Large language models have improved conversational AI by producing fluent, context-aware replies. Still, their short context windows and lack of persistent memory can break conversations: users often ...