Polaris AI - RAG-Powered Meeting Assistant

Submission By Team C0xFFEE OverFlow for the Murf Coding Challenge 4

📌 Overview

Polaris AI is an intelligent meeting assistant that integrates Recall.ai, Cerebras, and Murf AI to provide real-time meeting support.
It can:

• Join virtual meetings via Recall.ai.
• Listen and respond contextually when invoked with its name (“Polaris”).
• Use RAG (Retrieval-Augmented Generation) to answer from uploaded documents (PDF, DOCX, TXT).
• Generate MOM (Minutes of Meeting) after each session (as PDF).
• Provide real-time speech-to-speech responses using Murf AI voices.
• Allow mute/unmute, response mode switching, and voice customization.

---

⚙️ Features

• 🔗 Meeting Integration: Connects to online meetings through Recall.ai bots.
• 📝 Transcript Handling: Captures and processes live transcripts.
• 🧠 AI Assistant: Context-aware responses using Cerebras LLM.
• 📚 RAG System: Upload documents for contextual Q&A.
• 🎤 Voice Output: Generates natural speech with Murf AI voices.
• 📄 MOM Generation: Automatically creates concise minutes of meeting in PDF.
• 🎛️ UI Controls:
  • Start/Leave Meeting
  • Upload/Delete Documents
  • Change Voice
  • Switch Modes: Detailed, Quick, Document-based, General
  • Mute/Unmute

---

🚀 Setup

1. Clone Repository

git clone https://github.com/AayushBeura/Polaris.git
cd Polaris

2. Install Dependencies

pip install -r requirements.txt

3. API Keys

Polaris requires three API keys. Store them in user_config.json:

{
  "RECALL_API_KEY": "your_recall_api_key",
  "CEREBRAS_API_KEY": "your_cerebras_api_key",
  "MURF_API_KEY": "your_murf_api_key"
}

4. Set up Ngrok (for secure tunneling)

Polaris uses ngrok to create a public HTTPS tunnel for meeting connections.

1. Sign up at ngrok and get your auth token.
2. Set your auth token (replace YOUR_TOKEN with your actual key):

ngrok config add-authtoken YOUR_TOKEN

3. Polaris will use ngrok automatically when starting a meeting tunnel.

5. Run Application

python polaris.py

The app will start a Flask server with Socket.IO support.
Ngrok tunneling is available for external meeting connections.

---

🖥️ Interface

• Web-based dashboard (Flask + Socket.IO).
• Blurred background with animated UI elements.
• Sections:
  • Meeting Controls
  • Document Management
  • System Status
  • Transcript & Logs

---

🔄 Workflow

---

📦 Project Structure

polaris.py             # Main Flask + Socket.IO backend
user_config.json       # Stores API keys
uploads/               # Uploaded documents & generated MOM PDFs
static/                # Assets (e.g., logo, backgrounds)
requirements.txt       # Python dependencies

---

🛠️ Tech Stack

• Backend: Flask, Flask-SocketIO
• AI APIs: Cerebras LLM, Recall.ai, Murf AI
• Vector Search: FAISS + Sentence Transformers
• Document Parsing: PyPDF2, python-docx
• PDF Generation: FPDF
• UI: HTML, CSS, JavaScript (served by Flask)

---

📑 License

MIT License