🧠 Meeting Whisperer

Smart Meeting Assistant powered by LLMs, Google Calendar, and MCP

---

📌 Overview

Meeting Whisperer is an intelligent assistant that automates the end-to-end workflow of:

• Summarizing meeting transcripts using GPT-4
• Extracting action items for each participant
• Finding mutual availability across Google Calendars
• Detecting US national holidays
• Scheduling the next meeting via Google Calendar

The project is built using MCP (Model Context Protocol) for modular tool orchestration and is designed for real-world productivity and professional collaboration.

---

🚀 Features

Feature	Description
🧠 GPT-powered Summary	Extracts concise summaries and action items using OpenAI GPT-4
📅 Mutual Availability Check	Finds earliest shared availability using Google Calendar API
🏖️ Holiday Conflict Detection	Identifies US national holidays via separate MCP server
📤 Google Calendar Integration	Creates and sends calendar invites to all participants
🔌 Modular Design with MCP	Uses FastMCP to manage tool servers independently

---

📁 Project Structure

MeetingWhisperer/
├── streamlit_ui.py            ← Main Streamlit frontend
├── summarize_utils.py         ← GPT summarization logic
├── calendar_server.py         ← MCP tool server for Google Calendar
├── holiday_server.py          ← MCP tool server for holiday checking
├── authorize_google.py        ← Script to generate calendar tokens
├── token.json                 ← Creator's calendar token
├── tokens/                    ← Folder containing participant tokens (e.g., tokens/email1.json)
├── .env                       ← Contains your OpenAI API key

---

🛠️ Setup Instructions

1. Clone and Install Dependencies

git clone https://github.com/yourusername/meeting-whisperer.git
cd meeting-whisperer
uv venv
uv pip install -r requirements.txt

Add OpenAI API

Create a .env file in the root directory:

OPENAI_API_KEY=your_openai_key_here

3. Google Calendar Setup

Go to Google Cloud Console Create a project → Enable Google Calendar API Download your OAuth 2.0 credentials → save as client_secret.json Then authorize your own calendar:

uv run python authorize_google.py

For participants, place their tokens as:

tokens/<participant_email>.json

4. Run MCP Tool Servers

In separate terminals:

uv run python calendar_server.py

uv run python holiday_server.py

5. Run the Streamlit App

uv run streamlit run ui.py

🎥 Demo Flow

📤 Upload a .txt meeting transcript

📧 Enter participant email addresses

🧠 Generate summary & action items using GPT-4

🗓️ View the next suggested meeting date (from transcript)

🔍 Check for calendar availability + holiday conflicts

✅ Create a Google Calendar invite at the mutual time

📨 Event appears on all participant calendars

🧠 Future Plans

🔁 Recurring meeting detection & suggestions

💬 Slack/email delivery of summaries

📞 Zoom & Microsoft Teams integration

🏢 Support for custom org-wide holidays

📅 ICS calendar download and multi-timezone support

👤 Optional user authentication for enterprise use

---

🧬 What is MCP?

MCP (Model Context Protocol)

Introduced November 2024 by Anthropic as an open-source protocol, MCP allows for the integration between LLM applications and external data sources and tools.

In this project, calendar scheduling and holiday detection are offloaded to dedicated MCP tool servers — making the system highly modular and extensible.