Agent memory is the persistent layer that helps an AI system remember user preferences, prior decisions, and relevant context across sessions.
Agent memory is not the same as a context window or a retrieval index. It is the mechanism that stores what an agent should remember later and makes that information available again when it matters.
In practice, this usually includes preferences, instructions, goals, and selected facts tied to a project, user, or session.
Without memory, agents feel stateless. They can answer the current question but fail to build a relationship, preserve continuity, or improve the user experience over time.
That is why agent memory is one of the first infrastructure layers teams add when moving from demo to product.
Use these glossary pages and commercial landing pages to move from definition to implementation.