Email support agent

Client: Personal project
Type of project: AI
Role: AI Creative
Tool: N8N
Year: 2024-2025

The challenge – reducing inbox overload
Incoming emails often contain simple, repetitive support requests, but some require detailed answers. The challenge was to see if an AI agent could read, classify, and respond to emails automatically while leaving the complex cases for human review.

Solution
I built a prototype that read each incoming email and determined whether the question could be answered directly or needed escalation. The agent was connected to a knowledge base (database) for getting information related to the request. Simple requests triggered an automatic reply, while more complex ones generated a draft for a human to review and approve before sending.

Outcome
The email support agent demonstrated how automation can ease the burden of repetitive communication while keeping humans in the loop for sensitive or complex issues. It showed how efficiency and oversight can work together to create trust in AI-driven support flows.

The diagram shows the email agent flow. Incoming emails are read and analyzed by the agent, which connects to a knowledge base (database). Easy cases are answered automatically and sent to the user. More difficult cases generate a draft that is reviewed by a human before being sent. This structure balances automation with control.

A customer emails about a problem, the Email Support Agent analyzes the message, identifies it as a simple case based on the knowledge base, and sends back an automated response with a fix.

Contact

Email: pierre@bremell.com
Phone: +46 709353902
LinkedIn: bremell