Skip to main content

​
Welcome to Remem

Remem is a universal context layer that gives your AI agents persistent memory. Ingest personal and professional data, and Remem handles encryption, classification, and retrieval — so your agents always have the right context.

Quickstart

Create a tenant, ingest a document, and query it in 5 minutes.

Authentication

API keys, namespace grants, and security model.

Namespaces

Organize one workspace into isolated memory areas for agents and teams.

Core Concepts

Workspace hierarchy, encryption, query modes, and the processing pipeline.

MCP Integration

Connect Remem MCP tools to Claude, Codex, and other MCP clients.

CLI

Query and inspect Remem from terminal scripts and automation.

Agent Toolkit

Install remem-dev-sessions for Claude and Codex session memory.

​
Why Remem?

FeatureDescription
Hard tenant isolationPer-tenant encryption keys, dedicated vector collections, PostgreSQL RLS
Namespace isolation inside a workspacePer-key read/write grants and default namespace routing
AI-powered classificationAutomatic categorization, entity extraction, and sensitivity labeling
Two query modesFast (<100ms, hybrid search) and Rich (<2s, with LLM synthesis)
GDPR/CCPA readyBuilt-in data export and crypto-shredding deletion
MCP nativeFirst-class Model Context Protocol support for agent integrations

​
Architecture at a Glance

Client → FastAPI → Redis Streams → Worker Pipeline
                                      ├── Encrypt & Store (S3 + PostgreSQL)
                                      ├── Chunk & Embed (voyage-3.5-lite)
                                      ├── Classify & Extract (Grok 4 Fast)
                                      └── Index (Qdrant)