Microsoft Is Turning Windows 11 Into a System for AI Agents
Microsoft will keep Windows 11 as the base for turning Windows into an agentic OS, rather than shifting to Windows 12. The company described the plan at a Windows and Surface event on Oct. 7, 2026, focused on AI agents that can do tasks such as writing code. The Verge
The event started at 10 a.m. PT / 1 p.m. ET on Oct. 7. It was Microsoft's first Windows and Surface event in well over two years. BGR Microsoft gathered the announcements in a "Windows and Surface October 2026 news" roundup on its news site.
Windows chief Pavan Davuluri appeared onstage to explain agents and hybrid intelligence for Windows users. CEO Satya Nadella also spoke, referring to the "frontier ecosystem taking slightly different contours."
Hybrid intelligence is Microsoft's name for using free local models where possible instead of expensive cloud models. It includes Local Copilot AI, part of the fall Surface and Windows announcements. PCMag A local model runs on the PC itself, while a cloud model runs in a distant data center and charges per use.
For developers, Microsoft showed GitHub Copilot writing code locally without spending expensive cloud tokens. Routine code completion and agent steps use local inference on the device, with cloud fallback only when needed. Tokens are the small chunks of text cloud services measure for billing, so keeping everyday work local changes the cost equation.
Microsoft also showed agents running on Windows, including OpenClaw, Hermes, and Perplexity for Windows. The company wants Windows to serve as a platform where those agents handle tasks such as writing computer code. Reuters
Windows Search is being redesigned to look more modern and run faster, with results shown in-line so Search works more like a launcher. Search becomes a starting point for files, apps, settings, and agents, rather than a separate full-screen detour.
Microsoft listed Microsoft Execution Containers among its Windows and Surface October 2026 news topics. Agents that browse, install software, run code, and touch the file system need a sandbox with clear limits.
RTX Spark hardware for local models
The hardware anchor is Surface Laptop Ultra. It runs on Nvidia RTX Spark chips and can run AI models exceeding 120 billion parameters locally. Parameters are a rough measure of model size and capability. It starts at $2,599. The Verge
Microsoft opened preorders for what it called its most powerful Surface devices ever. Microsoft describes Surface devices as premium Windows laptops, computers, and 2-in-1 tablet PCs built for performance, flexibility and portability.
RTX Spark is not limited to Surface. Asus, Dell, HP, Lenovo, and MSI also have RTX Spark laptops.
What this changes for builders and IT
The broader context here is familiar to developers who have built with cloud AI services. Cloud tokens are quick for early tests and costly at scale. Response times vary. Data leaves the device. Costs grow when agents repeat steps to finish a job.
In my view, the durable part is the platform approach. Microsoft is treating agents as normal parts of Windows, with Search as launcher, Copilot as local helper, Execution Containers for isolation, and OEM hardware as the baseline. If that stack stays stable and well documented, it could open up everyday business automation that never fit per-seat cloud AI prices.
Worth flagging for IT teams is that controls will shape rollout. Agent permission models, audit trails, and update behavior will matter more than parameter counts. Teams will need to know what a Perplexity or OpenClaw agent can do on its own, how its actions are logged, and how local models are patched. The October event sets direction. The management tools will set pace.
