🎙️ Podcast Summarizer Agent

🚀 Overview

Podcast Summarizer Agent is an AI-powered application that automatically retrieves, transcribes, and summarizes podcast episodes. It uses a combination of LangChain, TogetherAI, and MongoDB to create an intelligent summarization pipeline.

🏗️ Architecture

The application uses LangGraph to create a sophisticated agent workflow. Here's how the components interact:

Core Components

1. Input Handler

   • Receives and processes user queries
   • Initializes the conversation state
   • Routes requests to appropriate components

2. Agent Node

   • Powered by TogetherAI
   • Processes input using advanced language models
   • Makes intelligent decisions about next actions
   • Maintains conversation context

3. Tool Node

   • YouTube Search: Finds relevant podcast episodes
   • Transcript Retrieval: Extracts captions from videos
   • Summary Generation: Creates concise episode summaries
   • Database Operations: Manages data persistence

4. Conditional Routing

   • Determines workflow progression
   • Handles error cases and retries
   • Manages conversation flow

✨ Key Features

• 🔍 Smart Search: Automated podcast discovery using SerpAPI
• 📝 AI-Powered Processing:
  • Automatic transcription via YouTube Transcript API
  • Intelligent summarization using TogetherAI
• 🗄️ Data Management:
  • MongoDB integration for persistent storage
  • Efficient metadata management
• 🤖 Advanced Workflow:
  • LangChain-powered agent system
  • Dynamic tool selection
  • Error recovery mechanisms
• 🌐 Dual Interface:
  • Agent works as a cron job on railway.com
  • User-friendly web interface (Streamlit)

🛠️ Technology Stack

Component	Technology	Purpose
Backend	Python 3.8+	Core processing
AI Framework	LangChain	Agent orchestration
Database	MongoDB	Data storage
AI Models	TogetherAI	Content summarization
Search	SerpAPI	YouTube search
Transcription	YouTube Transcript API	Caption extraction
UI	Streamlit	Web interface

📂 Project Structure

summarizer_container/
│── build_agent.py       # LangChain agent configuration
│── get_transcripts_tools.py  # YouTube search and transcript retrieval
│── prompt.py           # Prompt templates and configurations
│── streamlit_app.py    # Streamlit web interface
│── mongo_functions.py  # MongoDB operations
│── env_variables.py    # Environment variable management
│── requirements.txt    # Python dependencies
│── Dockerfile         # Container configuration

🚀 Getting Started

Prerequisites

• Python 3.8 or higher
• MongoDB instance
• Required API Keys:
  • TogetherAI
  • SerpAPI
  • MongoDB

Installation

1. Clone the repository
2. Install dependencies:

   pip install -r requirements.txt

3. Configure environment variables in .env file:

   TOGETHER_API_KEY=your_key_here
   SERPAPI_API_KEY=your_key_here
   MONGODB_URI=your_connection_string
   

Running the Application

Option 1: Streamlit Interface

streamlit run streamlit_app.py

Option 2: Docker Deployment

docker build -t podcast-summarizer .
docker run -p 5001:5001 podcast-summarizer

🔗 API Documentation

/podcast_agent

• Method: GET
• Purpose: Process podcast-related queries
• Parameters:
  • message: User input message
• Response: JSON with agent's response

/healthcheck

• Method: GET
• Purpose: Verify application status
• Response: JSON with system status

💰 Resource Costs

Service	Cost Model	Usage
TogetherAI	Pay-per-use	AI model calls
MongoDB	Free tier available	Data storage
SerpAPI	Pay-per-use	Search operations

🔮 Roadmap

• Enhanced error handling
• Improved summary quality
• Additional podcast sources
• Caching system
• Performance optimizations

---

🎯 Mission: Making podcast content more accessible through AI-powered summarization.