Odysseus AI: Windows setup with Ollama

This tutorial walks through a native Windows installation of Odysseus AI, identified in the metadata as PewDiePie's tool. The presenter installs Git and Python, demonstrates Python 3.14, and opens Command Prompt in a dedicated folder. Although the speaker says Docker is recommended, this walkthrough uses the native one-command launcher.

The launcher clones the repository, creates a virtual environment and installs dependencies, then asks for login credentials. The presenter opens the local server address in a browser and signs in. Chat requires a model connection; the speaker names Ollama and OpenRouter, then proceeds with Ollama to run models locally. OpenRouter is an optional hosted alternative.

The local LLM setup uses Ollama for Windows and a 0.8-billion-parameter model, called "Queen 3.5" in the transcript. The presenter downloads it with an Ollama pull command, scans for the server in Odysseus and adds the model to chat. A short greeting receives a response. The speaker says it can work offline, but does not demonstrate a disconnected session and expresses uncertainty about internet use. Slow responses are attributed to system resource use.

The final section creates start.bat in the installation directory and a desktop shortcut. The supplied script activates the virtual environment and starts Uvicorn at 127.0.0.1:7000 for later sessions; Ollama also needs to be running.