{ "name": "Telegram AI Expense Tracker", "nodes": [ { "parameters": { "content": "## Telegram AI Expense Tracker: Track Expenses via Bot to Google Sheets\n\nTrack expenses by sending simple messages to a Telegram bot (like 'Lunch $25'), have AI extract the details, and save everything to a Google Sheet for analysis.\n\n**Setup time:** 15 minutes · **Level:** Intermediate\n\n**You'll need accounts/credentials for:** Telegram, OpenAI, Google Sheets\n\n**Replace these before running:**\n- `REPLACE_WITH_BOT_TOKEN`\n- `REPLACE_WITH_SPREADSHEET`\n\n**Setup steps:**\n1. Connect Telegram, OpenAI, and Google Sheets in their respective nodes.\n2. Replace REPLACE_WITH_BOT_TOKEN with your Telegram bot token.\n3. Replace REPLACE_WITH_SPREADSHEET with your Google Sheet ID.\n4. Create a Google Sheet with columns: Date, Amount, Category, Merchant, Note.\n5. Set up your Telegram bot webhook to point to this workflow's Telegram Trigger URL.\n\nFree templates and updates: navkhan103.github.io", "height": 300, "width": 400, "color": 4 }, "id": "sticky-setup", "name": "Setup — read me first", "type": "n8n-nodes-base.stickyNote", "typeVersion": 1, "position": [ -440, -120 ] }, { "parameters": { "content": "### How it runs\n\nStarts from **Telegram Trigger**.\n\n1. A Telegram Trigger listens for messages sent to your bot.\n2. An OpenAI node extracts amount, category, and merchant from the message text.\n3. A Code node validates the extracted data and formats it for Sheets.\n4. A Google Sheets node appends the expense to your expense tracking spreadsheet.\n5. The bot sends a confirmation message back to the user.", "height": 300, "width": 400, "color": 7 }, "id": "sticky-how", "name": "How it runs", "type": "n8n-nodes-base.stickyNote", "typeVersion": 1, "position": [ -440, 220 ] }, { "parameters": { "content": "### If something breaks\n\nThe read and AI nodes here **retry 3× automatically** (5s apart), which absorbs rate limits and brief timeouts.\n\nNodes that **send or write deliberately do not retry** — a send that reached the far end but lost its response would otherwise fire twice, so you'd get duplicate messages or duplicate rows.\n\nTo be told when one of those fails, build a workflow starting with an **Error Trigger** node that posts to Slack or email, then set it under **three-dot menu > Settings > Error Workflow**. Do it once and point every workflow at it.", "height": 380, "width": 400, "color": 3 }, "id": "sticky-fail", "name": "If something breaks", "type": "n8n-nodes-base.stickyNote", "typeVersion": 1, "position": [ -440, 560 ] }, { "parameters": { "botToken": "REPLACE_WITH_BOT_TOKEN", "additionalFields": {} }, "id": "f9a0b1c2-0001-4000-8000-000000000001", "name": "Telegram Trigger", "type": "n8n-nodes-base.telegramTrigger", "typeVersion": 1.2, "position": [ 0, 0 ] }, { "parameters": { "modelId": { "__rl": true, "mode": "list", "value": "gpt-5.6-luna" }, "messages": { "values": [ { "content": "=Extract expense details from this message. Return only valid JSON with keys: amount (number), category (string), merchant (string), note (string). Example: {\"amount\": 25.5, \"category\": \"Food\", \"merchant\": \"Starbucks\", \"note\": \"Morning coffee\"}\n\nMessage: {{ $json.message.text }}", "role": "user" } ] } }, "id": "f9a0b1c2-0002-4000-8000-000000000002", "name": "Extract Expense", "type": "@n8n/n8n-nodes-langchain.openAi", "typeVersion": 1.8, "position": [ 220, 0 ], "retryOnFail": true, "maxTries": 3, "waitBetweenTries": 5000 }, { "parameters": { "jsCode": "// Parse the AI response and validate data\ntry {\n const aiResponse = $input.first().json.message.content;\n const expense = JSON.parse(aiResponse);\n \n // Validate required fields\n if (typeof expense.amount !== 'number' || isNaN(expense.amount)) {\n throw new Error('Invalid amount');\n }\n \n // Set default values if missing\n expense.category = expense.category || 'Uncategorized';\n expense.merchant = expense.merchant || 'Unknown';\n expense.note = expense.note || $('Telegram Trigger').first().json.message.text;\n \n // Format for Google Sheets\n const now = new Date();\n return [{ json: {\n date: now.toISOString().split('T')[0],\n amount: expense.amount,\n category: expense.category,\n merchant: expense.merchant,\n note: expense.note,\n raw_message: $('Telegram Trigger').first().json.message.text\n } }];\n} catch (error) {\n // If parsing fails, log the original message with error flag\n return [{ json: {\n date: new Date().toISOString().split('T')[0],\n amount: 0,\n category: 'ERROR',\n merchant: 'PARSE_FAILED',\n note: `Error: ${error.message}`,\n raw_message: $('Telegram Trigger').first().json.message.text\n } }];\n}" }, "id": "f9a0b1c2-0003-4000-8000-000000000003", "name": "Validate Data", "type": "n8n-nodes-base.code", "typeVersion": 2, "position": [ 440, 0 ] }, { "parameters": { "operation": "append", "documentId": { "__rl": true, "mode": "list", "value": "REPLACE_WITH_SPREADSHEET" }, "sheetName": "Expenses", "columns": "date,amount,category,merchant,note,raw_message" }, "id": "f9a0b1c2-0004-4000-8000-000000000004", "name": "Log to Sheets", "type": "n8n-nodes-base.googleSheets", "typeVersion": 4.5, "position": [ 660, 0 ] }, { "parameters": { "operation": "sendMessage", "chatId": "={{ $('Telegram Trigger').first().json.message.chat.id }}", "text": "={{ $json.amount ? `✅ Expense recorded: $${$json.amount} for ${$json.merchant} (${ $json.category })` : `❌ Could not parse expense. Please try format like: 'Lunch $25 at Restaurant'`}}", "additionalFields": {} }, "id": "f9a0b1c2-0005-4000-8000-000000000005", "name": "Confirm to User", "type": "n8n-nodes-base.telegram", "typeVersion": 1.2, "position": [ 880, 0 ] } ], "connections": { "Telegram Trigger": { "main": [ [ { "node": "Extract Expense", "type": "main", "index": 0 } ] ] }, "Extract Expense": { "main": [ [ { "node": "Validate Data", "type": "main", "index": 0 } ] ] }, "Validate Data": { "main": [ [ { "node": "Log to Sheets", "type": "main", "index": 0 } ], [ { "node": "Confirm to User", "type": "main", "index": 0 } ] ] } }, "settings": { "executionOrder": "v1" } }