GPT Pilot is revolutionizing the world of swift app development by providing a cutting-edge AI-powered developer companion. Unlike traditional AI tools that simply offer autocomplete or helper functions, GPT Pilot aims to create fully working, production-ready apps with minimal human intervention. This groundbreaking technology has far-reaching implications for the future of software development.

How Does It Work?

GPT Pilot is designed to research how much Large Language Models (LLMs) can be utilized to generate fully working, production-ready apps while a developer oversees the implementation. The main idea is that AI can write most of the code for an app (around 95%), but for the remaining 5%, a human developer is still needed until we achieve full Artificial General Intelligence (AGI).

Getting Started

If you're interested in using GPT Pilot, there are several ways to get started. The easiest method is by downloading the GPT Pilot VS Code extension if you use Visual Studio Code as your Integrated Development Environment (IDE). Alternatively, you can use the CLI tool.

To start using the CLI tool, follow these steps:

  1. Install Python 3.9+ and optionally PostgreSQL.
  2. Clone the GPT Pilot repository: git clone https://github.com/Pythagora-io/gpt-pilot.git
  3. Activate the virtual environment: python3 -m venv venv (on Windows, use venv\Scripts\activate)
  4. Install dependencies: pip install -r requirements.txt
  5. Create a config.json file and set your key and other settings.
  6. Start GPT Pilot: python main.py

Examples

Check out our GitHub repository for example apps created with GPT Pilot.

How to Run GPT Pilot in Docker?

To run GPT Pilot in Docker, follow these steps:

  1. Install Docker.
  2. Clone the GPT Pilot repository: git clone https://github.com/Pythagora-io/gpt-pilot.git
  3. Build and start the Docker container: docker build -t gpt-pilot . followed by docker run -it gpt-pilot

CLI Arguments

GPT Pilot offers several command-line options for managing projects and apps. These include:

  • python main.py --list: List all available projects and their branches.
  • python main.py --project : Start working on a specific project.
  • python main.py --delete : Delete a project with the specified app ID ( Warning: this cannot be undone!).
  • python main.py --help: Display help for all available options.

Architecture

GPT Pilot's architecture is designed to mimic real-life software development. The process begins with the Product Owner, who enters the app name and description. Next, the Specification Writer agent asks questions to clarify requirements. The Architect writes up technologies used in the app, while the Tech Lead defines development tasks for the Developer.

The Developer then takes each task and writes a human-readable description of what needs to be done. Code Monkey implements changes based on this description. Reviewer reviews every step of the task, sending it back to Code Monkey if necessary. Troubleshooter helps provide feedback when something goes wrong, while Debugger handles debugging issues as they arise.

Technical Writer generates documentation for the project, and GPT Pilot works with the developer to create a fully working production-ready app.

Contributing

If you're interested in contributing to GPT Pilot, join our Discord server, check out open GitHub issues, and see if anything interests you. We'd be happy to help resolve any of those issues. The best place to start is by reviewing blog posts mentioned above to understand how the architecture works before diving into the codebase.

Start your swift app development journey today with GPT Pilot!