# Telegram Inference Bot Refactoring

This repository contains a refactored version of the Telegram Inference Bot, which now uses a more flexible and maintainable approach for handling different AI providers.

## Changes

1. Introduced an abstract `AIProvider` class and concrete implementations for Anthropic and OpenAI.
2. Refactored the main bot code to use the new AI provider classes.
3. Implemented a factory function `create_ai_provider` for easy provider instantiation.
4. Updated command handlers to work with the new AI provider system.

## How to Use

1. Set up your environment variables in a `.env` file:
   ```
   TELEGRAM_BOT_TOKEN=your_telegram_bot_token
   ANTHROPIC_API_KEY=your_anthropic_api_key
   OPENAI_API_KEY=your_openai_api_key
   ```

2. Install the required dependencies:
   ```
   pip install -r requirements.txt
   ```

3. Run the bot:
   ```
   python telegram_inference_bot.py
   ```

## Commands

- `/start`: Start the bot and receive a welcome message.
- `/clear`: Clear the conversation history and any stored images.
- `/switch`: Switch between smart and regular models (OpenAI only).
- `/toggle`: Toggle between Anthropic and OpenAI providers.
- `/status`: Display the current AI provider and model being used.

## Extending the Bot

To add a new AI provider:

1. Create a new class in `ai_providers.py` that inherits from `AIProvider`.
2. Implement the required methods: `get_chat_response`, `format_messages`, `format_tool_calls`, etc.
3. Update the `create_ai_provider` function to include the new provider.

## Future Improvements

- Implement more robust error handling and logging.
- Add unit tests for the AI provider classes and main bot functionality.
- Extend the README with more detailed usage instructions and examples.