Someone finally built GLaDOS from Portal as an open source project. The GLaDOS Personality Core aims to create a physical version of the iconic AI character, complete with voice generation, vision tracking, and IoT device control that can express the AI’s intentions through connected devices.
The project runs on minimal hardware – even an 8GB single board computer can handle it. At its core, GLaDOS uses a local large language model for fast responses under 600ms. The system constantly records audio to catch voice commands, then quickly transcribes and processes them through the language model to generate natural responses.
The system connects to IoT devices throughout your space, allowing GLaDOS to control lights, appliances, and other smart devices. The vision system lets her track and turn toward people through connected cameras, adding an extra layer of eeriness to the interactions.
The installation process works on Windows, Mac, and Linux. You’ll need to install Ollama to run the language model locally, then follow the setup instructions for your operating system. The project uses Python and aims to minimize dependencies so it can run on basic hardware.
To avoid feedback loops where GLaDOS hears herself talk, you’ll want to use headphones or a conference room microphone with echo cancellation. You can also disable voice interruption in the config file if needed.
The entire project is available on GitHub under an MIT license. While it’s still experimental, the developer actively maintains it and provides support through Discord. If you want to test specific components before building the full system, check out the demo notebook included in the repository.
This project shows how far we’ve come with AI and IoT integration. What used to be science fiction is now something you can build at home with open source tools and some technical know-how. Just try not to give your GLaDOS access to deadly neurotoxins.
