Conversation Memory

The mechanism by which an AI agent retains context from prior interactions with the same user or about the same topic across separate sessions. Short-term memory (within a conversation) is handled by the context window. Long-term conversation memory uses external storage—databases, vector stores, or structured files—to persist facts, preferences, and history. Effective memory makes agents feel personalized: they remember your name, past issues, preferences, and ongoing projects without you repeating context.