Marketing Chatbot

A specialized AI-powered marketing assistant chatbot that helps users with marketing, branding, and business growth questions.

🚀 Features

• Specialized Knowledge: Strictly focused on marketing, branding, advertising, and growth strategies.
• RAG Architecture: Uses LangChain for Retrieval-Augmented Generation (RAG) to fetch relevant context from the vector store.
• Context Awareness: Leverages LangChain's History capabilities to remember conversation context for seamless follow-up questions.
• Text-to-Speech (TTS): Built-in TTS toggle to read responses aloud.
• Modern UI: Dark-themed, responsive, and animated chat interface.
• Voice Agent (Deprecated): The project previously supported a voice-to-voice agent but is now focused on text/chat interactions.

📂 Project Architecture

The project is structured into clear Frontend and Backend components:

marketing_chatbot/
├── backend/                # Python/Flask Backend
│   ├── app.py              # Main application entry point
│   ├── src/                # Source code for logic & prompts
│   │   ├── helper.py       # Embeddings & utility functions
│   │   └── prompt.py       # System prompts & instructions
│   ├── data/               # Data directory for RAG content
│   ├── store_index.py      # Script to process & index data
│   ├── requirements.txt    # Python dependencies
│   └── .env                # Environment variables
│
├── frontend/               # Web Interface
│   ├── templates/          # HTML templates (chat.html)
│   └── static/             # Static assets (CSS/JS/Images)
│
└── README.md               # Project Documentation

🛠️ Technology Stack

• Backend: Python, Flask, LangChain
• AI/LLM: Groq (Llama 3.1), Google Gemini (Embeddings), Pinecone (Vector Database)
• Frontend: HTML5, CSS3 (Bootstrap + Custom Animations), JavaScript (jQuery)
• Tools: Dotenv for configuration

⚡ How to Run

1. Prerequisites:

   • Python 3.8+ installed.
   • API Keys for Pinecone, Groq, and Google Generative AI.

2. Environment Setup:

   • Navigate to the backend folder.
   • Ensure your .env file is present containing:

     PINECONE_API_KEY=your_key
     GROQ_API_KEY=your_key
     GOOGLE_API_KEY=your_key
     

3. Install Dependencies:

   cd backend
   pip install -r requirements.txt

4. Run the Application:

   python app.py

   The server will start at http://localhost:8080.

🧠 Flow Overview

1. User Input: User sends a message via the web interface.
2. Backend Processing:
   • app.py receives the message.
   • Contextualization: If it's a follow-up question, it's rephrased to be standalone.
   • Retrieval: The system searches the Pinecone vector store for relevant marketing documents.
   • Generation: The LLM generates a concise, marketing-focused answer using the retrieved context.
3. Response: The answer is sent back to the frontend.
4. Display & TTS: The frontend displays the message with an animation and optionally reads it aloud using the Web Speech API.