Email AI Assistant

This project uses Python and AI to summarize and automate actions from emails.

Overview

The code connects to an email inbox, sends emails to an AI API (OpenAI or Azure), and processes the summaries and actions returned. Useful for automating parts of email workflow.

Key features:

• Summarizes email content with AI
• Extracts action items into TODOs
• Saves summaries for searchability
• Optionally, creates tasks in Monday.com from emails
• Handles some errors
• Avoids processing duplicate emails
• Tracks AI token usage

Setup

1. Clone this repo
2. Obtain API keys for OpenAI/Azure and add to .env
3. Configure settings in config.yml like email account
4. Run pip install -r requirements.txt
5. Schedule main.py to run, like with cron

Usage

• Set reset_items=True initially to clear existing files
• View output summaries in summary.txt
• View extracted actions in todos.txt
• Errors are logged to errors.txt
• Token usage tracked in token_usage.txt

Optional parameters to run

• -r True or -r False to reset all files. We use files for looking for duplicates, cheap, easy. Default is False because we do not want to process duplicate emails. If files are reset, they are deleted.
• -m True or -m False to save To Do's to Monday.com. Default is False

Configuration

Key settings in config.yml:

• TARGET_MAILBOX_NAME: Name of the mailbox folder to scan in your Outlook (in case of multiple mailboxes)
• TARGET_DOMAIN: Only process emails from this domain
• TARGET_RECEIVER: Who the assistant is processing mail for
• SMTP_ADDRESS_MARKER: Used to parse complex Outlook address formats, and help ensure we're only processing internal emails
• MONDAY_BOARD: Name of the Monday.com Board to save To-Dos to
• MONDAY_GROUP: Name of Group within Monday.com Boars to
• Note: This only processes the inbox, not folders.

See code comments for additional config options.

Environment Variables in .env

Use azure or open_ai to control which endpoint to use

• API_TYPE Set to 'azure' or 'openai' to direct which API to hit.

Environment Variables For Azure OpenAI Endpoint. Use with azure above. Given Microsofts LLM approach, can only call 1 API endpoint for one model, no need to specify model name

• AZURE_OPENAI_API_BASE
• AZURE_OPENAI_API_KEY
• AZURE_OPENAI_API_DEPLOYMENT_NAME
• AZURE_OPENAI_API_VERSION

Environment Variables for using OpenAI's ChatCompletion Endpoint. Use with openai above. With OpenAI's API You must specify the chat model to use

• OPENAI_API_KEY
• CHAT_MODEL

Monday.com token

• MONDAY_TOKEN

Contributing

Pull requests welcome!

License

MIT