Memory for AI assistants, chatbots, and OpenClaw agents

Out of stealth
The world's No.1 ranking
Memory as a Service company

Ahead of the industry leaders who require you to use more expensive answer models. 10x cheaper for you per query, enabling you to build better systems for lower cost and swap between answer models as needed with no degradation in performance.

m3mory
83.6%LongMemEval-S
Hybrid graph-retrieval engineBenchmarked onLongMemEval-SOpen SourceMCP Server, SDKs

Knowledge Graph

Connections, not just embeddings.

m3mory builds an entity graph from every conversation. People, places, projects, preferences. All connected. All searchable.

person
org
project
location
tech
event
preference

How it works

Five stages. One API call.

Every recall runs through our proprietary retrieval pipeline. Multiple search strategies, fusion scoring, and intelligent context assembly.

Understand

Analyses the query to determine what kind of memory retrieval is needed

Optimise

Rewrites and expands the query for maximum retrieval accuracy

Search

Runs multiple search strategies in parallel across your knowledge graph

Score

Proprietary fusion algorithm combines results from all search paths.

Assemble

Deduplicates, formats, and trims to your token budget. Quality over quantity

Three lines of code

Simple API. Powerful retrieval.

Two endpoints: ingest and context. The pipeline handles extraction, graphing, scoring, and assembly behind the scenes.

1import { M3mory } from "@m3mory/sdk";
2
3const m3 = new M3mory({ apiKey: process.env.M3MORY_KEY });
4
5// Ingest a conversation
6await m3.ingest({
7 agentId: "assistant-1",
8 userId: "user_abc",
9 messages: [
10 { role: "user", content: "I just moved to Bristol" },
11 { role: "assistant", content: "Welcome to Bristol!" }
12 ]
13});
14// -> 2 memories extracted, 3 entity edges created

Benchmarked, not just marketed

83.6%

LongMemEval-S Benchmark

The standard measure of AI memory retrieval quality. Higher is better.

m3morygpt-4o-mini
83.6%
SupermemoryGPT-4o
81.6%
ZepGPT-4o
71.2%
Full contextGPT-4o
60.2%

Edging ahead of the industry leader. The only system on the leaderboard running gpt-4o-mini end to end including the answer model, potentially 10x savings to your LLM budget.

Pricing

Start free. Scale when ready.

No credit card required. All plans include unlimited storage, unlimited users, and the full retrieval pipeline.

Free

Free
1M tokens/month tokens
10K queries/month search queries
Unlimited storage & users
Email support
Most popular

Pro

$15/mo
4M tokens/month tokens
100K queries/month search queries
Unlimited storage & users
Priority support

Scale

$299/mo
120M tokens/month tokens
20M queries/month search queries
Unlimited storage & users
Priority support

Open Source

Built in the open.

Our MCP server, SDKs, and integrations are MIT-licensed. The retrieval engine is proprietary.

@m3mory/mcp-server

Model Context Protocol server. Connect any MCP-compatible agent to m3mory in seconds.

TypeScript|MIT

@m3mory/sdk

Official TypeScript/JavaScript SDK. Type-safe, tree-shakeable, zero dependencies.

TypeScript|MIT

@m3mory/m3moryclaw-plugin

Drop-in plugin for OpenClaw. Auto-recall before every turn, auto-capture after.

TypeScript|MIT