AivexaNewsSearch
AI news for builders and product teamsChecked every hour

After challenging Google’s Android with Cyanogen, its co-founder takes aim at Windows for AI era

Collected Nov 10, 2026

Kora Systems, the Palo Alto company led by Cyanogen co-founder Kirt McMaster, unveiled a Linux-based desktop operating system on Tuesday that puts AI models and autonomous agents at the center of the computing experience. The software is slated to launch in 2027 as a free-to-use desktop OS, running on existing Intel- and AMD-powered PCs as well as machines built around Qualcomm's Snapdragon chips. The pitch is not a general-purpose Windows replacement so much as an environment for people whose daily work leans heavily on AI.

According to McMaster, the system lets different models and agents operate across applications, open and read files, and carry context from one task to the next. A chat interface called Chord provides access to both local and frontier models. Agents can be confined to separate workspaces with restricted access to files and other resources, which is meant to contain what an autonomous process can touch. Kora also plans an AI engine called Cortex that runs tasks on the device when it can and hands heavier work to cloud-hosted models.

The company is targeting a narrower audience than Microsoft's. McMaster estimates that 45 million to 50 million people currently account for much of global AI usage and revenue, and he expects that group to grow past 200 million users by 2033. His argument is that Kora does not need to win over Windows' installed base, only the slice of users willing to pay for capable models and agentic services. That points to subscriptions bundling model and agent access as the revenue model, rather than a license fee for the OS itself.

The market context is lopsided. Windows held roughly 77% of the worldwide desktop operating-system market in October, against close to 10% for Apple's OS X, per StatCounter's web traffic data. Competing with that from a standing start has defeated far larger firms, so Kora's bet is essentially that AI-heavy workloads create a distinct category where the incumbent's scale matters less.

McMaster frames the strategy through his Cyanogen experience, where he spent more than five years. Cyanogen tried to loosen Google's grip on Android but stayed dependent on Google Mobile Services even while adding features beyond Google's version of the platform, which he says capped its ability to build an independent ecosystem. His stated lesson is that a challenger still reliant on the incumbent has no chance. That history also explains the architecture: owning the OS layer, rather than shipping an app or browser on someone else's platform, is presented as the way to give agents real reach across a machine.

The path here was not a clean start. The company was previously Playtron, a gaming-OS venture led by McMaster, and rebranded as Kora in July 2025 as it pivoted to AI. The trigger, per McMaster, was the wave of AI-powered browsers from companies including Perplexity and OpenAI; he judged that browsers would not give agents enough access for complex tasks, making the operating system the better starting point. The pivot let Kora build on nearly three years of work on Playtron's Linux-based GameOS instead of beginning from zero.

On the technical side, Kora is built on Linux and uses Nix, replacing the Fedora Silverblue base of the earlier GameOS. Nix is a package manager and build system known for reproducible, declarative configuration, which in practice makes it easier to define and roll back a machine's state — a useful property for a system that lets agents modify files and settings. The open-source foundations do not extend to Kora's own layer: McMaster says its Cortex AI engine and desktop interface remain proprietary, and the company has not committed to open-sourcing its software. It does intend to support models and services from multiple providers.

The team numbers about 50 engineers, and McMaster credits AI coding tools — OpenAI's Codex and Anthropic's Claude — with accelerating development by at least fivefold, most of that gain coming in the past six months. None of the engineers came from Cyanogen; Kora recruited heavily from the Linux gaming community. Beyond the desktop, the roadmap includes AI-powered tools for creating, publishing and playing games inside the OS, a console-like experience called GameFrame, and support for Android apps.

Hardware partnerships matter for an OS with no install base. McMaster says Kora has worked closely with Qualcomm's engineers for nearly a year to optimize for Snapdragon PCs, including support for the Snapdragon X2 Elite chips. Separately, Kora is working with Saudi Arabia's AI company Humain on an enterprise edition called Humain OS, expected to run on Snapdragon laptops in 2027. Further PC-maker partnerships are promised for later, with no names or timeline given.

Investors listed are Samsung Next, Square Enix, Alumni Ventures, Polychain, VanEck and Circle, all of which previously backed Playtron. McMaster says the company raised additional funding after the AI pivot but declined to give the total or a valuation, stating only that it has enough to operate for several years.

Several things remain unsettled. A 2027 launch leaves a long window in which Microsoft is adding agentic features to Windows and device makers are shipping their own AI layers, so the differentiation Kora is claiming may narrow. The decision to keep the core proprietary while building on open components invites the same scrutiny Cyanogen faced over control of its stack. And the plan to support Intel, AMD and Snapdragon hardware means validating across three very different chip families and their NPUs, a substantial engineering load for a 50-person team. A likely trade-off is that deep, system-level agent access delivers the productivity Kora is selling while widening the security surface, which is presumably why the restricted agent workspaces exist; whether that containment holds up under real workloads is an open question.

Why it matters: For developers and product teams, Kora is an early test of whether AI agents belong at the OS layer rather than inside browsers or apps — if agents can reach files and applications directly, integrations built as web or desktop apps may look constrained by comparison. A subscription model bundling model access also signals competition shifting from per-seat software licenses toward metered AI usage. The timeline is distant and the incumbent's position strong, so the practical near-term effect is likely to be pressure on Windows and app developers to match system-level agent access — an inference, not a stated outcome.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

Kora is set to launch as a free-to-use alternative to Windows, targeting users who rely heavily on AI models and agents.