Is there a free plan?

Yes! Memwyre has a generous free tier that includes up to 50 memories, 10 document uploads, 30 AI chat messages, and full IDE integration via MCP. No credit card required — just sign up and start building your memory layer.

How do I connect Memwyre to Cursor or VS Code?

You can connect Memwyre to Cursor, VS Code, or Claude Desktop using our official Model Context Protocol (MCP) server. This gives your AI assistant direct, real-time access to your entire shared memory layer.

Do you offer a browser extension?

Yes! The Memwyre browser extension (available for Chrome and Edge) lets you instantly clip text, articles, and code snippets directly into your memory layer without breaking your flow.

Is my data private and secure?

Absolutely. Your memory layer is fully private and encrypted. Your data is only accessible to you via authenticated sessions and your personal API keys.

What is the OpenClaw plugin?

The OpenClaw plugin allows autonomous AI agents to read from and write to your Memwyre memory layer. This gives agents persistent memory across runs, letting them build on past debugging sessions and research.

Why not just use ChatGPT's built-in memory?

ChatGPT memory is locked entirely inside ChatGPT. It cannot be accessed by Claude, Cursor, your terminal agents, or other LLMs. Memwyre acts as a universal, shared memory layer that works across all of these tools simultaneously.

How is this different from traditional RAG?

Traditional RAG search takes static documents, chunks them, and searches for matches. Memwyre builds active, connected semantic profiles of your project guidelines, API models, and style rules, updating them dynamically in the background.

How does Memwyre compare to Mem0 or Zep?

Mem0 and Zep are developer databases designed as backend infrastructure for teams building custom agents. Memwyre is a ready-to-use memory platform that connects directly to everyday consumer tools.

Your AI
should remember everything.

Memwyre is the AI Memory Platform that gives every AI tool, agent, and conversation a persistent, shared memory.

Early Access Open

Get Started See Plans

$npx -y install-memwyre

/ PRODUCT FEATURES

[1/9]

All the tools to make AI remember.

Every feature you need to capture, organize, and query your shared memory platform seamlessly across your workflow.

01 / OMNIPRESENT OVERLAY

Access it from any app

Invoke Memwyre instantly over any app. Access your shared memory layer across ChatGPT, Claude, Notion, or email to capture thoughts mid-conversation and sync context in a single click.

/ WHAT WE DO

[2/9]

Bring your knowledge. We build memory. Your AI just knows.

Persistent memory infrastructure for AI agents. One platform, every capability.

For Developers & Teams

The Memwyre MCP Server

State-of-the-art retrieval and memory. Semantic search, fact extraction, and context pruning. One MCP server, every AI tool connected.

<200ms

RECALL LATENCY

88.5%

TOKEN REDUCTION

73.5%

LOCOMO ACCURACY

Self-hostable · Model-agnostic · TypeScript & Python SDKs

Start building in 5 minutes

For Everyone

Personal AI Memory

One memory across everything you use. What you teach one AI, every AI remembers.

Memwyre Dashboard

Control panel

Cursor · VS Code · Claude · OpenCode

MCP plugins

Chrome Extension

One-click save

Early Accessopen for power users

Get Personal Memory

/ PLATFORM COMPATIBILITY

[3/9]

One memory layer. Every tool connected.

Memwyre is not a standalone chatbot. It is a universal context layer that plugs directly into the models, editors, and terminal tools you already use every day.

01 / IDE INTEGRATION

Native IDE Memory

Connect your IDE directly via MCP. Past bug fixes, configuration choices, and API schemas become instantly searchable context inside Cursor, VS Code, or Windsurf.

memwyre

Add from memwyre

Manage connectors

Add connector

Tool accessLoad tools when needed

02 / TERMINAL AGENTS

Code Agents

Empower terminal tools like Claude Code and OpenClaw. Agents save debug results and project conventions, building cumulative operational memory.

claude-code — terminal

*Welcome to Claude Code!

while(curious) {

question_everything();

dig_deeper();

connect_dots(unexpected);

if (stuck) {

keep_thinking();

}

*Simmering...

03 / MULTI-MODEL

Works With Every LLM

One shared memory layer across Claude, ChatGPT, and Gemini. If you explain a code pattern in ChatGPT, your Cursor agent retrieves it instantly when you ask.

/ WHY MEMWYRE

[4/9]

Context for humans, not just databases.

Mem0 and Zep are built as backend database infrastructure for AI agents. Memwyre is built as the out-of-the-box context layer for individual developers, power users, and the tools they work in daily.

FEATURE	mMem0	ZZep
Memory Graph	—
User Profiles	—	—
Document Retrieval	—	—
Connectors	—	—
Document Extractors	—	—
Self-hostable		—
Consumer Plugins	—	—

/ INSTALLER

[5/9]

60-Second Setup.

No complex onboarding or data migrations. Install the MCP server and connect your workspace instantly.

Get Started

View Docs

claude_desktop_config.json

$npx -y install-memwyre --claude

/ HOW IT WORKS

[6/9]

How Memwyre works.

Read the docs

Memwyre is the unified memory layer built for your AI workflow. Save context once, and access it automatically across every tool.

01 . SAVE ONCE

Instant

context capture

Save prompts, guidelines, or database schemas with one click. They stream directly to your shared memory layer.

mcp & browser streams

02 . AI CONNECTS

Dynamic

no manual tagging

Our platform automatically parses facts, resolves entities, and links context in the background to build a connected memory graph.

continuous background indexing

03 . REMEMBER EVERYWHERE

Instant

context injection

Query from Cursor composition, Claude Desktop, ChatGPT, or terminal. The correct context is injected without manual copy-pasting.

p99 query recall

/ BENCHMARK

[7/9]

The LoCoMo Benchmark.

1,540 questions • 5 categories

If your memory platform returns noisy facts or misses key details, your AI writes broken code. Retrieval quality dictates the quality of your AI outcomes. Here is how Memwyre's accuracy compares on the MemoryBench long-context evaluation dataset:

Read Research

70.5%

Non-Adversarial Recall

Full 1,986 Benchmark Suite

4.9k

Mean Context Tokens

Vs 26k+ Raw Context

53.0

78.6

24.0

57.5

50.0

67.7

48.0

64.2

Full Benchmark Suite (1,540 Queries)Click category to inspect

Across the complete evaluation suite, Memwyre outperforms standard vector RAG by +24.4% average recall while reducing context footprint by 81%.

Baseline RAG

Memwyre

LOCOMO BENCHMARK

/ Accuracy Metrics

RAGMemwyre

Single-hopRAG: 53.0% / Memwyre: 78.6%

Multi-hopRAG: 24.0% / Memwyre: 57.5%

AdversarialRAG: 50.0% / Memwyre: 67.7%

TemporalRAG: 48.0% / Memwyre: 64.2%

Old

New

LOCOMO

Memory vs RAG

Latency (seconds)

403020100

12s

Memory

40s

RAG

40%

Fewer tokens

Compared to RAG baseline

View on X

/ PRICING

[8/9]

AI that works where your team works.

Simple, transparent pricing for individuals and teams. Start building your persistent AI memory layer today.

01 . FREE

forever

A great way to get started with building your persistent AI memory layer.

Up to 50 memories
Up to 10 document uploads
Standard AI token budget
IDE Integration (MCP)
Community Support

Get Started for Free

02 . PRO

Recommended

$12

/ month

Everything in Free, plus unlimited storage and priority features.

Integrations

Unlimited memories & documents
Google Drive & Notion sync
Claude Code, OpenCode & OpenClaw plugins
Browser Extension & MCP IDE Server
Extended daily AI chat quota & support

Start Building Your Brain

03 . ENTERPRISE

Custom

volume pricing

Tailored deployments for teams that need dedicated infrastructure, SSO, and volume pricing.

Everything in Pro
Dedicated infrastructure & SSO
Custom integrations & SLA
Dedicated support

Contact Sales

/ BLOG

[9/9]

Latest from the Memwyre Blog

/ FAQ

[HELP & PRIVACY]

Frequently asked questions.

Memwyre is a unified AI memory platform that captures your context across tools (browser, ChatGPT, Claude, Cursor) so you never lose an insight, code snippet, or research finding. It makes this knowledge instantly searchable and reusable by any tool.

The second brain
for your AI.

Get Started Compare Plans

Your AI should remember everything.