AI Memory Engine
3.9k 2026-04-13
CaviraOSS/OpenMemory
OpenMemory is a self-hosted, local-first cognitive memory engine providing real long-term memory for LLM applications and AI agents, distinct from RAG or vector databases.
Core Features
Provides real long-term, cognitive memory for LLMs and AI agents.
Self-hosted and local-first with support for SQLite and Postgres.
Offers Python and Node.js SDKs for easy integration into applications.
Integrates with popular AI frameworks like LangChain, CrewAI, and AutoGen.
Includes connectors to ingest data from sources like GitHub, Notion, and Google Drive.
Quick Start
pip install openmemory-pyDetailed Introduction
OpenMemory addresses the fundamental challenge of stateless LLMs by providing a persistent, cognitive memory engine. Unlike traditional RAG or vector databases, it aims to offer 'real' long-term memory, allowing AI agents to retain context and learn over time. It supports both embedded SDKs for local-first development and a central backend server for multi-user and organizational memory. This enables applications to overcome the 'amnesiac' problem, making AI interactions more intelligent and context-aware across sessions and users.