beclab/Olares
Open-Source Personal Cloud OS for Always-On Agents
What it solves
Olares addresses the privacy and cost concerns of using third-party cloud AI services. By bringing AI "home," it allows users to run agents and large language models on their own hardware, ensuring sensitive personal data (like files and messages) stays under their control rather than being stored on external servers.
How it works
Built on Kubernetes, Olares transforms a Linux host into a self-hosted AI platform. It provides a unified environment for compute, storage, and networking that can be managed via a web browser or a command-line interface. The system includes features like GPU pooling (with time-slicing and memory-slicing), a private VPN for secure access, and a marketplace for one-click installation of open-source AI applications.
Who it’s for
- Individual users wanting to run generative AI and personal agents privately.
- Small teams looking for a unified, self-hosted workspace for automation and data management.
- Developers wanting to build, test, and host agentic applications in isolated environments.
Highlights
- One-click local AI: Easily install open-source models and apps from a dedicated market.
- Accelerated computing: Manages and pools GPUs across different nodes.
- Agent-native management: Uses a CLI with "Agent Skills" so AI agents can operate the OS.
- Comprehensive system apps: Includes built-in tools for files, storage, and secure access via LarePass.