Core Concepts

What is AI Agent Memory?

The mechanisms by which an AI agent retains and recalls information across sessions and tasks.

Definition

AI agent memory refers to the systems that allow an agent to persist information across interactions, sessions, and time. There are several types: short-term memory (within a single context window), episodic memory (recalled experiences from past sessions), semantic memory (general facts about the user or environment), and procedural memory (learned patterns of how to accomplish tasks). Memory is what allows agents to build on previous work and personalize behavior.

Example

A CRM agent uses memory to remember which Salesforce fields it has permission to write to, which reps own which territories, and which data quality rules were configured — so it doesn't need to be told these things every run.

AI Agent Memory vs context-window: What's the difference?

AI Agent Memory

The mechanisms by which an AI agent retains and recalls information across sessions and tasks.

context-window

Context window is what the agent can see right now in a single interaction. Memory is what persists across interactions and is retrieved when relevant.

Related terms

Back to glossary