Hi all, I’m working on a research project exploring how stateless LLM-based chatbots handle long conversations and whether important earlier information is still reliably retained over time. My idea is to: Run a chatbot using an LLM API without any external memory system Introduce key facts early in
