Email Automation System

Business Challenge

A marketing agency was manually sending personalized emails to 1,000+ leads per campaign. A marketing coordinator would spend many hours per week writing individual emails, referencing each lead’s company, industry, and recent news to make them feel personal. It was effective — personalized emails performed well — but it didn’t scale.

The agency had tried mail merge tools, but the results were obviously templated and response rates dropped sharply. They needed something that combined the scale of automation with the personalization of human-written emails. They also needed to handle responses — when a lead replied, someone had to read it, categorize it, and respond appropriately, which consumed significant additional time each week.

Solution

We built an email automation system that uses AI to write genuinely personalized emails at scale. The system pulls each lead’s data from the CRM, researches their company and recent news, and generates a unique email that references specific details — a recent funding round, a product launch, a LinkedIn post. Each email feels hand-written.

The system also handles the full email lifecycle. It schedules sends at optimal times based on the recipient’s timezone and historical open patterns. When a response comes in, the AI reads it, categorizes it (interested, not interested, out of office, question), and either auto-responds or routes it to a human. The system tracks open rates, response rates, and meeting bookings, feeding data back to improve future campaigns.

Technologies

The system uses n8n as the orchestration layer, OpenAI for email personalization and response classification, the Gmail API for sending and receiving emails, and PostgreSQL for lead data, email history, and campaign analytics. n8n workflows handle the entire lifecycle: lead ingestion, email generation, send scheduling, response monitoring, and follow-up sequences.

The personalization engine is the core innovation. For each lead, an n8n workflow gathers context — CRM data, company website, recent news — and passes it to OpenAI with a prompt that produces a personalized email. The prompt includes examples of the agency’s writing style, so the AI’s output matches their voice. A response handler uses OpenAI to classify incoming replies and trigger the appropriate workflow: auto-respond to common questions, route interested leads to sales, and add out-of-office replies to a follow-up queue.

Results

The agency now sends far more personalized emails with the same team size, because the AI handles the writing and the coordinator just reviews and approves. Open rates increased meaningfully because the emails are genuinely personalized and sent at optimal times.

The team saved significant hours each week that were previously spent on manual email writing and response handling. Response rates remained nearly identical to the hand-written baseline — proving that the AI personalization was indistinguishable from human writing. The agency was able to take on considerably more clients without expanding their marketing team, and the system became a key selling point in their pitches.

Screenshots

AI-generated personalized email with lead context

Campaign performance dashboard with open and response rates

Architecture

The system is a set of n8n workflows orchestrated by a campaign manager. The personalization workflow pulls lead data from PostgreSQL, gathers external context, generates an email via OpenAI, and queues it for human review. The send scheduler releases approved emails at optimal times based on recipient timezone and historical patterns.

The response handler monitors the Gmail inbox, classifies each reply using OpenAI, and triggers the appropriate workflow: auto-response, sales routing, or follow-up scheduling. A campaign analytics dashboard tracks open rates, response rates, and meeting bookings, with A/B testing support for subject lines and email approaches. All data is stored in PostgreSQL for reporting and model improvement.