Chairman and CEO at Microsoft

Report this post

Today marks a new chapter for Windows, as we bring unmetered intelligence to every desk and every home, and make every PC a place where agents can work securely on your behalf. Some highlights of what we announced: • MAI-Code-1.1 Flash: 137B parameter coding model w/ 256K context window, which is now optimized to run on your PC! • GitHub Copilot now hands off work to local models like MAI-Code-1.1 Flash, helping projects cost a lot less without sacrificing quality. • With Hybrid Intelligence, Copilot can now take action directly on the PC and keep sensitive work on your device. • And with Code in Copilot, you can essentially build any software you need on your desktop, without any cloud token spend, and it’s just super at it. You’re no longer limited to what’s in an app store! Your PC becomes an infinite software factory. • Security is foundational to all this, which is why we are also bringing together Windows and Agent 365 so agents can work within secure boundaries on-device, including MXC a local sandbox for agent execution. Windows becomes your secure agent box! • All this comes to life on a new generation of devices, like Surface Laptop Ultra, powered by NVIDIA RTX Spark. Can’t wait to see what you build with all this. Read more about what we announced: https://lnkd.in/gFaRcv_X

Copy
LinkedIn
Facebook
Report this comment

This is a significant shift in how we think about the future of work. AI agents moving directly onto PCs could change not only how software is built, but also how teams operate, collaborate and solve problems. The combination of local AI, security and Copilot makes this especially interesting. The next question is: how quickly will businesses adapt their workforce and skills to this new model?

Report this comment

Lowering cloud token costs cannot come at the expense of auditability and governance defensibility. Microsoft’s own 2026 Digital Defense Report emphasizes that AI attacks now compress intrusion-to-impact timelines from days to minutes, while overreliance on Human-in-the-Loop controls introduces systemic oversight failure. Deploying trillion-parameter models and autonomous token factories directly onto endpoint silicon decentralizes execution into local memory—stripping away central visibility and granting persistent, high-privilege access to the device. 🛑 Cost efficiency is meaningless if the resulting execution architecture is not board defensible.

Report this comment

Huge moment Satya Nadella and the team. Turning every desk into a secure, self-contained workshop where you can actually build without handing your work off to someone else's servers. Exciting to watch this unfold.

Report this comment

Satya — this is where the governance problem becomes much more concrete. Once agents can act locally, continuously, and across files, applications, credentials, and software on behalf of a user, the question is no longer just whether the sandbox is secure. It is: does the authority for this exact action still stand now? A local agent may still have capability after consent changes, policy changes, account state changes, or the underlying evidence moves. So secure execution needs two distinct layers: can the agent execute safely? is the agent still legitimately authorized to execute this consequence now? The first is containment. The second is Temporal Standing. As PCs become agent execution environments, both are going to matter.

Report this comment

I should be able to choose if I need it or not. You leave consumers with no choice I dont need this extra load on my PC and unnecessary AI that I dont use. Seriously considering moving to Linux. Windows getting out of hands

Report this comment

Is this is a complete U-turn where individuals and businesses once again wanting to have more powerful computers/laptops instead of offloading the heavy lifting to cloud!

Report this comment

A lot to like here. Hybrid intelligence also means we'll need a much better understanding of what happens locally, what leaves the device, and who's responsible for each piece.

Report this comment

This is where agentic AI gets really interesting. Moving intelligence closer to the user means agents can actually operate within the environment where the work happens, with lower latency, better privacy, and less dependence on cloud inference. The idea of turning every PC into an agent execution environment feels like a major shift in how we think about software itself.