A guide to building long-term compounding knowledge infrastructure. See GitHub for details.

memU

memU is an open-source memory framework for AI companions, offering high accuracy, fast retrieval, and low cost for personalized AI experiences.

Introduction

memU is an open-source memory framework designed for AI companions, featuring high accuracy, fast retrieval, and low cost. It supports multiple retrieval strategies and helps AI companions truly “remember” users.

Key Features

  • 92% accuracy on Locomo benchmark
  • Up to 90% cost reduction
  • Advanced retrieval: semantic, hybrid, contextual
  • 24/7 enterprise support

Use Cases

  • Building personalized AI companions
  • Intelligent memory and user profiling
  • Long-term interaction and growth AI
  • Enterprise AI companion services

Technical Highlights

  • Optimized online platform for efficiency and cost
  • Multiple retrieval and memory strategies
  • Highly adaptable for various AI companion scenarios
  • Open-source architecture, easy integration and extension

Comments

memU
Resource Info
Author NevaMind-AI
Added Date 2025-09-04
Type
Tool
Tags
Agent OSS Utility