wtf is cairn?

Cairn is a desktop app that walks you through running an AI language model privately on your own computer, instead of sending your messages to someone else's servers. It detects your hardware, installs and manages the underlying software for you, and gives you a private chat window where nothing you type leaves your machine. The app works by installing and coordinating two existing tools it does not include by itself: Ollama, which runs the AI model using your computer's GPU when available, and Open WebUI, which provides the chat interface. Both run directly on your system rather than inside Docker or a virtual machine, since Docker cannot reach the graphics chip on Apple's own silicon and Cairn wants full GPU speed. In its simple mode, Cairn checks your available memory, graphics card, and disk space, then recommends a single AI model rated with a green, yellow, or red "will it run" indicator for your specific machine, downloads it, and opens the chat app in your browser. There is also an explore mode with a wider catalog of models for chat, coding, images, and even a medical focused model, each rated for your hardware and filterable by what you want to use it for. A remote access feature lets you reach your private assistant from other devices. By default the app is only reachable from the same computer, but you can choose to open it to your local Wi-Fi network or to your own devices anywhere through Tailscale, each option clearly explained, with a QR code shown so you can connect a second device easily. Cairn is available as a prebuilt installer for macOS, Windows, and Linux, and updates itself automatically once installed. An internet connection is only needed for the first setup, since everything runs offline after that. The project is licensed for free personal, research, and nonprofit use, but commercial use requires contacting the author for a separate license, except for early releases which remain under the Apache 2.0 license.