Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Agent memory

Agent memory is the set of techniques that let an AI agent keep information beyond a single context window, such as facts about a user or past work, by storing it outside the model and loading it back when needed.

Also called: long-term memory, memory layer

By Divya Sharma · Checked Nov 15, 2026 · How we review

A language model has no memory between calls. Everything it knows about the current task must sit in its context window, and that window is limited. Agent memory works around this by saving information elsewhere.

Kinds of memory

- Short-term memory is the conversation and tool results in the current session. - Long-term memory survives across sessions: user preferences, facts learned, summaries of past work.

LangGraph, for example, separates short-term thread state from longer-term memory across sessions.

How it works

The agent, or the code around it, decides what to write down: a note in a file, a row in a database or an embedding in a vector database. Later, relevant memories are retrieved and placed back in the prompt, much like retrieval-augmented generation. The MemGPT paper framed this like an operating system moving data between fast and slow memory, with the model managing its own memory tiers.

Example

A coding agent writes a project notes file at the end of each session and reads it at the start of the next, so it does not relearn the build commands every time.

Limits

Memory can go stale or contradict itself, and stored text from untrusted sources can carry instructions into later sessions. Tools such as Mem0 and Letta offer managed memory layers; the agent memory guide compares them.

Related terms

Context engineering
Context engineering is the practice of choosing and maintaining everything that goes into a language model's context window during a task: instructions, retrieved documents, tool definitions, tool results and conversation history.
Vector database
A vector database stores embeddings alongside their source records and finds the stored vectors closest to a query vector, usually with an approximate index so search stays fast across millions of items.
Context window
A context window is the maximum amount of text, counted in tokens, that a language model can consider at once, covering the system prompt, conversation, documents, tool results and the response it writes.
Context compaction
Context compaction is the step where an agent summarises a conversation that is nearing its context window limit and continues in a fresh window that starts from that summary, so a long task can keep going.
Retrieval-augmented generation
Retrieval-augmented generation (RAG) is a pattern where a system first searches a collection of documents for passages relevant to a question, then gives those passages to a language model so its answer draws on them.

Sources