Look into using Mozilla's [llamafile](https://github.com/mozilla-ai/llamafile) as a zero-dependency path to run Handler's embedded agent on [gemma4](https://ollama.com/library/gemma4). Right now, users have to run an Ollama server locally, and pull down gemma4 before they can interact with the embedded agent.
Look into using Mozilla's llamafile as a zero-dependency path to run Handler's embedded agent on gemma4.
Right now, users have to run an Ollama server locally, and pull down gemma4 before they can interact with the embedded agent.