MemU - An Open-Source AI Memory Framework for AI Emotional Companionship

What is MemU

MemU is an open-source AI memory framework for AI emotional companionship. MemU can remember every conversation between the user and the AI, extract key points, and build a knowledge graph, enabling the AI to truly understand the user. MemU's memory evolves autonomously, continuously optimizing with use. MemU supports rapid integration—just a few lines of code give the AI persistent memory. Compared with other frameworks, MemU achieves up to 92% accuracy, reduces costs by 90%, and delivers retrieval speeds as fast as 50 milliseconds. In emotional companionship, education, and other scenarios, MemU can become the user's thoughtful intelligent companion.

Main Features of MemU

Memory Storage and Management: MemU stores conversations as "self-growing" folders, persisting memory and evolving autonomously over time, ensuring information retention and continuous optimization.
Memory Extraction and Retrieval: MemU provides high accuracy (95% single-hop, 88% multi-hop) and fast retrieval (50 milliseconds), supporting semantic, hybrid, and contextual retrieval strategies to efficiently extract information.
Memory Understanding and Evolution: Automatically extracts key points from conversations, builds a knowledge graph, and can analyze memories offline to generate new insights, continuously optimizing the knowledge base.
Personalized Experience: Builds personalized profiles based on user conversations and behavior, providing deep emotional companionship, remembering stories and preferences, and thoughtfully supporting user needs.
Integration and Extension: Integrates through simple code, is compatible with mainstream platforms such as OpenAI and Anthropic, and supports enterprise-level customized development and intelligent analysis tools.

MemU Project Address

Project Website：https://memu.pro/
GitHub Repository：https://github.com/NevaMind-AI/memU

How to Use MemU

Create an Account and Get an API Key
- Visit the Official Website: Go to the MemU official website https://memu.pro/。
- Register an Account: Register a free account on the official website.
- Get an API Key: After logging in to your account, visit the API key page to generate your dedicated API key.
Install the MemU Library: In your code environment, use the following command to install the MemU Python library:

pip install memu-py

Integrate MemU into Your Code: The following is a simple code example showing how to integrate MemU into a project:

import os
from memu import MemUClient

# 初始化 MemU 客户端
memu_client = MemUClient(
    base_url="https://api.memu.so",
    api_key=os.getenv("MEMU_API_KEY")  # 从环境变量中获取 API 密钥
)

# 准备对话内容
conversation_text = """
User: 你好，MemU！
Assistant: 你好！有什么可以帮你的吗？
User: 我最近在学习编程，你能给我一些建议吗？
Assistant: 当然可以！编程是一个很棒的技能。你可以从 Python 开始，它非常适合初学者。
"""

# 将对话内容传入 MemU，让其开始记忆
memu_client.memorize_conversation(
    conversation=conversation_text,
    user_id="user001",  # 用户 ID
    user_name="User",  # 用户名称
    agent_id="assistant001",  # AI 助手 ID
    agent_name="Assistant"  # AI 助手名称
)

print("对话已成功记录到 MemU！")

View and Manage Memories: After integration is complete, on the MemU management page view and manage all conversations and memories remembered by the AI.

MemU Application Scenarios

AI Emotional Companionship: MemU can record the user's emotional journey and preferences, providing thoughtful emotional support, making the AI a companion more caring than a lover and more understanding than family.
AI Education: Provides personalized learning plans and tutoring for educational applications, helping students learn more efficiently.
AI Therapy: In medical and psychological treatment, helps record patients' medical history and treatment progress, assisting doctors and therapists in making more accurate diagnoses and treatments.
AI Robots: MemU enables AI robots to remember interactions with users and environmental information, providing smarter, more personalized services and interactive experiences.
Intelligent Customer Service: With MemU, intelligent customer service systems can record customers' historical issues and preferences, enabling more coherent multi-turn conversations and providing more efficient and personalized customer service.