Agent Memory
A mechanism for an agent to store and retrieve user preferences, past tasks and key facts across sessions.
In plain language
Agent memory is usually split into short-term (current task context), long-term (preferences/facts) and working memory (intermediate results). Implementations include vector stores, structured databases, summaries and knowledge graphs. The hard part is not storage but what to write, when to forget and how to prevent bad memories from contaminating future decisions. Memory must be inspectable, deletable and correctable, and must distinguish explicit user statements from model inferences. Private memories need consent and retention limits.
Example
A writing agent remembers “no exclamation marks” and applies it next time, but should not permanently store a one-off tone request.
Recommended tools
Related terms
Agentic Workflow
Breaking a task into steps planned by AI, executed by tools and reviewed by humans.
AI Agent
An AI system that perceives a goal, plans steps, calls tools and executes tasks.
Retrieval-Augmented Generation
Retrieve relevant passages from a knowledge base, then have the model answer from them.
AI Upscaling
Using models to increase image resolution and add detail.