n8n AI Agent Workflow for Local Business Lead Follow-Up

n8n vs. Make.com: When to Choose Which

Setting Up n8n

Option 1: n8n Cloud (Easiest Start)

Option 2: Self-Hosted on a VPS (Free)

# Install Docker
curl -fsSL https://get.docker.com | sh

# Run n8n with Docker Compose
cat > docker-compose.yml << 'EOF'
version: '3'
services:
  n8n:
    image: docker.n8nio/n8n
    restart: unless-stopped
    ports:
      - "5678:5678"
    environment:
      - N8N_HOST=your-domain.com
      - N8N_PORT=5678
      - N8N_PROTOCOL=https
      - WEBHOOK_URL=https://your-domain.com/
      - GENERIC_TIMEZONE=America/Chicago
    volumes:
      - n8n_data:/home/node/.n8n

volumes:
  n8n_data:
EOF

docker compose up -d

The Workflow: AI Lead Follow-Up Agent

1. Watches your Gmail inbox for new inquiry emails (use a dedicated inquiries address if you can)
2. Uses an LLM to classify lead quality (GOOD_FIT / MAYBE / NOT_FIT)
3. Generates a personalized reply for GOOD_FIT leads
4. Sends the email reply (or, safer at first, saves it as a draft for you to approve)
5. Logs everything to a Google Sheet
6. Sends you a Slack/email alert for high-priority leads only

Workflow JSON (Import Directly into n8n)

{
  "name": "AI Lead Follow-Up Agent",
  "nodes": [
    {
      "parameters": {
        "pollTimes": {
          "item": [{"mode": "everyMinute"}]
        },
        "filters": {
          "readStatus": "unread",
          "sender": ""
        }
      },
      "name": "Gmail Trigger",
      "type": "n8n-nodes-base.gmailTrigger",
      "typeVersion": 1,
      "position": [250, 300]
    },
    {
      "parameters": {
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "openAiApi",
        "resource": "chat",
        "model": "gpt-4o-mini",
        "messages": {
          "values": [
            {
              "role": "system",
              "content": "You are a lead qualifier. Classify the email as GOOD_FIT, MAYBE, or NOT_FIT. Reply only with valid JSON: {\"classification\": \"GOOD_FIT\", \"reason\": \"one sentence\", \"priority\": \"high|medium|low\", \"first_name\": \"extracted name or Unknown\"}"
            },
            {
              "role": "user",
              "content": "=Classify this lead inquiry:\n\nFrom: {{ $json.from }}\nSubject: {{ $json.subject }}\nBody: {{ $json.text }}\n\nGOOD_FIT: local business, clear service need, budget signals, professional inquiry\nNOT_FIT: requests free work, spam, wrong service entirely, student project\nMAYBE: everything else"
            }
          ]
        },
        "options": {
          "maxTokens": 150,
          "temperature": 0.2
        }
      },
      "name": "Classify Lead",
      "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
      "typeVersion": 1,
      "position": [500, 300]
    },
    {
      "parameters": {
        "jsCode": "const raw = $input.first().json.choices[0].message.content;\ntry {\n  return [{ json: JSON.parse(raw) }];\n} catch(e) {\n  return [{ json: { classification: 'MAYBE', reason: 'parse error', priority: 'low', first_name: 'there' } }];\n}"
      },
      "name": "Parse Classification",
      "type": "n8n-nodes-base.code",
      "typeVersion": 2,
      "position": [750, 300]
    },
    {
      "parameters": {
        "conditions": {
          "string": [
            {
              "value1": "={{ $json.classification }}",
              "value2": "GOOD_FIT"
            }
          ]
        }
      },
      "name": "Is Good Fit?",
      "type": "n8n-nodes-base.if",
      "typeVersion": 1,
      "position": [1000, 300]
    },
    {
      "parameters": {
        "authentication": "predefinedCredentialType",
        "nodeCredentialType": "openAiApi",
        "resource": "chat",
        "model": "gpt-4o-mini",
        "messages": {
          "values": [
            {
              "role": "system",
              "content": "You write warm, personalized business emails. Under 180 words. Sound like a real person, not a marketing email. Sign as Nataliia from DataLatte."
            },
            {
              "role": "user",
              "content": "=Write a reply to this inquiry.\n\nFrom: {{ $('Gmail Trigger').first().json.from }}\nSubject: {{ $('Gmail Trigger').first().json.subject }}\nInquiry: {{ $('Gmail Trigger').first().json.text }}\n\nInclude:\n- Reference something specific from their message\n- Suggest a free 20-min discovery call\n- Calendly link: https://calendly.com/datalatte/discovery\n- Warm and specific, not generic"
            }
          ]
        },
        "options": {
          "maxTokens": 350,
          "temperature": 0.7
        }
      },
      "name": "Generate Reply",
      "type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
      "typeVersion": 1,
      "position": [1250, 200]
    },
    {
      "parameters": {
        "operation": "reply",
        "messageId": "={{ $('Gmail Trigger').first().json.id }}",
        "message": "={{ $json.choices[0].message.content }}"
      },
      "name": "Send Reply",
      "type": "n8n-nodes-base.gmail",
      "typeVersion": 2,
      "position": [1500, 200]
    },
    {
      "parameters": {
        "operation": "append",
        "documentId": "YOUR_GOOGLE_SHEET_ID",
        "sheetName": "Leads",
        "columns": {
          "mappingMode": "defineBelow",
          "value": {
            "timestamp": "={{ new Date().toISOString() }}",
            "from": "={{ $('Gmail Trigger').first().json.from }}",
            "subject": "={{ $('Gmail Trigger').first().json.subject }}",
            "classification": "={{ $('Parse Classification').first().json.classification }}",
            "priority": "={{ $('Parse Classification').first().json.priority }}",
            "reason": "={{ $('Parse Classification').first().json.reason }}",
            "replied": "=YES"
          }
        }
      },
      "name": "Log to Sheets",
      "type": "n8n-nodes-base.googleSheets",
      "typeVersion": 4,
      "position": [1750, 200]
    }
  ],
  "connections": {
    "Gmail Trigger": {
      "main": [[{"node": "Classify Lead", "type": "main", "index": 0}]]
    },
    "Classify Lead": {
      "main": [[{"node": "Parse Classification", "type": "main", "index": 0}]]
    },
    "Parse Classification": {
      "main": [[{"node": "Is Good Fit?", "type": "main", "index": 0}]]
    },
    "Is Good Fit?": {
      "main": [
        [{"node": "Generate Reply", "type": "main", "index": 0}],
        [{"node": "Log to Sheets", "type": "main", "index": 0}]
      ]
    },
    "Generate Reply": {
      "main": [[{"node": "Send Reply", "type": "main", "index": 0}]]
    },
    "Send Reply": {
      "main": [[{"node": "Log to Sheets", "type": "main", "index": 0}]]
    }
  }
}

Using n8n's Native AI Agent Node

• Maintains conversation context
• Can call multiple tools in sequence
• Supports function calling (takes actions based on LLM decisions)
• Works with many LLM providers, including OpenAI-compatible APIs such as Groq

Trigger (Gmail) 
    → AI Agent node
        - System prompt: "You are a business development assistant. When given a new email lead, classify it and draft a reply."
        - Tools: [Gmail Reply, Google Sheets Append, Slack Notify]
        - LLM: Groq (llama-3.3-70b) or OpenAI gpt-4o-mini
    → Done

Connecting to Groq (Low-Cost LLM Option)

1. In n8n, go to Credentials → New → OpenAI API
2. Set API key: your Groq API key
3. Set base URL: https://api.groq.com/openai/v1
4. In any LLM node, set model to: llama-3.3-70b-versatile

Safety Checklist Before Going Live

• Start with draft mode or send replies only to yourself for the first week.
• Never let the agent promise prices, availability or guarantees you have not approved. Put those rules in the system prompt.
• Route complaints, refunds and sensitive messages to a human.
• Keep API keys in n8n credentials, not in the workflow text.
• Tell customers they may be talking to an automated assistant where the law or good practice requires it.

FAQ

Next Step

Related Articles

• Make.com AI Agent Workflow for Local Business: Complete Blueprint
• AI Appointment Reminder Agent: Python Script That Cuts No-Shows
• Make.com vs Zapier for Local Business: Which Platform Is Better?
• AI Agent for Lead Generation: How to Automate Your Follow-Ups
• Custom GPT for Local Business: Real System Prompts That Work