Fixed the demo URL by allowing direct messages to the bot and using that. Also because I can DM the bot now, I added an OLLAMA model running QWEN2:5b, although because it it running in CPU only mode it is extremally slow