← Back to Glossary
Agent Memory
Definition
Agent memory is a mechanism that allows an AI agent to store, retrieve, and use information from past interactions or executions across multiple inference steps or sessions.
Purpose
The purpose of agent memory is to enable continuity, statefulness, and long-term task execution by allowing an agent to reference prior context beyond a single prompt or context window.
Key Characteristics
- Persistence of information across multiple agent executions
- Selective storage and retrieval based on relevance or priority
- Separation between short-term context and long-term memory
- Ability to update, overwrite, or forget stored information
- Controlled access to memory to prevent leakage or misuse
Usage in Practice
In practice, agent memory is used to maintain task state, remember prior decisions, store intermediate results, track user preferences, or enable long-running workflows that cannot be completed within a single model invocation.
One implementation of this concept is offered by Kenaz through the Semantic Engineering service.