KORA Systems launches KORA OS, claiming to be the first bottom-up operating system designed for AI PCs
KORA Systems released the KORA operating system, built on NixOS with a neural workspace and new filesystem, featuring the in-house Cortex inference engine that schedules CPU, GPU and NPU resources to run open-weight models while keeping workspace memory and context on the local device.
KORA Systems announced on November 10 the release of a new operating system called KORA OS, which it claims is the first operating system designed bottom-up specifically for AI PCs. Rather than layering an AI assistant on top of a traditional system, KORA OS integrates agentic workflows, contextual memory and local AI inference at the system layer, aiming to shift the computer from opening apps to complete a task toward expressing intent and letting AI help carry it out. At the base, KORA OS is built on NixOS and introduces a neural workspace and a new filesystem, letting users organize workspaces by task goal while the system manages files around intent, context and memory, with traditional folders and desktop apps retained but no longer at the core. On AI capability, KORA uses its own Cortex inference engine to schedule CPU, GPU and NPU resources and run open-weight models locally, with a built-in Lemma inference engine based on llama.cpp to improve inference efficiency, and the system decides whether a task should be handled in the cloud or kept on the device. On privacy, KORA says workflow memory and context remain on the desktop device of the user and do not leave the machine. To reduce the safety risks of autonomous AI operations, each workspace runs in an isolated sandbox, with applications and AI agents holding separate permissions.