I built an AI cold email machine in n8n — here's exactly how it works

A few months ago I was spending 3-4 hours a day manually researching leads, writing personalised emails, and following up. Classic agency grind.

Then I built a fully automated cold email workflow in n8n that handles the entire process — from lead research to personalised copy to sending — without me touching it.

Here's the exact flow:

Step 1 — Lead inputI feed a list of business names + URLs into a Google Sheet. n8n watches the sheet and picks up new rows automatically.

Step 2 — AI enrichmentFor each lead, n8n scrapes the company website and pulls key info: what they do, recent blog posts, any pain points mentioned. This context goes to GPT-4o.

Step 3 — Personalised email generationGPT-4o writes a custom opening line referencing something specific about that company — not generic "I loved your website" stuff. Real specifics. Then it slots that into a proven email framework.

Step 4 — Send + trackThe email goes out via SMTP (or whatever sending tool you use). Lead status in the sheet updates automatically to "Sent."

Step 5 — Follow-up sequenceIf no reply after 3 days, a follow-up gets triggered automatically. Different angle, shorter copy. Works like a drip.

The result? I went from spending ~4 hours/day on outreach to about 15 minutes reviewing what went out. Reply rates actually went UP because the personalisation improved.

The whole thing runs on n8n (self-hosted or cloud), costs almost nothing to operate, and took me about a weekend to build the first version.

If you're running an AI agency or selling any kind of B2B service, this is probably the highest-ROI automation you can build right now.

Drop a comment below if you want me to break down any specific step in more detail — happy to go deeper on the AI prompt structure or the follow-up logic 👇

5 comments