Llama3 RAG Enabled Chatbot

Overview

The bot uses a vector store to provide relevant answers based on the data fetched from Intercom articles. It also serves a web server and utilizes ngrok for exposing the local server to the internet.

Repository Structure

• main.py: The main script to start the bot, server, and manage subprocesses for ollama and ngrok.
• telegram_bot.py: Handles Telegram bot functionality and communication.
• data_processor.py: Fetches and processes data from Intercom.
• utils.py: Utility functions including HTML stripping.
• vector_store.py: Manages the vector store and embedding generation.
• web_server.py: Serves the web API using Quart and Hypercorn.

Setup

Prerequisites

• Python 3.10+
• pip package manager
• An Intercom account and API token
• Telegram Bot API credentials
• ngrok for exposing the local server

Installation

1. Clone the repository:

   git clone https://github.com/your-username/custom-intercom-bot.git
   cd custom-intercom-bot

2. Create and activate a virtual environment:

   python3 -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install the required dependencies:

   pip install -r requirements.txt

4. Create a .env file in the root directory and add your environment variables:

   API_ID=your_telegram_api_id
   API_HASH=your_telegram_api_hash
   BOT_TOKEN=your_telegram_bot_token
   CHAT_ID=your_chat_id
   INTERCOM_TOKEN=your_intercom_token
   PROMPT_TEMPLATE="Your prompt template here"

5. Create an Ollama modelfile.

   ollama pull llama3
   ollama create custom-chat-bot -f custom-bot

Usage

To start the bot and the server, run:

python3 main.py