Kora Systems has unveiled a Linux-based desktop operating system designed around AI models and autonomous agents, with a planned launch in 2027. The company is led by Kirt McMaster, who previously co-founded Cyanogen, an effort to build an alternative Android operating system. Kora is targeting people who use AI heavily in their everyday work rather than trying to replace Windows for every PC user. Its software aims to let models and agents operate across applications, access files and retain context between tasks. The company says the operating system will support existing Intel- and AMD-powered PCs and computers based on Qualcomm Snapdragon chips.
An operating system designed around agents
The core idea is to make the operating system an active environment for AI agents rather than limiting them to a browser tab or an individual application. A user could ask an agent to work across documents, retrieve information from local files and carry out a multi-step task that spans several programs. Kora argues that browsers do not provide agents with enough access to a computer for more complex workflows. An operating-system-level approach could give agents a broader view of the user’s environment, but that capability also raises security concerns. A system that can read files and interact with applications needs strict permission boundaries and clear visibility into what the agent is doing.
Kora plans to isolate agents in separate workspaces with restricted access to files and other resources. That separation is important because an agent should not automatically inherit the full permissions of the person using the computer. A user might allow an assistant to work on a project folder without granting access to personal documents, credentials or unrelated business data. Sandboxing can reduce the damage caused by mistakes or malicious instructions, but it must be implemented carefully. The operating system will need to manage permissions, network access, application interaction and the transfer of information between an agent workspace and the user’s normal desktop.
Isolation and permissions are central to the design
The operating system is based on Linux and uses Nix, with the company moving away from the Fedora Silverblue foundation used by its earlier GameOS project. Kora’s own technology, including its Cortex AI engine and desktop interface, will remain proprietary even though Linux and Nix are open source. The company says it intends to support models and services from different providers. That flexibility could help users avoid being locked into one model vendor, but the experience will depend on how consistently those models work with the same tools and permissions. Local models may handle some tasks, while cloud services can be used when more computing power is required.
Kora is being developed by a team of about 50 engineers, and McMaster told TechCrunch that AI coding tools such as OpenAI Codex and Anthropic Claude have accelerated development by at least fivefold. The company has also worked with Qualcomm for nearly a year to optimise the system for Snapdragon-powered PCs, including Snapdragon X2 Elite chips. A separate enterprise version, Humain OS, is being developed with Saudi AI company Humain and is expected to run on Snapdragon laptops in 2027. These relationships may help Kora reach hardware and enterprise customers, although the company has not yet named PC manufacturers that will ship the system.
The 2027 launch faces a difficult desktop market
The commercial plan combines a free-to-use desktop operating system with subscriptions that bundle access to AI models and agent services. That model could lower the barrier to trying the OS while creating recurring revenue from users who want integrated AI capabilities. It will also need to compete with the convenience and application compatibility of Windows, macOS and established Linux distributions. Kora’s previous experience with Cyanogen provides a cautionary lesson: McMaster says dependence on the platform being challenged can undermine an alternative ecosystem. The company’s long-term prospects will depend on application support, hardware stability, privacy protections and whether its agent-first interface solves real workflow problems better than software running on existing operating systems.