Churn Monitor

A CrewAI Flow demo for identifying and preventing customer churn through AI-powered analysis and automated email outreach.

Demoed at CrewAI Signal 2025 - Complex Event-based Agentic Workflows Workshop.

Overview

This project demonstrates an agentic system that:

• Analyzes customer data - Reviews account information and support interactions
• Classifies churn risk - Identifies customers at high, medium, or low risk of churning
• Generates personalized emails - Creates targeted outreach to prevent churn
• Produces reports - Outputs an HTML report with findings and recommendations

Note: All data used in this project is fictional and generated for demonstration purposes only.

Quick Start

Prerequisites

• Python >=3.10 <3.14
• UV package manager
• CrewAI
• OpenAI or Anthropic API key

Installation

pip install uv
crewai install

Create a .env file in the project root:

OPENAI_API_KEY=your_key_here
ANTHROPIC_API_KEY=your_key_here # alternative
SERPER_API_KEY=your_key_here # optional

Running

crewai flow kickoff

The flow will analyze the data and generate a churn risk report at reports/churn_risk_report.html.

Questions?

Feel free to reach out at vini[at]hey[dot]com 🙃