Technology · Agentic → Memory · Open Source · wiki:deep

mem0

Taxonomy

Mem0 (“mem-zero”) is a memory layer for AI assistants and agents: multi-level User / Session / Agent memory with search/add APIs, Python and JS SDKs, optional self-hosted server, and a managed Cloud platform. Default stack uses an LLM plus embeddings (OpenAI defaults documented; other providers supported).

README (2026) describes a new memory algorithm (single-pass ADD-only extraction, entity linking, multi-signal retrieval, temporal reasoning) with published LoCoMo / LongMemEval / BEAM scores — managed platform numbers; OSS users should expect directionally similar, not identical, results. Apache-2.0.

RepoDocsStack: agentic-memory

Why it matters here

Cross-framework agents need persistent, searchable preferences and facts without each framework inventing storage. Mem0 is the research default for a drop-in memory API (library → self-host → Cloud) integrable with LangGraph, CrewAI, etc. Prefer LangMem when already all-in on LangGraph stores; prefer Zep/Graphiti for temporal knowledge graphs; prefer Letta for full stateful agent OS.

How it works

  1. Choose deployment: pip install mem0ai library, docker compose self-host, or Cloud.
  2. Memory.search retrieves relevant memories for a user/agent; inject into the system prompt.
  3. After the turn, Memory.add writes new memories from the conversation.
  4. Optional NLP extras enable hybrid BM25 + entity boosting; CLI/mem0 init --agent for agent signup flows.

Related: langmem · letta · zep · topics/31-memory-architecture

Flow

Scroll inside the canvas to pan

When to reach for it

  • Use when: you need a framework-agnostic memory API with OSS and hosted paths.
  • Skip when: LangGraph-native tools suffice (langmem), you need Zep’s graph/timeline product, or MemGPT-style agent harness (letta).
  • Prefer instead: self-host when tenancy/compliance blocks Cloud; Cloud when zero-ops wins.

Limits

  • Platform vs OSS gap: benchmark tables are managed-platform; don’t cite them as OSS guarantees.
  • ADD-only algorithm change: migration guides matter — old UPDATE/DELETE assumptions break.
  • LLM/embed defaults: quality and cost track provider choice; pin models in installs.
  • Not the ledger: remembered chat facts ≠ declared company truth.

Linked from

Catalog backlinks — what points here (wiki “what links here”).

Features

Stacks

Peers

What we checked

Claims below are backed by science sources on disk.

Product / memory layer

mem0ai/mem0 — project README / research highlights · MODERATE

“Drop-in memory infrastructure for AI agents and apps. Context that persists. Built for production.”

Research paper

Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory · STRONG

“We introduce Mem0, a scalable memory-centric architecture that addresses this issue by dynamically extracting, consolidating, and retrieving salient information from ongoing conversations.”

Wave / landscape note

Wave-9 note — Mem0 as memory layer (not skill registry) · ANECDOTAL

“Mem0 Platform is the fully managed memory layer for your AI apps and agents.”

In this research stack

Features and peers linked from the catalog map — not a second product surface.

Features it supports

Same cell

Stack

Document management

No DM vendor crosswalk edges yet.

Source

Primary repo github.com/mem0ai/mem0 · Open Source

technologies/mem0/README.md

Research inventory

9 tags · 13 out · 14 in · 3 artifacts · 1 gaps · 37 corpus docs

Catalog tags

landscape.layer
Agentic
landscape.subcategory
Memory
license_tag
Open Source
maps.dm
absent
maps.features
10
one_liner
Agentic
review.depth
science
slug
mem0
title
mem0

Artifacts

  • dm_map · absent
  • features_map · present · technologies/mem0/features.md
  • readme · present · technologies/mem0/README.md

Out · alternative_to

Out · maps_to

In · alternative_to

In · in_stack

In · maps_to

Corpus tags

category
Agentic → Memory
dedication
open-source
feature
job-assistant
job-support
learning-adaptation
memory-state-persistence
os-company-model-memory
r2-10-sleep-improvement
r2-31-memory-architecture
r2-35-context-engineering
r2-36-importance-consolidation
wt-data-context
kind
company_review
interface_notes
map_edge
tech_features
tech_quote
tech_readme
tech_science_source
tech_section
needs_deepen
false
quality
ok
section
Evidence
Features map
GitHub map
How it works
Limits & failure modes
Links
Scientific notes
What it is
When to use / skip
Why it matters here
slug
mem0
source_id
arxiv-2504-19413
github-readme-mem0
wave9-mem0-note
surface
interface-review
yc-capture
technology
mem0

Corpus documents (37)

company_review · 2

  • Mem0 · interface review
  • Mem0

interface_notes · 1

  • YC capture · mem0

map_edge · 10

  • mem0 → job-assistant
  • mem0 → job-support
  • mem0 → learning-adaptation
  • mem0 → memory-state-persistence
  • mem0 → os-company-model-memory
  • mem0 → r2-10-sleep-improvement
  • mem0 → r2-31-memory-architecture
  • mem0 → r2-35-context-engineering
  • … +2 more

tech_features · 1

  • mem0 · features

tech_quote · 9

  • mem0 · arxiv-2504-19413
  • mem0 · arxiv-2504-19413
  • mem0 · arxiv-2504-19413
  • mem0 · github-readme-mem0
  • mem0 · github-readme-mem0
  • mem0 · github-readme-mem0
  • mem0 · wave9-mem0-note
  • mem0 · wave9-mem0-note
  • … +1 more

tech_readme · 1

  • mem0

tech_science_source · 3

  • mem0 · arxiv-2504-19413
  • mem0 · github-readme-mem0
  • mem0 · wave9-mem0-note

tech_section · 10

  • mem0 · Evidence
  • mem0 · Features map
  • mem0 · GitHub map
  • mem0 · How it works
  • mem0 · Limits & failure modes
  • mem0 · Links
  • mem0 · Scientific notes
  • mem0 · What it is
  • … +2 more