Catalyst - AI-Powered Desktop Assistant

Catalyst is a dynamic AI-powered desktop assistant designed to help users with various tasks efficiently and intuitively. With its extensible architecture and the capability to execute complex workflows, Catalyst is more than just a personal assistant—it's a powerful tool for managing your digital life.

✨ Features

1. General Assistance and Conversation:

   • Engage in general conversations and receive contextual answers to your questions.
   • Supports Markdown and LaTeX for mathematical and scientific explanations.

2. File and Folder Management:

   • Create, edit, and manage files seamlessly.

3. Google Calendar Integration:

   • List and create calendar events effortlessly.

4. Email Management:

   • List and send Gmail messages, making communication a breeze.

5. GitHub Repository Management:

   • Create and list GitHub repositories directly from the assistant.

6. News Retrieval:

   • Fetch top news headlines based on queries or categories.

7. Contacts Management:

   • List contacts from your Google account for easy access.

---

🛠️ Demos

Example 1: Python Code

The assistant can fetch your recent Gmail messages and save them to a file:

Input:

Can you create a python file with the code for a simple Tic Tac Toe game

Result:

Example 2: Fetching Emails and adding them to a file (Chained Tasks)

Input:

List my last 10 emails and add them to a file named emails.txt

Result:

---

🔧 How It Works

Catalyst combines cutting-edge AI models with carefully designed task parsers to perform complex operations. It achieves this through:

1. Manual Function Calling:

   • A unique approach where the AI outputs tasks as JSON commands, which are parsed and executed by the backend. This system allows precise control and transparency over task execution.

2. Sequential Task Execution:

   • The AI can chain tasks using the output of one as the input for the next, enabling complex workflows like fetching emails and saving them to a file.

3. Custom Parsers:

   • Each feature, from file handling to GitHub management, has a dedicated parser that interprets AI commands and executes them.

4. Modular Design:

   • The project is structured into independent modules for better maintainability and scalability.

---

🚀 Getting Started

Prerequisites

• Node.js for backend and frontend
• Python 3.x for the text-to-speech service
• Ollama for the models (install it using WSL of you're on windows and install the model used in the ollama.js file)
• Google API credentials for Gmail, Calendar, and Contacts
• GitHub token for repository management
• News API key for fetching news headlines

Installation

1. Clone the Repository:

   git clone https://github.com/yourusername/catalyst.git
   cd catalyst

2. Set Up Backend:

   cd server
   npm install

3. Set Up Frontend:

   cd client
   npm install

4. Set Up Python Microservice:

   cd python_microservice
   python -m venv venv
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`
   pip install -r requirements.txt

5. Environment Variables: Create a .env file in the root directories for server and python_microservice. Include keys like:

   GOOGLE_API_CREDENTIALS_PATH=path/to/credentials.json
   GOOGLE_API_TOKEN_PATH= "path/to/token.json"
   GITHUB_TOKEN=your_github_token
   NEWS_API_KEY=your_news_api_key

---

🖥️ Running Catalyst

1. Start the Python Microservice:

   cd python_microservice
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`
   python app.py

2. Start the Backend Server:

   cd server
   npm start

3. Start the Frontend Client:

   cd client
   npm start

---

📚 Design Highlights

1. Manual Function Calling: Catalyst's unique approach to task execution ensures precision and flexibility. The AI outputs tasks in a structured JSON format, which are interpreted by dedicated handlers.

2. Chained Task Execution: Tasks can depend on the results of previous actions, allowing for workflows like fetching news and emailing them to a contact.

3. Error Handling and Modular Parsers: Each feature is independently parsed, ensuring robust error handling and clear separation of concerns.

4. Customizable AI Model: Powered by Ollama's API, the assistant can adapt to user-specific needs, with further potential for fine-tuning.

---

📅 Future Plans

• Sequential Functions: Improve how the model deals with chained tasks (allow it to see the result of each task and, according to it, define the input for the next task)

• Enhanced Voice Interaction: Improve the quality of Speach-To-Text and Text-To-Speach.

• Electron Integration: Package Catalyst as a cross-platform desktop application.

• More Models: Add support for other models like GPT3.5 and 4.

• Extended API Support: Add support for more services like Dropbox, Slack, and Trello.

---

🤝 Contributing

Catalyst is a work in progress, and contributions are welcome! If you find a bug or have an idea for improvement, feel free to open an issue or submit a pull request.

---

📝 License

This project is licensed under the MIT License.