This is a memory layer infrastructure designed specifically for AI and LLM applications, aiming to provide various agents or AI applications with writable, readable and retrievable personalized'memory' capabilities. It can be easily integrated into various AI applications through APIs, thereby breaking through the context length limitations of traditional LLM and realizing the persistent storage and retrieval of user information and preferences.