AI Outreach Pipeline — Personalized at Scale by Ameer HamzaAI Outreach Pipeline — Personalized at Scale by Ameer Hamza

AI Outreach Pipeline — Personalized at Scale

Portfolio project
Ameer Hamza

The Problem

A consulting firm was doing cold outreach manually. Their process: pull a list of prospects from LinkedIn, research each one, write a personalized email, send it, track it in a spreadsheet. One person doing this could handle maybe 15–20 emails a day, and the quality dropped fast after the first 5 because research fatigue is real.
The bigger problem was consistency. Some emails were good. Some were lazy. The ones sent on Friday afternoons were noticeably worse than Monday morning ones. Human output varies. A system doesn't.

What I Built

An end-to-end outreach automation built in n8n that runs on a schedule. Here's the flow:
It starts by pulling the next batch of uncontacted prospects from an Airtable base. For each prospect, it makes an API call to fetch publicly available data about their company — recent news, company size, industry. It feeds that data into an OpenAI prompt I engineered specifically for cold outreach — the prompt uses their company name, recent activity, and a specific pain point mapped to their industry. The AI writes a personalized first paragraph. The template handles the rest.
The email goes out through Gmail API with proper sending delays between each one (to avoid spam flags). The prospect's record in Airtable gets updated automatically — status changes to "Contacted", timestamp logged, email body saved. When a reply comes in, a second workflow detects it, classifies it (interested / not interested / out of office), and sends a Slack notification only for the "interested" ones.

The Result

Output went from 15–20 manual emails/day to 80–100 AI-personalized emails/day with one person doing zero writing. Reply rate held steady compared to manual outreach because the personalization quality was consistent — not dependent on how tired the person was. The team went from spending 4 hours/day on outreach to 20 minutes reviewing replies.
Tech used: n8n, Airtable API, OpenAI GPT-4o API, Gmail API, scheduling nodes, email classification prompt, Slack webhook notifications, conditional logic for reply routing.

1 like · 0 views · May 2, 2026

Built an AI system that writes and sends personalized cold outreach — so the client only touches replies worth reading

Tags