Great post. I wonder how much this can improve if you RAG-ify a diverse set of contextual data, for example calendar, meals, recent conversations from the real world, etc.
It's also interesting that бля was translated to 'damn'. :)
It's also interesting that бля was translated to 'damn'. :)