GitHub - NevaMind-AI/memU: Memory for 24/7 proactive agents like moltbot (clawdbot).
2 months ago
- #AI
- #Memory Framework
- #Proactive Agents
- memU is a memory framework designed for 24/7 proactive agents, reducing LLM token costs for always-on agents.
- It continuously captures and understands user intent, enabling agents to act proactively without explicit commands.
- Key capabilities include 24/7 operation, user intention capture, and cost efficiency by caching insights.
- memU features a three-layer system for reactive queries and proactive context loading, enhancing agent performance.
- The framework supports custom LLM and embedding providers, including OpenRouter for multiple LLM access.
- Proactive benefits include auto-categorization, pattern detection, and context prediction for anticipatory actions.
- memU supports real-time processing and immediate memory updates, with both RAG and LLM-based retrieval methods.
- It is suitable for various applications, including personal AI assistants, customer support, and research assistants.
- The framework is open for contributions, with detailed setup and contribution guidelines provided.