First memories

Memory types

Storing memories

Beyond extraction

Give a shop's support assistant long-term memory with LangMem: memory managers, the extraction request, collections, updates, profiles, episodes, the LangGraph store with Gemini embeddings, store managers, memory tools, an agent that remembers across threads, background processing, running summaries and prompt optimization. Every output is a real run on Groq's gpt-oss-120b.

LangMem

LangMem tutorial: long-term memory for AI agents in Python

LangMem overview

Install LangMem with free Groq and Gemini keys

Installation and setup

Extract memories from a conversation with create_memory_manager

Extracting memories

What a LangMem memory manager sends to the model

The extraction request

Chat models for LangMem: model strings, objects and keys

Chat models

Memory collections in LangMem: many memories that grow over time

Collections

Update and delete memories in LangMem with PatchDoc and RemoveDoc

Updates and deletes

User profiles in LangMem: one memory with fixed fields

Profiles

Episodic memory in LangMem: learn from past interactions

Episodic memory

The LangGraph store: where LangMem keeps memories

The LangGraph store

Semantic search in the LangGraph store with Gemini embeddings

Semantic search

create_memory_store_manager: extract and save memories per user

Store managers

LangMem memory tools: manage_memory and search_memory

Memory tools

An agent that remembers across conversations with LangMem

Agent memory

Background memory with LangMem's ReflectionExecutor

Background memory

Running summaries in LangMem: summarize_messages for long chats

Running summaries

Procedural memory in LangMem: optimize a prompt from feedback

Prompt optimization

LangMem integrations: persistent stores, other models and embeddings

Integrations

Build a support assistant with long-term memory using LangMem

Project: support assistant

Storing memories quiz

store.search(("memories", "asha")) with no query returns what?