I built an AI cold email machine that runs 24/7 — here's exactly how it works

Most people spend 2–3 hours a day on cold outreach. I spent 2 days building a system that does it for me — indefinitely.

Here's the full breakdown:

Step 1: Lead sourcing I use a multi-source scraper to pull leads from Apollo, LinkedIn, and Google Maps depending on the niche. Each lead gets enriched with company info, tech stack, and recent activity signals.

Step 2: AI personalization Every email gets a unique opening line generated by GPT based on the prospect's LinkedIn bio, recent posts, or company news. Not just "Hey [FirstName]" — actually personalized first lines that reference real things.

Step 3: n8n orchestration The whole workflow runs in n8n. It handles the scheduling, sends via SMTP, tracks opens/clicks, and routes hot replies to a Slack notification so I can jump in manually when someone responds.

Step 4: Auto follow-ups If no reply after 3 days, it sends a follow-up. Day 7, another one. The sequence stops automatically the moment someone replies. No double-emailing.

Results after 30 days: — 847 emails sent — 34% open rate (average in my niche is ~22%) — 11 interested replies — 3 paid clients

The whole thing costs about $0.003 per email to run (OpenAI API + SMTP). Way cheaper than any cold email tool with a monthly subscription.

The key insight: personalization at scale is the unlock. Generic blasts don't work anymore. AI-generated first lines that actually reference real context — that's what moves the needle.

Drop a comment if you want me to break down the n8n workflow in more detail — happy to share the logic behind each node 👇

4 comments