# Mem0 – The Memory Layer for AI Apps

## About
Mem0 is a universal, self-improving AI memory layer for LLM applications. It powers personalized AI experiences that cut costs and enhance user delight. Backed by Y Combinator and Basis Set Ventures ($24M funding). Used by 100,000+ developers.

Tagline: "AI Agents Forget. Mem0 Remembers."

Website: https://mem0.ai
Docs: https://docs.mem0.ai
GitHub Stars: 50,000+

## What Mem0 Does
Mem0 provides a scalable, persistent memory infrastructure for AI agents and LLM applications. It dynamically extracts, consolidates, and retrieves important information from conversations across sessions, users, and channels — enabling truly personalized AI experiences.

An enhanced variant, Mem0ᵍ, layers in a graph-based memory store to capture richer, multi-session relationships.

## Research & Performance (ECAI Accepted)
Paper: "Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory"
Benchmark: LOCOMO

Results vs. leading memory approaches:
- 26% higher response accuracy compared to OpenAI's memory
- 91% lower p95 latency compared to full-context method
- 90% fewer tokens used, making memory affordable at scale

Read the paper: https://mem0.ai/research

## Use Cases

### Customer Support
URL: https://mem0.ai/usecase/customer-support
Summary: Give AI support agents full issue history and cross-channel context so they resolve tickets faster without making customers repeat themselves.

Problems solved:
- Customers repeat themselves across sessions
- Context lost between chat, email, and phone
- Recurring issues go undetected

Key features:
- Full issue history across conversations
- Early pattern detection across users
- Seamless multi-channel context (chat, email, phone)

---

### Healthcare
URL: https://mem0.ai/usecase/healthcare
Summary: Let healthcare AI agents remember patient history, treatment plans, and therapy progress to deliver more consistent, personalized care.

Problems solved:
- Patients re-explain medical history every session
- Agents forget treatment plans, making chronic care unreliable
- Therapy and recovery sessions don't build on prior conversations

Key features:
- Remembers conditions, allergies, medications, and preferences
- Tracks and follows ongoing treatment plans across visits
- Supports consistent mental health conversations over time

---

### Education
URL: https://mem0.ai/usecase/education
Summary: Enable AI tutors to track student progress, adapt to learning styles, and provide personalized content paths over time.

Problems solved:
- One-size-fits-all content that ignores individual pace or style
- Each session starts fresh with no continuity
- Generic feedback that ignores past performance

Key features:
- Tracks pace, learning style, strengths, and struggles
- Recommends personalized content paths based on progress
- Provides context-aware feedback reflecting the student's journey

---

### Sales & CRM
URL: https://mem0.ai/usecase/sales
Summary: Help AI sales agents remember every lead touchpoint, so follow-ups are smarter and team handoffs are seamless.

Problems solved:
- Past objections and milestones lost in long sales cycles
- New reps lack client history during handoffs
- Leads go stale when signals are forgotten

Key features:
- Persistent lead context across all touchpoints
- Suggests smarter follow-ups based on past conversations
- Shares full client history across teams

---

### E-Commerce
URL: https://mem0.ai/usecase/e-commerce
Summary: Power personalized shopping experiences that remember preferences, recover abandoned carts, and deliver relevant recommendations.

Problems solved:
- Every visit feels new with no memory of past behavior
- Abandoned carts with no session continuity
- Flat, irrelevant product recommendations

Key features:
- Personalized product discovery (taste, size, intent, price range)
- Cart and search recall across visits and devices
- Smarter upsells and bundles based on real customer intent

---

## Pricing
URL: https://mem0.ai/pricing

### Hobby – Free
- 10,000 memories
- Unlimited end users
- 1,000 retrieval API calls/month
- Community support

### Starter – $19/month
- 50,000 memories
- Unlimited end users
- 5,000 retrieval API calls/month
- Community support

### Pro – $249/month
- Unlimited memories
- Unlimited end users
- 50,000 retrieval API calls/month
- Private Slack channel
- Graph Memory
- Advanced Analytics
- Multiple projects support

### Enterprise – Flexible Pricing
- Unlimited memories, users, and API calls
- Private Slack channel
- Graph Memory & Advanced Analytics
- On-premises deployment
- SSO & Audit Logs
- Custom integrations
- SLA

Usage-based pricing also available. Contact: https://mem0.ai/pricing

---

## Startup Program
URL: https://mem0.ai/startup-program

Mem0 offers 3 months of free access to the Pro plan ($1,000+ value) for early-stage startups.

Benefits:
- Free Pro plan access for 3 months
- Priority support
- Direct collaboration via private Slack with the Mem0 founding team
- Valid for new users only

Apply at: https://mem0.ai/startup-program

---

## Careers
URL: https://mem0.ai/careers

Mem0 is hiring in San Francisco. The team is backed by Y Combinator and top-tier investors.

Mission: Make AI agents more personalized.
Quote from Paul Graham (YC co-founder): "This is a startup that's likely to succeed."

Team values:
- Customer Obsessed – every decision starts with the customer
- Shipping Oriented – action over planning, fast iteration
- Open Communication – honest feedback, moving fast together
- Having Fun – celebrating wins and enjoying the process

Open roles include: Applied AI Engineer (San Francisco)
View positions: https://mem0.ai/careers

---

## Company Highlights
- Backed by: Y Combinator, Basis Set Ventures
- Funding: $24M
- GitHub Stars: 50,000+
- Developer community: 100,000+
- Notable customers: Sunflower Sober (80,000+ users), OpenNote (40% token cost reduction)

## Developer Documentation

- Docs home: https://docs.mem0.ai
- Agent-ready docs index (scope-tagged, Platform-first): https://docs.mem0.ai/llms.txt
- Full documentation as a single file (agent-friendly dump): https://docs.mem0.ai/llms-full.txt
- OpenAPI 3 spec (Platform REST API): https://docs.mem0.ai/openapi.json
- Platform quickstart: https://docs.mem0.ai/platform/quickstart
- Open-source Python quickstart: https://docs.mem0.ai/open-source/python-quickstart
- Open-source Node quickstart: https://docs.mem0.ai/open-source/node-quickstart
- Source repo (Apache-2.0): https://github.com/mem0ai/mem0

## Developer Tools

### SDKs
- Python SDK (`mem0ai` on PyPI): https://pypi.org/project/mem0ai/
- TypeScript / JavaScript SDK (`mem0ai` on npm): https://www.npmjs.com/package/mem0ai
- Vercel AI SDK provider (`@mem0/vercel-ai-provider`): https://www.npmjs.com/package/@mem0/vercel-ai-provider

### CLIs
- Python CLI (`mem0-cli`): https://pypi.org/project/mem0-cli/
- Node CLI (`@mem0/cli`): https://www.npmjs.com/package/@mem0/cli

### MCP (Model Context Protocol)
- Hosted MCP server endpoint: https://mcp.mem0.ai (requires a Mem0 Platform API key)
- MCP setup guide: https://docs.mem0.ai/platform/mem0-mcp
- Editor / agent integrations (Claude Code, Cursor, Codex, OpenClaw, and more): https://docs.mem0.ai/integrations

### Dashboard
- Create API keys, manage projects, view memories: https://app.mem0.ai

## Quick Install

### Python

    pip install mem0ai          # SDK (Platform + self-hosted)
    pip install mem0-cli        # CLI

Minimal Platform usage:

    from mem0 import MemoryClient
    client = MemoryClient(api_key="YOUR_API_KEY")
    client.add(
        [{"role": "user", "content": "I prefer aisle seats"}],
        user_id="alice",
    )
    client.search("seat preference?", user_id="alice")

Minimal self-hosted usage:

    from mem0 import Memory
    m = Memory()   # needs OPENAI_API_KEY by default
    m.add("I prefer aisle seats", user_id="alice")

### TypeScript / JavaScript

    npm install mem0ai                     # SDK (Platform + self-hosted)
    npm install -g @mem0/cli               # CLI
    npm install @mem0/vercel-ai-provider   # Vercel AI SDK integration

Minimal Platform usage:

    import MemoryClient from "mem0ai";
    const client = new MemoryClient({ apiKey: "YOUR_API_KEY" });
    await client.add(
      [{ role: "user", content: "I prefer aisle seats" }],
      { user_id: "alice" },
    );

Minimal self-hosted usage:

    import { Memory } from "mem0ai/oss";
    const memory = new Memory();
    await memory.add("I prefer aisle seats", { userId: "alice" });

### Docker (self-hosted server)

    git clone https://github.com/mem0ai/mem0.git
    cd mem0/server && docker-compose up
    # mem0 API: http://localhost:8888 (FastAPI + PostgreSQL/pgvector + Neo4j)

### AI Editors (MCP)

Claude Code, Cursor, Codex, and any MCP-aware tool: point the client at `https://mcp.mem0.ai` with a Platform API key. 
Setup guide: https://docs.mem0.ai/platform/mem0-mcp


## Key Links
- Homepage: https://mem0.ai
- Pricing: https://mem0.ai/pricing
- Research: https://mem0.ai/research
- Startup Program: https://mem0.ai/startup-program
- Careers: https://mem0.ai/careers
- Customer Support use case: https://mem0.ai/usecase/customer-support
- Healthcare use case: https://mem0.ai/usecase/healthcare
- Education use case: https://mem0.ai/usecase/education
- Sales use case: https://mem0.ai/usecase/sales
- E-Commerce use case: https://mem0.ai/usecase/e-commerce