🧠 Multi-Agent Outlook Orchestration App

This repository provides a lightweight multi-agent orchestration system for Microsoft Outlook using local Ollama LLMs and CrewAI. It continuously monitors Outlook inboxes and processes emails with specialized agents for classification, summarization, urgency detection, tone analysis, and auto-reply generation.

✨ Features

• ✅ CrewAI-powered orchestration of independent agents
• 📥 Multiple inbox support
• 🧠 Email categorization into Work, Personal, Promotional, etc.
• ✍️ Summary, tone, urgency detection, and response drafting
• 📨 Outlook draft auto-reply generation
• ⚡ Uses quantized LLMs (via Ollama) for CPU-only inference
• 🔁 Runs continuously in the background with polling
• 🧩 Modular and easily customizable agent/task configuration
• 💻 Designed to be lightweight for laptop/desktop users — no GPU required
• 📤 The application uses COM automation (via pywin32) to interact with Outlook. Ensure Outlook is running and accessible.
• 🦙 Ollama must be running in the background. First-time setup may pull the model.

🧰 Prerequisites

• Windows with Microsoft Outlook installed and configured
• Python 3.10 or later
• Git + pip

🧪 How to Run

1. Clone the repo and create a virtual environment:

   git clone https://github.com/bearded-impala/outlook-agents.git && cd outlook-agents
   python -m venv outlook-agents && outlook-agents\Scripts\activate && python -m pip install --upgrade pip

2. Install dependencies:

   pip install -r requirements.txt
   

3. Update src/config/config.py Update IDENTITY as per your preference

4. Run the application: The app will keep running and monitor your Outlook inbox for unread emails.

   python src/main.py
   

⚙️ Configuration (src/config/config.py)

IDENTITY: Your Persona
MODEL_NAME: The model must be from https://ollama.com/search (e.g., gemma:2b, llama3, mistral, etc.)
POLLING_INTERVAL: How frequently the app checks for new emails in seconds

🧩 Components and Their Purposes

This project leverages several libraries, models, and tools to create a lightweight, CPU-efficient multi-agent orchestration app for Microsoft Outlook. Below is an overview of everything used and its purpose:

• pywin32
  Purpose: Enables COM automation to interact with Microsoft Outlook. It allows the app to read emails, mark them as read, and access other Outlook functionalities.

• crewai
  Purpose: Serves as the multi-agent orchestration framework. Crew AI is used to register and manage the execution of our agents (Inbox Scanner, Summarizer, Urgency Detector, Tone Analyzer, Auto-Responder, Follow-up Tracker, and Task Manager).

• BeautifulSoup (bs4)
  Purpose: Cleans up and extracts readable text from email HTML content. Used to strip HTML tags and formatting artifacts while preserving readable structure.

• Windows Task Scheduler Purpose: Configures the application to run automatically at system startup or on a defined schedule, ensuring continuous background operation.

⏰ How to Schedule the Python Script Using Task Scheduler

Follow these steps to run your Python script automatically using Windows Task Scheduler:

1. Open Task Scheduler

Press Win + R, type taskschd.msc, and press Enter.

2. Create a New Task

In the right-hand pane, click Create Task.

3. Configure the General Tab

• Provide a Name for the task.
• Select "Run whether user is logged on or not".
• (Optional) Check "Run with highest privileges" if your script requires admin access.

4. Set Up the Trigger

• Go to the Triggers tab and click New...
• In the Begin the task dropdown, select At startup (or choose your preferred schedule).
• Click OK.

5. Set Up the Action

• Go to the Actions tab and click New...
• For Action, select Start a program.
• In the Program/script field, enter the path to your Python executable. Example:

  C:\Users\YourName\AppData\Local\Programs\Python\Python39\python.exe
  

• In the Add arguments field, enter the path to your Python script. Example:

  "C:\path\to\your_script.py"
  

• In the Start in field, enter the directory where your script is located. Example:

  C:\path\to\
  

• Click OK.

---

Your script will now run automatically based on the trigger you've set!