{ "workflow": { "id": 3379, "name": "Automate product training & customer support via WhatsApp, GPT-4 & Google Sheets", "views": 37653, "recentViews": 5, "totalViews": 37653, "createdAt": "2025-03-30T16:46:52.017Z", "description": "![Workflow Screenshot](https://www.dr-firas.com/workflow_Automate_Product.png)\n\n# WhatsApp AI Agent: Auto-Train Product Data & Handle Customer Support\n\n## Who Is This For\n\nThis workflow is ideal for **eCommerce founders**, **product managers**, **customer support teams**, and **automation builders** who rely on WhatsApp to manage product information and interact with clients.\n\nIt’s perfect for businesses that want to **automate product data entry and support responses** directly from WhatsApp messages using GPT-4 and Google Sheets.\n\n## What Problem Does This Workflow Solve\n\n- **Manual Product Data Entry**: Collecting and organizing product data from links is tedious and error-prone. \n- **Slow Customer Response Times**: Responding to client questions manually leads to delays and inconsistent support. \n- **No Logging System for Issues**: Without automation, support issues often go undocumented, making it harder to learn and improve.\n\n## What This Workflow Does\n\n### Step 1 – Incoming Message Detection\n- Listens for incoming messages via WhatsApp.\n- If the message starts with `train:`, it routes to the product training process.\n- Otherwise, it routes to the customer support assistant.\n\n### Step 2 – Product Data Training\n- **Extracts URL** from the message using a regex script.\n- **Fetches HTML content** from the URL.\n- **Cleans HTML data** to extract readable product description.\n- **Saves raw data** (URL + description) into Google Sheets.\n- **Uses GPT-4** to enhance product data: \n → Name, price (one-time or subscription), topic, and FAQs. \n- **Updates the product row** in Google Sheets with structured information.\n\n### Step 3 – Customer Support Flow\n- Analyzes user messages with GPT-4 to understand the request or issue.\n- Looks up relevant product info in Google Sheets.\n- Detects potential problems (e.g. payment, login, delivery).\n- Suggests an appropriate solution.\n- Logs the problem, solution, and category to the `Customer Issues` sheet.\n- Sends a response back to the client via WhatsApp.\n\n### Step 4 – Client Response\n- Sends the AI-generated response to the client via WhatsApp.\n- Keeps the communication fast, clear, and professional.\n\n## Setup Guide\n\n### Prerequisites\n\n- **WhatsApp Business API access**\n- **OpenAI API Key**\n- **Google Account** with Google Sheets access\n- A hosted instance of **n8n (Cloud or self-hosted)**\n\n### Setup Steps\n\n1. **Import the Workflow** into your n8n instance.\n2. **Connect your credentials** for WhatsApp, OpenAI, and Google Sheets.\n3. **Customize Google Sheet IDs and names** as needed.\n4. **Test** by sending a `train:` message or a regular customer message to WhatsApp.\n5. **Activate the workflow** to make it live.\n\n## How to Customize This Workflow\n\n- **Edit AI prompts** to reflect your product type, language style, or tone.\n- **Change the trigger keyword** (e.g. from `train:` to `add:` or anything else).\n- **Add integrations** like Notion, Airtable, or CRM tools.\n- **Expand the Sheets structure** with more product fields (e.g. stock status, image link).\n- **Add notifications** to Slack or email after product updates or issue logging.\n\n📄 **Documentation**: [Notion Guide](https://automatisation.notion.site/WHATSAPP-WORKFLOW-FULL-1-1c63d6550fd9804d8303c08cd1083587?pvs=4)\n\n---\n\n### Need help customizing?\nContact me for consulting and support : [Linkedin](https://www.linkedin.com/in/doctor-firass/) / [Youtube](https:/https://www.youtube.com/@DRFIRASS)\n", "workflow": { "id": "GvlLeSOUPfar0DQS", "meta": { "instanceId": "a2b23892dd6989fda7c1209b381f5850373a7d2b85609624d7c2b7a092671d44", "templateCredsSetupCompleted": true }, "name": "Automate Product Data Collection & Customer Support via WhatsApp + GPT-4 + Google Sheets", "tags": [], "nodes": [ { "id": "370611ee-2725-4ee4-a73a-68d4eaabfe3a", "name": "WhatsApp Business Cloud", "type": "n8n-nodes-base.whatsApp", "position": [ 1020, 220 ], "webhookId": "ef261fd4-e3b8-42e9-a0e2-2c3c44b6b15e", "parameters": { "textBody": "={{ $json.output }}", "operation": "send", "phoneNumberId": "574583382412256", "additionalFields": {}, "recipientPhoneNumber": "+1234567890" }, "credentials": { "whatsAppApi": { "id": "credential-id", "name": "whatsAppApi Credential" } }, "typeVersion": 1 }, { "id": "30520fe8-a423-4942-85b2-9e8776b45939", "name": "Sticky Note", "type": "n8n-nodes-base.stickyNote", "position": [ -740, -20 ], "parameters": { "width": 400, "height": 440, "content": "## 🟡 STEP 1 – Incoming WhatsApp Message\nListens for new WhatsApp messages.\nIf the message starts with train:, it triggers the product training flow.\nOtherwise, it goes to the customer support flow.\n**WhatsApp Business Cloud node** : [Here](https://www.notion.so/automatisation/WHATSAPP-WORKFLOW-1c63d6550fd980559679e7535938a68d?pvs=4#1c63d6550fd980f9a2a5e25a3654da82)" }, "typeVersion": 1 }, { "id": "d3259e0e-5368-407d-8ac6-993784c939c9", "name": "Sticky Note1", "type": "n8n-nodes-base.stickyNote", "position": [ -300, 280 ], "parameters": { "color": 6, "width": 1180, "height": 420, "content": "## 🔵 STEP 3 – Customer Support Flow\nTriggered when the message **does NOT start with `train:`**.\n\n1. **AI Analyzes the Message** \n Understands the customer's question or problem.\n\n2. **Fetch Product Data if Needed** \n Reads relevant product info from Google Sheets.\n\n3. **Detect Issues Automatically** \n Identifies if the user is facing a specific issue.\n\n4. **Suggest Solutions** \n Proposes a helpful, clear response to solve the issue.\n\n5. **Log Customer Problems** \n Saves the problem, suggested solution, and category.\n\n6. **Respond to the User** \n Sends a professional and helpful WhatsApp reply." }, "typeVersion": 1 }, { "id": "455ce4ab-6d14-4d4e-8ff8-04612cf532b9", "name": "Sticky Note2", "type": "n8n-nodes-base.stickyNote", "position": [ -300, -400 ], "parameters": { "color": 3, "width": 1180, "height": 660, "content": "## 🔴 STEP 2 – Product Data Training (Triggered by train:)\nTriggered when the message starts with `train:`.\n**OpenAI API keys** : [Here](https://platform.openai.com/api-keys)\n\n\n\n\n\n\n\n\n\n\n\n\n1. **Extract URL from Message** \n Uses regex to detect and extract the product link.\n\n2. **Fetch HTML Content** \n Sends an HTTP request to retrieve the full page content.\n\n3. **Clean HTML Data** \n Strips HTML tags, scripts, and formats the text.\n\n4. **Save Raw Product Info** \n Appends the cleaned content and URL into Google Sheets.\n\n5. **AI: Enhance Product Details** \n - Extracts Product Name \n - Detects Price (subscription or one-time) \n - Identifies Product Topic \n - Generates FAQs\n\n6. **Update Product Sheet** \n Enriches the row in Sheets with structured product data.\n" }, "typeVersion": 1 }, { "id": "15251a04-a139-4a71-9e96-d5ecd9c2e66c", "name": "Sticky Note3", "type": "n8n-nodes-base.stickyNote", "position": [ 920, -20 ], "parameters": { "color": 4, "width": 320, "height": 440, "content": "## 🟢 STEP 4 – Client Response\nFinal step of the flow.\n\n- Sends the AI-generated response back to the customer via WhatsApp.\n- Ensures the message is clear, helpful, and personalized." }, "typeVersion": 1 }, { "id": "f028f5a3-fed3-4f90-89cc-2af211da7362", "name": "Incoming Message Trigger", "type": "n8n-nodes-base.whatsAppTrigger", "position": [ -700, 220 ], "webhookId": "48ce9bac-c7ea-4cb7-a40a-b010ea5ac743", "parameters": { "options": {}, "updates": [ "messages" ] }, "credentials": { "whatsAppTriggerApi": { "id": "credential-id", "name": "whatsAppTriggerApi Credential" } }, "typeVersion": 1 }, { "id": "54edaa0c-005e-4810-a2fa-76fbcc009b2b", "name": "Check If Training", "type": "n8n-nodes-base.switch", "position": [ -520, 220 ], "parameters": { "rules": { "values": [ { "outputKey": "train", "conditions": { "options": { "version": 2, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "9865cb5b-33da-490c-afc3-186457d5b564", "operator": { "type": "string", "operation": "startsWith" }, "leftValue": "={{ $json.messages[0].text.body }}", "rightValue": "train:" } ] }, "renameOutput": true }, { "outputKey": "text", "conditions": { "options": { "version": 2, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "5a9a1fee-b408-469f-a08c-e8d690fc9792", "operator": { "type": "string", "operation": "notStartsWith" }, "leftValue": "={{ $json.messages[0].text.body }}", "rightValue": "train:" } ] }, "renameOutput": true } ] }, "options": {} }, "typeVersion": 3.2 }, { "id": "b4187b28-efd9-46cb-8d1c-23d89d3f6125", "name": "Extract URL from Text", "type": "n8n-nodes-base.code", "position": [ -260, -320 ], "parameters": { "jsCode": "// Récupérer le texte à analyser depuis l'input\nconst texteExemple = $input.first().json.messages[0].text.body;\n\n// Expression régulière modifiée pour capturer les URL avec ou sans protocole\nconst regex = /((?:https?:\\/\\/)?(?:www\\.)?[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,6}(?:\\/\\S*)?)/g;\n\n// Extraction des URL à partir du texte\nconst matches = texteExemple.match(regex) || [];\n\n// Si des URL ont été trouvées, on les mappe en objets\nif (matches.length > 0) {\n const output = matches.map(url => ({ url }));\n return output;\n} else {\n // Si aucune URL n'est trouvée, on retourne les items d'entrée pour que le workflow continue\n return $input.all();\n}\n" }, "typeVersion": 2 }, { "id": "dbbcc8cf-2544-4de1-bcfd-6292d96a1017", "name": "Fetch HTML Page", "type": "n8n-nodes-base.httpRequest", "position": [ -100, -320 ], "parameters": { "url": "={{ $json.url }}", "options": { "response": { "response": { "responseFormat": "text" } } } }, "typeVersion": 4.2 }, { "id": "91e711d0-26c0-41dd-a9c5-28100f0f44f6", "name": "Clean HTML Content", "type": "n8n-nodes-base.code", "position": [ 60, -320 ], "parameters": { "jsCode": "// Exemple : récupération du contenu HTML depuis le premier item\n// Si vous avez un champ nommé \"html\" dans un noeud précédent.\nconst htmlContent = $input.first().json.html;\n\n// Fonction de nettoyage du HTML\nfunction nettoyerHTML(input) {\n if (typeof input !== 'string') {\n throw new Error(\"Expected HTML content as a string.\");\n }\n let cleanedText = input;\n\n // 1. Retirer les liens et leur contenu\n cleanedText = cleanedText.replace(/]*>.*?<\\/a>/gs, '');\n\n // 2. Retirer