Short answer
Agent memory is an AI agent's ability to retain and reuse information across turns and conversations, so it remembers context, history, and prior outcomes instead of starting fresh each time. It typically combines short-term memory within a session and longer-term memory stored and retrieved across sessions.It matters because resolution depends on context. An agent that remembers a customer's plan, prior tickets, and recent order can resolve an issue in one pass instead of asking the customer to repeat themselves. Memory is what makes an agent feel continuous rather than amnesiac.
Agent memory is a general AI concept, not an Aide-owned term. In Aide, the AI platform for customer service, the relevant context is supplied by the Customer Context Engine, which pulls history and signals from connected systems such as Shopify, WooCommerce, and Salesforce. The aim is durable, accurate context, not a black box of accumulated state.
What the agent recalls and acts on stays traceable rather than hidden, and the same context is surfaced to people, so the team's shared understanding of the customer stays complete instead of being locked inside the agent.