BEIJING, March 13, 2026 /PRNewswire/ — At the recent FOSSASIA Summit 2026 in Bangkok, Thailand, developers and open source communities from across Asia and beyond gathered to share the latest advancements in open technologies. China’s open source operating system community openKylin participated with multiple technical talks and an interactive booth, highlighting its latest work on integrating artificial intelligence into operating system architecture.
AI-Native Architecture: Redefining the Core Capabilities of Operating Systems
As large language models and multimodal AI continue to mature, operating systems are evolving from passive resource management platforms into intelligent systems capable of understanding and assisting users. openKylin 2.0 is exploring this shift through a full-stack AI approach aimed at building an AI-native operating system for the intelligent computing era.
During the summit, the openKylin technical team introduced its Linux-native AI subsystem architecture currently under development. The design treats AI as a fundamental capability of the operating system rather than an add-on at the application layer, enabling unified intelligent services for both applications and system components.
A Three-Layer Decoupled Design to Simplify AI Development
To address challenges such as diverse hardware platforms, fragmented model frameworks, and complex integration processes, openKylin proposes a three-layer architecture consisting of a Unified Inference Framework, an AI Runtime Layer, and an AI SDK Layer. This structure decouples models from hardware and applications from models, allowing developers to build AI applications without managing underlying infrastructure complexity.
Device–Cloud Collaboration with Built-in Privacy Protection
The subsystem also supports hybrid device–cloud inference. Through an AI Engine module, tasks can dynamically run either locally or in the cloud depending on computing resources, network conditions, and privacy requirements — ensuring both performance and data protection.
From “AI on OS” to “AI for OS”
Looking ahead, openKylin is promoting a shift from “AI on OS” to “AI for OS,” pursuing deeper integration between AI and operating systems while exploring technologies such as multi-agent collaboration, lightweight device-side models, and system-level AI interfaces.
Through its talks, demonstrations, and booth interactions at FOSSASIA, openKylin signaled its ambition to contribute to the global evolution of AI-native open source operating systems. More information about the distribution can also be found on DistroWatch.
