- Meet Mind-Bus — a persistent AI agent platform with memory, RAG/CAG, and Adaptive Context Compression (ACC).
- Designed for long-term intelligent systems that remember, retrieve, and self-improve.
- Production-ready architecture with modular agent orchestration and scalable deployment.
Mind-Bus is a persistent AI agent platform designed to operate in long-term contexts, learn continuously, and integrate modern AI technologies in a production-ready environment.
It scales from a local AI assistant to a full enterprise-grade intelligent system.
- Persistent AI agent with long-term and short-term memory
- Hybrid RAG + CAG knowledge retrieval system
- Adaptive Context Compression (ACC) for long conversations
- Self-correcting learning from user feedback and system failures
- Modular agent orchestration and tool integration
- Supports local and cloud LLMs
- Scalable production deployment
- Fully open-source and self-hostable
- Built for modern AI infrastructure and future integrations
Mind-Bus is designed to function as a long-term intelligent system rather than a temporary chatbot.
Live demo coming soon.
- Persistent conversational AI agent
- Long-term semantic memory
- Episodic and correction memory
- Hybrid RAG knowledge retrieval
- CAG cached context system
- Adaptive Context Compression (ACC)
- Autonomous task execution
- Tool integration (web, files, APIs, databases)
- Self-correcting learning system
- Feedback-driven improvements
- Modular architecture
- Production deployment support
- Monitoring and observability
- Security and audit logging
More detailed documentation coming soon.
You can run Mind-Bus locally or on your own infrastructure.
- Python 3.10+
- Docker
- Postgres
- Redis
- Qdrant
git clone https://github.com/Iro96/Mind-Bus.git
cd Mind-Bus
docker-compose up --buildThen open:
http://localhost:8000
Full setup guide will be available in the documentation.
Mind-Bus is built using a modular AI agent architecture.
Core components:
- API Server (FastAPI)
- Agent Orchestrator (LangGraph)
- Memory System
- Retrieval System (Qdrant)
- Adaptive Context Compression (ACC)
- Worker Pipeline
- Self-learning Reflection Engine
- Deployment Infrastructure
The system is designed for scalability, reliability, and long-term learning.
Mind-Bus can be deployed as:
- Local AI system
- Private cloud agent
- Enterprise knowledge platform
- Autonomous AI infrastructure
- Hybrid cloud AI system
Enterprise features may include:
- team memory
- secure deployments
- private model hosting
- monitoring dashboards
- access control
- multi-agent orchestration
More information coming soon.
Yes. Mind-Bus is fully self-hostable and can run on your local machine or server.
No. It can run with API-based LLMs or local CPU models. GPU is optional for local models.
Mind-Bus is model-agnostic and supports:
- local LLMs
- cloud LLMs
- open-source models
- custom AI models
Yes. The system includes a self-correcting memory and reflection pipeline that allows continuous improvement while maintaining safety and auditability.
Made with contrib.rocks.
Mind-Bus is open source and community-driven.
- Build a next-generation AI agent platform
- Work with modern AI technologies
- Gain experience in large-scale AI systems
- Help shape the future of persistent AI
You can help by:
- building new features
- improving architecture
- writing documentation
- fixing bugs
- suggesting ideas
Good first issues and contribution guidelines will be added soon.
- Core agent system
- Memory architecture
- RAG + CAG integration
- ACC context compression
- Self-learning system
- Deployment infrastructure
- Monitoring and evaluation
- Multi-agent support
- Enterprise features