Cross-Session Memory
The capability of an AI assistant to retain user information across separate conversation sessions. It typically encompasses short-term context memory, long-term explicit memory, and implicit preference memory, and involves trade-offs between privacy protection and personalized experience.
Timeline (last 90 days)
Even if the context window expands to millions of tokens, cross-session memory remains an engineering challenge, usually requiring an external vector database or a structured memory layer to achieve.
Cross-session persistence requires the system to uniquely identify user or problem entities; otherwise, historical memory cannot be shared.
Cross-session memory systems are typically divided into several layers: short-term context memory, long-term explicit memory, and implicit preference memory.
All Facts (3)
Even if the context window expands to millions of tokens, cross-session memory remains an engineering challenge, usually requiring an external vector database or a structured memory layer to achieve.
50%UnverifiedCross-session persistence requires the system to uniquely identify user or problem entities; otherwise, historical memory cannot be shared.
50%UnverifiedCross-session memory systems are typically divided into several layers: short-term context memory, long-term explicit memory, and implicit preference memory.
50%