Build an AI-Powered Cold Email Outreach Agent with n8n (Free Template)

Cold email outreach at scale requires personalization that actually converts. Generic templates get ignored. Manual research takes hours per prospect. This n8n agent solves both problems by combining AI-powered prospect research with automated, personalized email sequences that feel human-written. You'll learn how to build a complete cold outreach system that researches prospects, generates custom emails, and manages follow-up sequences automatically.

The Problem: Cold Email Outreach That Doesn't Scale

Current challenges:

Manual prospect research takes 15-30 minutes per lead
Generic email templates achieve <2% response rates
Follow-up sequences require constant manual tracking
Personalization at scale is nearly impossible without a large team
CRM data entry and campaign management consume hours daily

Business impact:

Time spent: 20+ hours/week on outreach for 100 prospects
Cost: $2,000-4,000/month for VA or SDR time
Conversion loss: 60-80% lower response rates with generic emails
Missed opportunities: 70% of prospects never receive follow-ups

Manual cold email processes create a bottleneck between lead generation and actual conversations. Sales teams either sacrifice personalization for volume or spend unsustainable time researching each prospect.

The Solution Overview

This n8n agent automates the entire cold email workflow from prospect intake to follow-up sequences. It uses AI to research each prospect's company and role, generates personalized email copy based on custom templates, and manages multi-touch follow-up campaigns automatically. The system integrates with your CRM, email provider, and AI research tools to create a hands-off outreach machine that maintains the personal touch of manual research. Built entirely in n8n, this workflow handles everything from data enrichment to delivery tracking without requiring custom code.

What You'll Build

This cold email agent delivers complete outreach automation with AI-powered personalization:

Component	Technology	Purpose
Prospect Intake	Google Sheets/Airtable	Centralized lead list management
AI Research	OpenAI GPT-4	Company and prospect background analysis
Email Generation	Custom Prompts + GPT-4	Personalized email copy creation
Email Delivery	Gmail/SendGrid	Automated sending with tracking
Follow-up Logic	n8n Schedule + Function Nodes	Multi-touch sequence management
CRM Sync	HubSpot/Salesforce	Contact and activity logging
Response Detection	Email Webhook	Automatic reply monitoring

Key capabilities:

Automatic prospect research using AI web scraping
Custom email generation based on industry and role
Multi-step follow-up sequences (3-5 touches)
Response tracking and campaign analytics
CRM integration for seamless data flow
Personalization variables (company name, role, pain points)
A/B testing support for email variants

Prerequisites

Before starting, ensure you have:

n8n instance (cloud or self-hosted version 1.0+)
OpenAI API account with GPT-4 access
Gmail account with App Password or SendGrid API key
Google Sheets or Airtable for prospect management
CRM account (HubSpot, Salesforce, or Pipedrive)
Basic understanding of email deliverability best practices
Domain with proper SPF/DKIM/DMARC records configured

Step 1: Set Up Prospect Data Source

Configure your prospect intake system to feed leads into the automation workflow.

Create the prospect spreadsheet:

Set up Google Sheets with these required columns: Email, First Name, Last Name, Company, Title, LinkedIn URL, Status, Last Contact Date
Add optional enrichment columns: Company Size, Industry, Pain Points, Recent News
Create a "Status" column with values: New, Researched, Email Sent, Follow-up 1, Follow-up 2, Responded, Unsubscribed

Node configuration:

{
  "name": "Google Sheets",
  "type": "n8n-nodes-base.googleSheets",
  "parameters": {
    "operation": "read",
    "sheetId": "YOUR_SHEET_ID",
    "range": "A:H",
    "options": {
      "headerRow": 1
    }
  }
}

Why this works:

Google Sheets provides a simple interface for sales teams to add prospects while giving n8n structured data to process. The status column enables the workflow to track each prospect's position in the outreach sequence and prevent duplicate emails. Using a spreadsheet instead of direct CRM integration allows for easy manual review and editing before automation kicks in.

Step 2: AI-Powered Prospect Research

Build the research engine that gathers personalized context about each prospect.

Configure the research workflow:

Add an HTTP Request node to scrape the company website using Apify or Bright Data
Connect an OpenAI node with a custom research prompt
Set up a Function node to extract and structure key insights
Add error handling for prospects with limited online presence

Research prompt configuration:

// Function node to prepare research prompt
const companyName = $input.item.json.Company;
const prospectTitle = $input.item.json.Title;
const linkedinUrl = $input.item.json['LinkedIn URL'];

const researchPrompt = `Research this prospect and company:
Company: ${companyName}
Prospect Title: ${prospectTitle}
LinkedIn: ${linkedinUrl}

Provide:
1. Company's main product/service (1 sentence)
2. Recent company news or achievements (last 3 months)
3. Likely pain points for someone in the ${prospectTitle} role
4. Relevant industry trends affecting this company
5. Personalization angle for cold outreach

Format as JSON with keys: product, news, painPoints, trends, angle`;

return {
  json: {
    prompt: researchPrompt,
    prospectEmail: $input.item.json.Email
  }
};

OpenAI node settings:

{
  "model": "gpt-4",
  "temperature": 0.7,
  "maxTokens": 500,
  "messages": [
    {
      "role": "system",
      "content": "You are a B2B sales research assistant. Provide concise, actionable insights for cold email personalization."
    },
    {
      "role": "user",
      "content": "={{ $json.prompt }}"
    }
  ]
}

Why this approach:

AI research replaces 15-30 minutes of manual work per prospect. The structured prompt ensures consistent output format that downstream nodes can parse reliably. Using GPT-4 instead of GPT-3.5 improves research quality significantly—it better understands company contexts and identifies relevant pain points. The JSON output format makes it easy to inject research findings directly into email templates.

Step 3: Generate Personalized Email Copy

Create the email generation system that turns research into compelling outreach messages.

Build the email template system:

Create a Function node with your base email template structure
Add dynamic variables for research insights
Connect to OpenAI for final copy polish
Include A/B testing logic for subject lines

Email generation function:

// Extract research data
const research = JSON.parse($input.item.json.researchOutput);
const firstName = $input.item.json['First Name'];
const company = $input.item.json.Company;

// Base template with personalization
const emailTemplate = `Subject: Quick question about ${company}'s ${research.product}

Hi ${firstName},

I noticed ${research.news} - congratulations on that milestone.

Given your role as ${$input.item.json.Title}, I imagine ${research.painPoints} might be on your radar.

We've helped similar companies in ${research.trends} by [YOUR VALUE PROP]. Specifically, [SPECIFIC BENEFIT].

Would it make sense to explore how this could work for ${company}? I have a 15-minute slot [DAY] at [TIME] if you're open to it.

Best,
[YOUR NAME]

P.S. ${research.angle}`;

return {
  json: {
    emailBody: emailTemplate,
    prospectEmail: $input.item.json.Email,
    firstName: firstName,
    company: company
  }
};

Polish with AI:

{
  "name": "Polish Email Copy",
  "type": "n8n-nodes-base.openAi",
  "parameters": {
    "model": "gpt-4",
    "temperature": 0.8,
    "messages": [
      {
        "role": "system",
        "content": "You are an expert cold email copywriter. Make this email more conversational and compelling while keeping it under 150 words. Maintain all personalization elements."
      },
      {
        "role": "user",
        "content": "={{ $json.emailBody }}"
      }
    ]
  }
}

Why this works:

The two-stage approach (template + AI polish) balances consistency with natural language. Your template ensures key messaging stays on-brand while AI polish removes robotic phrasing. The research variables inject genuine personalization that recipients can verify—not generic flattery. Keeping emails under 150 words dramatically improves mobile readability and response rates.

Step 4: Email Delivery and Tracking

Configure the sending infrastructure with proper deliverability and tracking.

Set up Gmail sending:

{
  "name": "Send Email",
  "type": "n8n-nodes-base.gmail",
  "parameters": {
    "operation": "send",
    "to": "={{ $json.prospectEmail }}",
    "subject": "={{ $json.emailSubject }}",
    "message": "={{ $json.polishedEmail }}",
    "options": {
      "sendAsHtml": false,
      "ccList": "your-tracking-email@company.com"
    }
  }
}

Add tracking pixel:

// Function node to add tracking
const trackingId = `${$json.prospectEmail}-${Date.now()}`;
const emailWithTracking = $json.polishedEmail + `

<img src="https://your-tracking-domain.com/pixel/${trackingId}" width="1" height="1">`;

return {
  json: {
    finalEmail: emailWithTracking,
    trackingId: trackingId,
    prospectEmail: $json.prospectEmail
  }
};

Update CRM and spreadsheet:

{
  "name": "Update Status",
  "type": "n8n-nodes-base.googleSheets",
  "parameters": {
    "operation": "update",
    "sheetId": "YOUR_SHEET_ID",
    "range": "G{{ $json.rowNumber }}",
    "options": {
      "valueInputMode": "USER_ENTERED"
    },
    "dataToSend": "defineInNode",
    "fieldsUi": {
      "values": [
        {
          "fieldId": "Status",
          "fieldValue": "Email Sent"
        },
        {
          "fieldId": "Last Contact Date",
          "fieldValue": "={{ $now.toISO() }}"
        }
      ]
    }
  }
}

Critical configuration:

Sending limits: Cap at 50 emails/day per Gmail account to avoid spam flags
Delay between sends: Add 2-5 minute random delays using Wait node
Reply-to address: Use a monitored inbox, not a no-reply address
Unsubscribe link: Include in email footer for compliance

Step 5: Automated Follow-Up Sequences

Build the multi-touch follow-up system that increases response rates by 3-5x.

Schedule node configuration:

{
  "name": "Daily Follow-up Check",
  "type": "n8n-nodes-base.scheduleTrigger",
  "parameters": {
    "rule": {
      "interval": [
        {
          "field": "hours",
          "hoursInterval": 24
        }
      ]
    },
    "triggerTimes": {
      "item": [
        {
          "hour": 9,
          "minute": 0
        }
      ]
    }
  }
}

Follow-up logic:

// Function node to determine follow-up stage
const lastContactDate = new Date($json['Last Contact Date']);
const daysSinceContact = Math.floor((Date.now() - lastContactDate) / (1000 * 60 * 60 * 24));
const currentStatus = $json.Status;

let followUpAction = null;

if (currentStatus === 'Email Sent' && daysSinceContact >= 3) {
  followUpAction = 'Follow-up 1';
} else if (currentStatus === 'Follow-up 1' && daysSinceContact >= 5) {
  followUpAction = 'Follow-up 2';
} else if (currentStatus === 'Follow-up 2' && daysSinceContact >= 7) {
  followUpAction = 'Follow-up 3';
}

if (followUpAction) {
  return {
    json: {
      ...($json),
      nextAction: followUpAction,
      sendFollowUp: true
    }
  };
}

return { json: { sendFollowUp: false } };

Follow-up email templates:

const followUpTemplates = {
  'Follow-up 1': `Hi ${firstName},

Following up on my email from ${daysSinceContact} days ago about ${company}'s ${research.product}.

I know inboxes get busy. If this isn't a priority right now, just let me know and I'll check back in a few months.

Still interested in that 15-minute chat?

Best,
[YOUR NAME]`,

  'Follow-up 2': `${firstName},

Last one from me - I promise!

I've attached a quick case study showing how [SIMILAR COMPANY] solved [PAIN POINT] in just 6 weeks.

Worth 15 minutes to see if this applies to ${company}?

[YOUR NAME]`,

  'Follow-up 3': `Hi ${firstName},

I'll take your silence as a "not right now" and close my loop.

If anything changes, here's my calendar link: [LINK]

Best of luck with ${company}'s growth!

[YOUR NAME]`
};

Why this approach:

The 3-5-7 day cadence balances persistence with respect. Each follow-up adds new value (case study, different angle) rather than repeating the same ask. The final "breakup" email often triggers responses from prospects who were interested but busy. Automated follow-ups recover 40-60% of responses that would otherwise be lost.

Workflow Architecture Overview

This workflow consists of 24 nodes organized into 5 main sections:

Data ingestion (Nodes 1-4): Reads prospect list from Google Sheets, filters for "New" status, validates email format, and prepares data structure
Research phase (Nodes 5-10): Scrapes company websites, calls OpenAI for prospect research, parses JSON responses, handles API errors, and stores research data
Email generation (Nodes 11-16): Applies email templates, injects personalization variables, polishes copy with AI, generates subject line variants, and adds tracking pixels
Delivery system (Nodes 17-20): Sends emails via Gmail, updates CRM records, logs activity in Google Sheets, and implements sending delays
Follow-up automation (Nodes 21-24): Runs daily schedule checks, identifies prospects needing follow-ups, generates follow-up emails, and manages sequence progression

Execution flow:

Trigger: Manual execution for initial send, scheduled daily for follow-ups
Average run time: 45-60 seconds per prospect (including AI calls)
Key dependencies: OpenAI API, Gmail API, Google Sheets API

Critical nodes:

OpenAI Research Node: Handles prospect and company research with GPT-4
Email Polish Node: Refines copy for natural language and tone
Follow-up Logic Function: Determines next action based on days since last contact
Status Update Node: Prevents duplicate sends by tracking campaign progress

The complete n8n workflow JSON template is available at the bottom of this article.

Key Configuration Details

OpenAI API Settings

Required fields:

API Key: Your OpenAI API key from platform.openai.com
Model: gpt-4 (not gpt-3.5-turbo for research quality)
Temperature: 0.7 for research, 0.8 for email polish
Max Tokens: 500 for research, 300 for email generation

Common issues:

Using gpt-3.5-turbo → Results in generic, low-quality research
Temperature above 0.9 → Creates inconsistent, sometimes inappropriate copy
Insufficient max tokens → Truncates research output mid-sentence

Gmail Integration

Authentication setup:

Enable 2-factor authentication on your Google account
Generate App Password at myaccount.google.com/apppasswords
Use App Password (not regular password) in n8n credentials
Set sending limit to 50 emails/day maximum

Deliverability configuration:

Warm up new sending accounts: Start with 10 emails/day, increase by 10 every 3 days
SPF record: Add v=spf1 include:_spf.google.com ~all to DNS
DMARC policy: Set to p=none initially, monitor for issues
Custom tracking domain: Use subdomain like track.yourdomain.com for pixel tracking

Variables to customize:

followUpDelay: Change from 3-5-7 to 2-4-6 for faster cadence or 4-7-10 for slower
maxFollowUps: Adjust from 3 to 2 (less aggressive) or 5 (more persistent)
emailTemplate: Modify base template to match your value proposition and tone
researchPrompt: Add industry-specific research questions for better personalization

Testing & Validation

Test the research phase:

Add a single test prospect to your Google Sheets with complete data
Run the workflow manually and examine the OpenAI research output
Verify JSON parsing works correctly—check for malformed responses
Review research quality: Does it identify real pain points and recent news?

Validate email generation:

Send test emails to your own address with different prospect profiles
Check personalization variables populate correctly (no undefined or null)
Review AI-polished copy for tone and length (should be <150 words)
Test on mobile devices—80% of prospects read emails on phones

Monitor deliverability:

Check spam folder placement using mail-tester.com (aim for 8/10+ score)
Track open rates—healthy cold email campaigns achieve 40-60% opens
Monitor bounce rates—above 5% indicates list quality issues
Watch for spam complaints—above 0.1% requires immediate investigation

Common troubleshooting:

Issue	Cause	Solution
Emails not sending	Gmail API rate limit hit	Add Wait node with 3-5 min delay between sends
Generic research output	GPT-3.5 instead of GPT-4	Switch to gpt-4 model in OpenAI node
Follow-ups not triggering	Schedule node timezone mismatch	Set timezone to your local time in Schedule node
Duplicate emails sent	Status not updating	Add error handling to Status Update node

Deployment Considerations

Production Deployment Checklist

Area	Requirement	Why It Matters
Error Handling	Retry logic with exponential backoff on API failures	Prevents data loss when OpenAI or Gmail APIs timeout
Rate Limiting	50 emails/day per Gmail account, 2-5 min delays	Maintains sender reputation and avoids spam flags
Monitoring	Daily email report of sent/failed/responded counts	Detects issues within 24 hours vs discovering after campaign ends
Backup	Weekly export of prospect data and email logs	Enables recovery if Google Sheets is accidentally deleted
Compliance	Unsubscribe link in footer, honor opt-outs within 24 hours	Meets CAN-SPAM requirements, avoids legal issues
Documentation	Node-by-node comments explaining logic	Reduces troubleshooting time from hours to minutes

Scaling considerations:

Multiple sending accounts: Rotate between 3-5 Gmail accounts for 150-250 emails/day capacity
Dedicated IP: Switch to SendGrid with dedicated IP at 500+ emails/day for better deliverability control
Research caching: Store company research for 30 days to avoid duplicate OpenAI calls for repeat prospects
Batch processing: Process prospects in groups of 25 to prevent workflow timeouts on large lists

Real-World Use Cases

Use Case 1: SaaS Sales Team

Industry: B2B SaaS selling to marketing teams
Scale: 200 prospects/week, 3-person sales team
Modifications needed: Add LinkedIn Sales Navigator integration for prospect enrichment, customize research prompt for marketing pain points, integrate with HubSpot for deal tracking
Results: 18% response rate (vs 4% with generic emails), 60% time savings on outreach

Use Case 2: Agency New Business

Industry: Digital marketing agency targeting e-commerce brands
Scale: 500 prospects/month, 1 BD person
Modifications needed: Research focuses on website performance and ad spend, email templates emphasize ROI case studies, follow-up sequence includes free audit offer
Results: 12% response rate, 8 qualified calls/week, $40K in new business first quarter

Use Case 3: Recruiting Outreach

Industry: Technical recruiting for software engineers
Scale: 1,000 candidates/month, 2 recruiters
Modifications needed: Research pulls GitHub activity and tech stack, emails emphasize company culture and growth, shorter follow-up cadence (2-3-5 days)
Results: 25% response rate, 15% conversion to phone screen, 40% reduction in time-to-hire

Customizations & Extensions

Alternative Integrations

Instead of Gmail:

SendGrid: Best for high-volume sending (500+ emails/day)—requires 8 node changes to swap Gmail nodes for HTTP Request nodes with SendGrid API
Mailgun: Better deliverability analytics and webhook support—swap Gmail nodes, add webhook listener for opens/clicks
Amazon SES: Lowest cost at scale ($0.10 per 1,000 emails)—requires AWS credential setup, similar HTTP Request node configuration

Instead of Google Sheets:

Airtable: Better for team collaboration with rich field types—change Google Sheets nodes to Airtable nodes, add view filtering for campaign segments
PostgreSQL: Best for 10,000+ prospects with complex querying—requires database setup, SQL query nodes replace spreadsheet reads
HubSpot Lists: Direct CRM integration eliminates sync step—use HubSpot nodes for both read and write operations

Workflow Extensions

Add LinkedIn automation:

Connect to Phantombuster or Apify for LinkedIn profile scraping
Extract recent posts and engagement data
Reference LinkedIn activity in email personalization
Nodes needed: +4 (HTTP Request for Phantombuster, Function for data parsing, Set for variable storage)

Implement A/B testing:

Create subject line variants (3-5 options)
Randomly assign prospects to test groups
Track open rates by variant
Auto-select winning subject line after 50 sends
Nodes needed: +6 (Function for random assignment, Split in Batches, Merge for results, IF for winner selection)

Add response classification:

Set up Gmail webhook to catch replies
Use OpenAI to classify response sentiment (Positive/Neutral/Negative/Out-of-office)
Auto-create tasks in CRM for positive responses
Pause follow-up sequences for any reply
Nodes needed: +8 (Webhook trigger, OpenAI classification, Switch for routing, CRM task creation)

Scale with multi-channel outreach:

Add LinkedIn connection requests after email 1
Send LinkedIn message after email 2 if no response
Include SMS for high-value prospects after email 3
Coordinate timing across channels
Performance improvement: 35-50% higher response rates with multi-channel
Nodes needed: +12 (Phantombuster for LinkedIn, Twilio for SMS, complex timing logic)

Integration possibilities:

Add This	To Get This	Complexity
Clearbit Enrichment	Company size, tech stack, funding data	Easy (3 nodes)
Calendly Integration	One-click meeting booking in emails	Easy (2 nodes)
Slack Notifications	Real-time alerts for responses	Easy (2 nodes)
Zapier Webhook	Connect to 3,000+ apps	Medium (4 nodes)
Custom AI Training	Fine-tuned model on your best emails	Advanced (requires OpenAI fine-tuning)

Get Started Today

Ready to automate your cold email outreach?

Download the template: Scroll to the bottom of this article to copy the complete n8n workflow JSON
Import to n8n: Go to Workflows → Import from File, paste the JSON
Configure your services: Add API credentials for OpenAI, Gmail, and Google Sheets
Test with 5 prospects: Verify research quality and email generation before scaling
Deploy to production: Set your daily schedule and activate the workflow

This agent transforms cold outreach from a time-consuming manual process into a scalable, personalized system. The AI research ensures every email feels custom-written while the automated follow-ups recover responses you'd otherwise miss.

Need help customizing this workflow for your specific outreach strategy? Schedule an intro call with Atherial at https://atherial.ai/contact.