
その他
ワークフロー情報
難易度
上級
ノード数18
カテゴリー2
ノードタイプ13
作成者
InfyOm Technologies
@infyomWe are an AI Automation Agency focusing on various AI Automation Services like, workflow automation, shopify automation, chatbot developerment, custom ai agent development and more.
外部リンク
n8n.ioで表示 →
このワークフローを共有