Odysseus and Ollama: Windows setup guide

This tutorial walks through installing Odysseus on Windows and connecting it to Ollama for local AI chat. The presenter describes Odysseus as an open source, self-hosted workspace that keeps chats and files private. The walkthrough covers installation and a basic chat test; it does not independently verify those privacy claims or demonstrate the document and agent features mentioned in the introduction.

The setup starts with Git and a clone of the Odysseus repository. It uses Python 3.11, with python.exe added to PATH during installation, followed by a Python environment and the required packages. Running python setup.py prepares the application and prompts for a username and password. After starting the server, the presenter opens the dashboard at 127.0.0.1:7000.

The built-in cookbook produces model download errors in the demonstration, so the presenter uses Ollama instead. The example is Qwen 2.5:3B, described as roughly a 1.9 GB download. Odysseus finds the model through Settings and Scan for servers. Gemma 2 2B is another suggested model, with a fresh scan needed after adding it.

Closing the server terminal stops Odysseus. The final section creates an open.bat file with the installation folder path and a desktop shortcut to start the server and open the dashboard.