Autonomous Assistant

An experimental autonomous AI assistant that plans and executes tasks to achieve a given objective using OpenAI’s API.

Overview

This project implements an AI assistant capable of autonomously planning and executing tasks to achieve a specified objective. It leverages the beta features of the OpenAI API, including assistants, threads, and vector stores.

The assistant interacts with functions defined in a separate functions module, allowing it to perform various actions during its planning and execution phases.

Features

• Autonomous Task Planning: The assistant can determine the next logical steps to achieve the defined objective without manual intervention.
• Function Calling: Utilizes function calling to perform specific tasks, enabling more complex operations.
• Persistent State: Stores assistant, thread, and vector store IDs in a .env file for continuity between runs.
• Command-Line Interface: Offers a simple CLI to run, reset, update, or list messages from the assistant.
• Logging: Records function outputs and assistant responses for review and debugging.

Prerequisites

• Python: Version 3.7 or higher.
• OpenAI API Key: An API key with access to the beta features (assistants, threads, vector stores).
• Dependencies: Listed in requirements.txt.

Installation

1. Clone the Repository

   git clone https://github.com/yourusername/autonomous-ai-assistant.git
   cd autonomous-ai-assistant

2. Create a Virtual Environment (Optional)

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install Dependencies

   pip install -r requirements.txt

4. Set Up Environment Variables

   Create a .env file in the root directory and add your OpenAI API key:

   OPENAI_API_KEY=your-openai-api-key

Configuration

The assistant’s behavior and settings can be configured via the config.py file:

• OBJECTIVE: The main objective for the assistant to achieve.
• MODEL: The OpenAI model to use (e.g., 'gpt-4').
• ASSISTANT_ID, THREAD_ID, VECTOR_STORE_ID: Managed automatically but can be set manually if needed.

Usage

The script provides a command-line interface with several actions:

python assistant.py [action]

Actions

• run: Start the assistant’s autonomous execution loop.
• thread: List the messages in the current thread, including the assistant’s responses.
• reset: Reset the current thread and clear logs and outputs.
• update: Update the assistant’s configuration.

python assistant.py [action]

Function Definitions

The assistant can call various functions defined in the functions.py module. These functions are registered with the assistant and can perform specific tasks.

• Adding Functions: You can add or modify functions in functions.py to extend the assistant’s capabilities.
• Function Registration: Ensure new functions are added to the DEFINITIONS list in functions.py for the assistant to recognize them.

Logging and Outputs

• Logs: Outputs of function calls are stored in the ./logs directory.
• Outputs: Assistant's outputs or results are stored in the ./outputs directory.

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any changes or enhancements.