AI Email Automation

Outbound campaigns generated more reply volume than one person could keep up with. Every inbound reply routed to the founder regardless of intent, each one written from scratch with no template system. Product knowledge and objection handling lived in Google Drive and never made it into replies. Campaign cadence had to stay artificially low because more outreach meant more manual work.

We built an n8n pipeline that watches the inbox, classifies each reply by intent, and retrieves the right context from a vector store of internal Drive docs before writing a draft. Meeting requests get real calendar availability pulled in automatically. The founder gets a ready-to-review draft sitting in Gmail on the original thread, same inbox, no new tools, final send stays with him.

Intent-routed RAG draft engine writing directly into Gmail. The system classifies each reply as an objection, meeting request, question, or out-of-office before retrieval so the right context is pulled every time. Drafts are grounded in the actual product knowledge base, not guessed, no hallucinated claims, no generic responses. The Drive ingestion pipeline keeps the vector store current automatically.

n8n orchestrates two workflows: reply pipeline and document ingestion. Everything else is a signal source, retrieval layer, or output target.