Open WebUI and Ollama: local AI server setup on Mac

Mike Murphy walks through a self-hosted Open WebUI setup on macOS, with Ollama supplying the models. He uses Python rather than the Docker installation method he also mentions. The tutorial creates a folder in the home directory, sets up and activates a virtual environment, then installs Open WebUI with pip.

Much of the walkthrough covers everyday startup: adding a terminal alias to .zshrc, reloading that file, and using the shortcut to launch the server. Murphy opens the interface at http://localhost:8080 and creates an admin account for this installation. In his demonstration, Open WebUI automatically detects the running Ollama instance and lists its models.

The server stays attached to the terminal session. Murphy keeps that session open while chatting and uses Control-C to stop it. He also demonstrates access from an iPhone on the same Wi-Fi network by entering the Mac's IP address and port 8080 in a browser.

For viewers who want to run models locally, the tutorial explains the basic installation and access workflow. Murphy describes built-in document retrieval and optional external API connections, but does not configure those features here. He says the local model setup can work offline; his privacy claims concern that setup, while external connections are optional.