Player not loading? Watch on YouTube
This tutorial adds Open WebUI to an existing Gemma 4 installation in Ollama. Jamie describes the open source interface as a way to give a local LLM document collections and saved personas alongside chat. Ollama and Gemma 4 must already be installed.
The setup uses Docker Desktop and a single container command, then opens localhost:3000 in a browser. Jamie demonstrates Windows setup, explains that Docker may require WSL 2 and a restart, and also gives Mac instructions. After creating a local account, Open WebUI detects his running Ollama instance.
The document example uses three school policy PDFs. Jamie creates a knowledge base and attaches it to fresh chats through the attachment menu or # shortcut. He explains that Open WebUI chunks and indexes documents, then retrieves relevant passages for the model. He repeats a question with Wi-Fi disabled and reports that it works offline. His privacy claims describe this demonstrated setup; adding a web page requires a connection.
An image description example follows, then a saved email-writing persona built with a system prompt. The demos use Gemma 4 E4B, which Jamie says runs well with 8 to 16 GB of RAM. He recommends trying the 26B mixture-of-experts model with at least 16 GB of RAM and describes an approximately 18 GB download. Gemma 4 is absent from the model list shown, so he explains how to pull it through Ollama in the terminal.
