Descriptions:
Nate Herk walks through the full architecture and live construction of an AI-powered email management system he built in n8n and sold to a client for $1,650. The system is composed of two distinct parts: an inbox routing agent that classifies incoming emails by category—finance, sales, opportunities—and logs them to Google Sheets, and a Slack-based personal assistant that can retrieve emails from Gmail, look up contacts via Google Contacts, and draft outbound messages that match the client’s writing voice.
The most technically substantive section covers the email drafting subworkflow, where Herk makes an explicit architectural decision to isolate drafting in a sub-agent rather than handling everything in a single node. He explains why writing style belongs in the system prompt rather than a vector database, and describes a practical method for generating a tone-and-style guide: feeding 50–100 of the client’s real emails into an LLM to extract patterns, phrasing tendencies, and vocabulary.
Throughout the build, Herk covers n8n-specific implementation details including structured outputs, memory configuration, and Gmail tool setup. The video doubles as both a developer tutorial for building multi-step automation agents and a freelancer guide for how to scope, price, and pitch AI workflow services to small business clients—making it useful across a range of experience levels.
📺 Source: Nate Herk · Published January 07, 2026
🏷️ Format: Hands On Build
