Yusuke This workflow corresponds to n8n.io template #10539 — we link there as the canonical source.
This workflow follows the Agent → Google Sheets 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 →
{
"meta": {
"templateCredsSetupCompleted": true
},
"nodes": [
{
"id": "ac9638bd-e82e-473c-935d-77387f0827c6",
"name": "Test Input (Change Me)1",
"type": "n8n-nodes-base.set",
"position": [
832,
576
],
"parameters": {
"mode": "raw",
"options": {},
"jsonOutput": "={\n \"text\": \"I'm really frustrated with your service. This is the third time I've had issues and nobody is helping me!\"\n}"
},
"typeVersion": 3.4
},
{
"id": "3972b891-6ef7-4f31-915c-7b290932b36e",
"name": "Set Config1",
"type": "n8n-nodes-base.set",
"position": [
1056,
576
],
"parameters": {
"mode": "raw",
"options": {},
"jsonOutput": "={\n \"MAX_LEN\": 600,\n \"ADD_FOLLOWUP_QUESTION\": true,\n \"FORMALITY\": \"auto\",\n \"EMOJI_ALLOWED\": false,\n \"SIGNOFF\": \"\",\n \"BLOCK_LINKS\": true,\n \"RISK_WORDS\": [\n \"refund\",\"chargeback\",\"lawsuit\",\"harassment\",\"self-harm\",\"suicide\",\"abuse\",\n \"threat\",\"racist\",\"illegal\",\"hate\",\"scam\"\n ]\n}",
"includeOtherFields": true
},
"typeVersion": 3.4
},
{
"id": "546540d3-8889-4a5b-a06f-19322d9b386b",
"name": "Ensure Session1",
"type": "n8n-nodes-base.code",
"position": [
1280,
576
],
"parameters": {
"jsCode": "const j = $input.first().json || {};\nconst sid = j.sessionId || 'test-session-1';\nconst text = j.text || j.message || j.input || JSON.stringify(j);\nreturn [{ json: { ...j, sessionId: String(sid), text } }];"
},
"typeVersion": 2
},
{
"id": "9f4b03fa-aa22-4cca-a71b-b8d5fa505e45",
"name": "AI Agent (Empathy)1",
"type": "@n8n/n8n-nodes-langchain.agent",
"position": [
1568,
576
],
"parameters": {
"text": "=You are an Empathy Reply Assistant for English-language messages.\nAnalyze the incoming text:\n<Text>{{ $json.text || $json.message || $json.input || JSON.stringify($json) }}</Text>\n\nGoals:\n1) Detect `sentiment` (positive|neutral|negative|mixed)\n2) Choose a natural human `tone` that fits the sender (warm, calm, upbeat, apologetic, confident, concise)\n3) Draft a short `reply` (<= {{ $('Set Config1').item.json.MAX_LEN }} chars) that:\n - acknowledges the emotion first in plain English\n - provides a practical next step or clarification\n - {{ $('Set Config1').item.json.ADD_FOLLOWUP_QUESTION ? 'ends with a gentle question' : 'may end without a question' }}\n - uses **no links**, **no hashtags**\n - matches formality: {{ $('Set Config1').item.json.FORMALITY }} (auto|casual|polite)\n\nReturn **JSON only** with keys: sentiment, tone, reply, confidence (0.0\u20131.0), needs_handover (boolean).",
"options": {
"systemMessage": "Role: Empathetic, practical, concise. Avoid corporate cliches. Keep it in English. No URLs or hashtags."
},
"promptType": "define",
"hasOutputParser": true
},
"typeVersion": 1.6
},
{
"id": "a9afe8de-3810-4b2b-978e-a30f22f487f5",
"name": "Anthropic Chat Model1",
"type": "@n8n/n8n-nodes-langchain.lmChatAnthropic",
"position": [
1504,
800
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "claude-3-5-sonnet-latest",
"cachedResultName": "Claude 3.5 Sonnet (latest)"
},
"options": {}
},
"typeVersion": 1.3
},
{
"id": "1badd599-1d2a-484f-8219-3f03951c2ad4",
"name": "Memory (Recent 4)",
"type": "@n8n/n8n-nodes-langchain.memoryBufferWindow",
"position": [
1632,
800
],
"parameters": {
"contextWindowLength": 4
},
"typeVersion": 1.3
},
{
"id": "6799d4c5-da73-4415-9d26-c87adfe0d990",
"name": "Structured Output Parser",
"type": "@n8n/n8n-nodes-langchain.outputParserStructured",
"position": [
1760,
800
],
"parameters": {
"jsonSchemaExample": "{\n \"sentiment\": \"positive|neutral|negative|mixed\",\n \"tone\": \"warm|calm|upbeat|apologetic|confident|concise\",\n \"reply\": \"string\",\n \"confidence\": 0.0,\n \"needs_handover\": false\n}"
},
"typeVersion": 1.2
},
{
"id": "61a412c9-801c-4d5d-822a-8499a26d2607",
"name": "Post-Process & Sanitize1",
"type": "n8n-nodes-base.code",
"position": [
1968,
576
],
"parameters": {
"jsCode": "// Read config from Set Config node (stable across items)\nconst cfgNode = $nodes['Set Config1']?.data?.json || {};\nconst cfg = {\n MAX_LEN: Number(cfgNode.MAX_LEN ?? 600),\n EMOJI_ALLOWED: cfgNode.EMOJI_ALLOWED === false ? false : Boolean(cfgNode.EMOJI_ALLOWED),\n SIGNOFF: String(cfgNode.SIGNOFF ?? ''),\n BLOCK_LINKS: cfgNode.BLOCK_LINKS === undefined ? true : Boolean(cfgNode.BLOCK_LINKS)\n};\n\nconst input = $input.first().json || {};\nlet out = { ...input };\n\nout.sentiment = (out.sentiment || '').toLowerCase();\nout.tone = (out.tone || '').toLowerCase();\nout.reply = (out.reply || '').toString();\nout.confidence = (typeof out.confidence === 'number') ? out.confidence : 0.6;\nout.needs_handover = !!out.needs_handover;\n\nif (cfg.BLOCK_LINKS) {\n out.reply = out.reply\n .replace(/https?:\\/\\/\\S+/gi, '')\n .replace(/#[\\w-]+/g, '')\n .replace(/\\s{2,}/g, ' ');\n}\n\nif (cfg.EMOJI_ALLOWED === false) {\n out.reply = out.reply.replace(/[\\p{Extended_Pictographic}\\p{Emoji_Component}]/gu, '');\n}\n\nout.reply = out.reply\n .replace(/[A-Z0-9._%+-]+@[A-Z0-9.-]+\\.[A-Z]{2,}/ig, '[email hidden]')\n .replace(/\\b\\+?\\d[\\d\\s().-]{6,}\\b/g, '[phone hidden]');\n\nconst maxLen = Number(cfg.MAX_LEN || 600);\nif (out.reply.length > maxLen) {\n out.reply = out.reply.slice(0, maxLen - 1).replace(/\\s+\\S*$/, '') + '\u2026';\n}\n\nif (cfg.SIGNOFF && !/\\b(best|thanks|regards)\\b/i.test(out.reply)) {\n out.reply = out.reply.trim() + \"\\n\" + cfg.SIGNOFF;\n}\n\nreturn [{ json: out }];"
},
"typeVersion": 2
},
{
"id": "c255267c-5dfb-47d9-9c9a-9a7ae4ada264",
"name": "Risk & Handover Rules1",
"type": "n8n-nodes-base.code",
"position": [
2192,
576
],
"parameters": {
"jsCode": "// Read RISK_WORDS from Set Config node\nconst cfgNode = $nodes['Set Config']?.data?.json || {};\nconst riskWords = Array.isArray(cfgNode.RISK_WORDS) ? cfgNode.RISK_WORDS : [];\n\nconst escapeRegex = (s) => String(s).replace(/[.*+?^${}()|[\\]\\\\]/g, '\\\\$&');\nconst riskPattern = riskWords.length ? '(' + riskWords.map(escapeRegex).join('|') + ')' : '(?!)';\nconst risk = new RegExp(riskPattern, 'i');\n\nconst j = $input.first().json || {};\nconst originalText = j.text || j.message || j.input || '';\n\nconst lowConfidence = (j.confidence || 0) < 0.45;\nconst highRisk = risk.test(originalText) || risk.test(j.reply || '');\nconst veryNegative = (j.sentiment || '') === 'negative';\n\nconst needsHandover = Boolean(j.needs_handover || lowConfidence || highRisk || veryNegative);\n\nreturn [{ json: { ...j, needs_handover: needsHandover, _debug: { lowConfidence, highRisk, veryNegative, originalText: String(originalText).substring(0,120), riskWordsUsed: riskWords } } }];"
},
"typeVersion": 2
},
{
"id": "7aae0aff-287e-4c5c-b6d4-c2e5b42e2efb",
"name": "If needs_handover1",
"type": "n8n-nodes-base.if",
"position": [
2416,
576
],
"parameters": {
"options": {},
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "strict"
},
"combinator": "and",
"conditions": [
{
"id": "e2876b8a-248f-41a9-95f2-cbb235ad7071",
"operator": {
"type": "boolean",
"operation": "true"
},
"leftValue": "={{ $json.needs_handover }}"
}
]
}
},
"typeVersion": 2.2
},
{
"id": "42085b59-d2d1-443b-9699-dbbe3ce86acf",
"name": "Draft (Needs Review)",
"type": "n8n-nodes-base.set",
"position": [
2640,
480
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "a1",
"name": "status",
"type": "string",
"value": "REVIEW"
},
{
"id": "a2",
"name": "note",
"type": "string",
"value": "Escalate to human: risk/low-confidence/negative tone detected."
}
]
},
"includeOtherFields": true
},
"typeVersion": 3.4
},
{
"id": "529f1815-7913-424d-8908-ff282ceabcdd",
"name": "Draft (Auto-OK)",
"type": "n8n-nodes-base.set",
"position": [
2640,
672
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "b1",
"name": "status",
"type": "string",
"value": "OK"
}
]
},
"includeOtherFields": true
},
"typeVersion": 3.4
},
{
"id": "9517622f-10ae-4597-81e6-4b426d266e4c",
"name": "for real-time replies",
"type": "n8n-nodes-base.webhook",
"position": [
608,
672
],
"parameters": {
"path": "7a62325a-0000-4724-8fe4-8829e3dea2fb",
"options": {}
},
"typeVersion": 2.1
},
{
"id": "50a47075-c245-46f5-a103-9c240c03f519",
"name": "(5\u201310 min) for missed or queued",
"type": "n8n-nodes-base.scheduleTrigger",
"position": [
608,
848
],
"parameters": {
"rule": {
"interval": [
{}
]
}
},
"typeVersion": 1.2
},
{
"id": "66b91934-ac03-4d78-a816-052aebae5d8a",
"name": "Update a message",
"type": "n8n-nodes-base.slack",
"position": [
2864,
688
],
"parameters": {
"channelId": {
"__rl": true,
"mode": "list",
"value": ""
},
"operation": "update",
"otherOptions": {},
"updateFields": {}
},
"typeVersion": 2.3
},
{
"id": "0dc9f0d7-38ff-4cf0-919e-c6d87b93168d",
"name": "Save to Google Sheets1",
"type": "n8n-nodes-base.googleSheets",
"position": [
2848,
464
],
"parameters": {
"columns": {
"value": {
"Nodes": "={{ $json.nodes }}",
"Stars": "={{ $json.stars }}",
"File URL": "={{ $json.file_url }}",
"Repo URL": "={{ $json.repo_url }}",
"Use Case": "={{ $json.ai_use_case }}",
"Workflow": "={{ $json.workflow_name }}",
"Key Nodes": "={{ $json.key_nodes }}",
"AI Powered": "={{ $json.has_ai }}",
"AI Summary": "={{ $json.description }}",
"Difficulty": "={{ $json.ai_difficulty }}",
"Node Types": "={{ $json.node_types }}",
"Repository": "={{ $json.repository }}"
},
"mappingMode": "defineBelow",
"matchingColumns": [
"Workflow"
],
"attemptToConvertTypes": false,
"convertFieldsToString": false
},
"options": {},
"operation": "appendOrUpdate",
"sheetName": "={{ $('Config1').item.json.SHEET_NAME }}",
"documentId": "={{ $('Config1').item.json.SPREADSHEET_ID }}"
},
"credentials": {
"googleSheetsOAuth2Api": {
"name": "<your credential>"
}
},
"typeVersion": 4.5
},
{
"id": "96ce7708-d101-4cb8-a242-1ca6e8727fd3",
"name": "for testing",
"type": "n8n-nodes-base.manualTrigger",
"position": [
608,
480
],
"parameters": {},
"typeVersion": 1
},
{
"id": "3f86e343-aec8-4bd3-8f8d-ad09a1aae6e5",
"name": "Sticky Note",
"type": "n8n-nodes-base.stickyNote",
"position": [
144,
160
],
"parameters": {
"width": 384,
"height": 368,
"content": "## Empathy Reply Assistant\n**What it does** \nGenerates empathetic reply drafts, sanitizes content, and auto-escalates risky cases.\n\n**Flow** \n1) Input (Manual / Webhook) \n2) AI (Claude) \u2192 { sentiment, tone, reply, confidence } \n3) Sanitize (links / PII / length) \n4) Risk rules (keywords, negativity, low confidence) \n5) Route \u2192 **OK** / **Needs Review**\n\n**Who for** \nSupport, Community, Sales, CX, Automation builders"
},
"typeVersion": 1
},
{
"id": "ab112a1e-bf36-4ee6-b8e5-9b0a500c4a5e",
"name": "Sticky Note1",
"type": "n8n-nodes-base.stickyNote",
"position": [
144,
608
],
"parameters": {
"width": 384,
"height": 352,
"content": "## Routes & Thresholds\n**Escalate (Needs Review)** when: \n- negative sentiment \n- matches `RISK_WORDS` \n- confidence < 0.45 \n- mentions legal / harassment / self-harm \n\n**Auto-OK** when: \n- no risk matched \n- confidence \u2265 0.45 \n\n**Outputs** \n- OK \u2192 sendable draft \n- REVIEW \u2192 human check required"
},
"typeVersion": 1
},
{
"id": "fceb4bb7-0e2b-460e-901f-8dcb415fdb7e",
"name": "Sticky Note2",
"type": "n8n-nodes-base.stickyNote",
"position": [
832,
160
],
"parameters": {
"color": 7,
"width": 384,
"height": 304,
"content": "## Setup (3\u20135 min)\n1. Connect **Anthropic** \u2192 Anthropic Chat Model \n2. Edit **Set Config1**: \n - `MAX_LEN`, `FORMALITY`, `ADD_FOLLOWUP_QUESTION` \n - `EMOJI_ALLOWED`, `BLOCK_LINKS` \n - `RISK_WORDS` (add your terms) \n3. Choose trigger: \n - **Manual** \u2192 testing \n - **Webhook** \u2192 real-time replies \n - **Schedule** \u2192 every 5\u201310 min \n\n\ud83d\udca1 Tip: keep secrets in **Credentials**, not nodes."
},
"typeVersion": 1
},
{
"id": "b789e28c-1349-46fd-bbcd-161102405afa",
"name": "Sticky Note3",
"type": "n8n-nodes-base.stickyNote",
"position": [
832,
768
],
"parameters": {
"color": 7,
"width": 384,
"height": 224,
"content": "## Notes / Safety\n- \ud83d\udd10 No hardcoded API keys \u2192 use **Credentials** \n- \ud83e\uddea Treat outputs as **drafts** before sending \n- \ud83e\uddf9 Remove test data before sharing publicly \n- \u2699\ufe0f Tune thresholds & `RISK_WORDS` to your org policy \n- \ud83d\uddc2\ufe0f Optionally log to Slack / Sheets for review \n\nAuthor: **Yusuke (@yskautomation)** \nLicense: **MIT**"
},
"typeVersion": 1
}
],
"connections": {
"Set Config1": {
"main": [
[
{
"node": "Ensure Session1",
"type": "main",
"index": 0
}
]
]
},
"Ensure Session1": {
"main": [
[
{
"node": "AI Agent (Empathy)1",
"type": "main",
"index": 0
}
]
]
},
"Memory (Recent 4)": {
"ai_memory": [
[
{
"node": "AI Agent (Empathy)1",
"type": "ai_memory",
"index": 0
}
]
]
},
"If needs_handover1": {
"main": [
[
{
"node": "Draft (Needs Review)",
"type": "main",
"index": 0
}
],
[
{
"node": "Draft (Auto-OK)",
"type": "main",
"index": 0
}
]
]
},
"AI Agent (Empathy)1": {
"main": [
[
{
"node": "Post-Process & Sanitize1",
"type": "main",
"index": 0
}
]
]
},
"Anthropic Chat Model1": {
"ai_languageModel": [
[
{
"node": "AI Agent (Empathy)1",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"for real-time replies": {
"main": [
[
{
"node": "Test Input (Change Me)1",
"type": "main",
"index": 0
}
]
]
},
"Risk & Handover Rules1": {
"main": [
[
{
"node": "If needs_handover1",
"type": "main",
"index": 0
}
]
]
},
"Test Input (Change Me)1": {
"main": [
[
{
"node": "Set Config1",
"type": "main",
"index": 0
}
]
]
},
"Post-Process & Sanitize1": {
"main": [
[
{
"node": "Risk & Handover Rules1",
"type": "main",
"index": 0
}
]
]
},
"Structured Output Parser": {
"ai_outputParser": [
[
{
"node": "AI Agent (Empathy)1",
"type": "ai_outputParser",
"index": 0
}
]
]
},
"(5\u201310 min) for missed or queued": {
"main": [
[]
]
}
}
}
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.
googleSheetsOAuth2Api
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
About this workflow
Generate empathetic, professional reply drafts for customer or user messages. The workflow detects sentiment, tone, and risk level, drafts a concise response, sanitizes PII/links/emojis, and auto-escalates risky or low-confidence cases to human review. Input — Manual Test or…
Source: https://n8n.io/workflows/10539/ — 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.
Enhance your support, onboarding, and internal knowledge workflows with an intelligent RAG-powered chatbot that responds using live data stored in Google Sheets. 🤖📚 Built for teams that rely on struct
Chat Trigger, Output Parser Structured, Memory Buffer Window +6
Fully automates your service order pipeline from incoming booking to supplier confirmation — with built-in SLA enforcement and automatic escalation if a supplier goes silent. 📥 Receives orders via web
HTTP Request, Google Sheets, Agent +4
This workflow automates CSV data processing from upload to database insertion.
Agent, Anthropic Chat, Output Parser Structured +3
This workflow automates document understanding by accepting uploaded PDF or TXT files, extracting their text, generating a structured summary and question–answer set using GPT-4o, validating the AI ou
Agent, Output Parser Structured, Lm Chat Azure Open Ai +3
⏺ 🚀 How it works
Agent, Anthropic Chat, Output Parser Structured +6