AI Automation Workflow (Python + Gemini)

Description This project demonstrates an end-to-end AI automation workflow built with Python.

It simulates a real-world business process where data is:

• fetched from an external API
• analyzed using AI (Google Gemini)
• transformed into structured insights
• saved as a CSV report
• and triggers email alerts for high-priority items

Technologies Used

• Python
• Requests (API calls)
• Google Gemini API (google-genai)
• CSV module
• SMTP (email automation)

Features

• Fetches data from a public API
• Processes and filters text data
• Uses AI to generate:
  • Summary
  • Tone classification
  • Priority level
  • Recommended action
• Saves structured output into CSV
• Sends automated email alerts when high-priority items are detected

Project Structure

AI_automation/ │ ├── main.py ├── config.py ├── requirements.txt ├── README.txt ├── report_ai_v2.csv │── get.py │── save_csv.py │── email_service.py │── parse.py

How to Run

1. Clone the repository

git clone https://github.com/pisnictudor/AI_automation.git cd AI_automation

2. Install dependencies

pip install -r requirements.txt

3. Set environment variables

Gemini API key (PowerShell):

$env:GEMINI_API_KEY="YOUR_API_KEY"

Email credentials (recommended):

$env:SENDER_EMAIL="[email protected]" $env:APP_PASSWORD="your_app_password" $env:RECEIVER_EMAIL="[email protected]"

4. Run the script

python main.py

Output

The script generates a CSV file:

report_ai_v2.csv

With the following columns:

• ID
• Title
• Summary
• Tone
• Priority
• Recommended Action

Email Automation Logic

The system sends an email only if high-priority items are detected.

Behavior:

• No "high" priority → No email sent
• At least one "high" → Email alert triggered

Email contains:

• ID
• Title
• Summary
• Recommended action

Example Use Case

This project simulates a real business workflow such as:

• marketing content analysis
• customer feedback triage
• lead prioritization
• automated reporting and alerting systems

Key Concepts Demonstrated

• API integration
• AI-powered data analysis
• prompt engineering
• data parsing and structuring
• CSV report generation
• event-driven automation
• email notifications

Future Improvements

• Add logging system
• Add retry mechanisms for API calls
• Export data to Google Sheets
• Build a dashboard (Streamlit)
• Schedule execution (cron / task scheduler)
• Improve AI response validation

Portfolio Value

This project demonstrates the ability to:

• build automation pipelines
• integrate AI into real workflows
• design modular Python applications
• create business-oriented solutions

Author

Tudor Pisnic GitHub: https://github.com/pisnictudor