Microsoft Launches the Hybrid Intelligence Era: Windows is Reshaped for AI Agents
Quick Look
- Microsoft announced Microsoft Execution Containers (MXC) infrastructure for Windows 11, which offers a dynamic isolation layer that controls file and network access at runtime for autonomous artificial intelligence agents.
- MXC is supported by leading agents such as OpenAI Codex, GitHub Copilot and Anthropic Claude Code.
- It also optimizes token costs with its hybrid intelligence architecture and smart routing technology, which combines local and cloud intelligence.
AI-generated summary
Why It Matters
Since autonomous artificial intelligence agents have the ability to edit files, write code, and run continuously in the background, different security and control requirements arise from traditional applications. Microsoft developed the MXC infrastructure to meet this need.
The Most Secure Platform for Agents: Microsoft Execution Containers (MXC)
Unlike traditional applications, autonomous artificial intelligence agents have the ability to edit files, write code and run continuously in the background. This situation necessitates the rewriting of security and auditing standards. Microsoft is making the Microsoft Execution Containers (MXC) infrastructure, designed to meet this need, generally available for Windows 11.
MXC; It provides a dynamic isolation layer that controls at runtime which files and networks an agent can access. This structure, which provides flexible security options for institutions and individual users from process and session isolation to virtual machines, is supported by industry-leading agents such as OpenAI Codex, GitHub Copilot, Anthropic Claude Code and Meta Muse.
Cloud and Local Power Combine: Hybrid Intelligence and Intelligent Routing
As the costs and resource consumption of large language models and artificial intelligence tools increase, users are looking for the best performance without straining their cloud budgets. Microsoft provides a solution to this problem with its hybrid intelligence architecture that blends local artificial intelligence and cloud intelligence. Advanced open source models such as MAI Code 1.1 Flash, NVIDIA Nemotron and DeepSeek V4 Flash work locally on the device, producing instantaneous responses without sending data to the cloud.
Smart routing technology, announced in collaboration with GitHub and directing tasks to the most appropriate model, optimizes token costs by activating local models. Additionally, llama.cpp support added to the Windows ML platform allows developers to instantly adapt to the latest open source projects.
A New Era in Copilot and Hardware Ecosystem
Supercharged Copilot: Local context, local actions and local models are integrated in Copilot+ PCs. With the user's permission, the assistant can access files on the computer and perform tasks such as file editing and troubleshooting directly on the device.
RTX Spark and Powerful Hardware: New devices from manufacturers such as Acer, ASUS, Dell, HP, Lenovo, MSI and Microsoft Surface, based on the NVIDIA RTX Spark architecture, are available for pre-order. Surface Laptop Ultra handles the heaviest AI loads locally with up to 128 GB of combined memory.
Open Questions
- What is MXC's pricing model?
- In which sectors will MXC be used first?
- Are performance benchmarks available for MXC?
