Windows Developer reposted this
Agents are only as valuable as the trust you can place in them. 🤝 Microsoft is building Windows for hybrid intelligence — and with MXC, giving AI agents a secure place to run, with containment that adapts to each prompt and tool call. We're glad to be among the partners bringing support for it. For Egnyte, it's another step in a long relationship with Microsoft, and another way to extend our approach right to the agent layer on Windows: https://bit.ly/4hBMz7F
Windows Developer reposted this
Hermes Agent is now live (and featured) in the Microsoft Store for seamless one-click install. More updates for the Windows ecosystem coming soon!
Windows Developer reposted this
We're excited to support Microsoft’s launch of its new Surface Laptop Ultra devices! Our work with Microsoft brings native Python, conda, and Anaconda’s vetted packages to Windows on ARM from day one. AI developers and data scientists can build without emulation, create reproducible environments, and use a trusted software supply chain for AI. Kilo Desktop powers Local AI all on the device.: set up an environment, run a model locally, and build or explore with AI agents. A project started on one Surface can be rebuilt by a teammate using the same environment file. From unboxing to a working local AI application in minutes. Join 50M+ Anaconda users and 5M+ Kilo developers building with AI. Get started with Kilo Desktop and Anaconda packages: https://lnkd.in/g62qfJq4
More local compute means more room to experiment. Surface RTX Spark Dev Box brings powerful local AI compute to the desk, helping developers and AI engineers experiment, evaluate models, and iterate locally. It’s another way Windows is giving builders the compute they need for demanding local AI workflows, right at their desk. Surface RTX Spark Dev Box is available for preorder now and begins shipping to customers in the U.S. in November. Learn more and preorder: https://msft.it/6044as8ns
The future of AI is hybrid intelligence. Not everything needs to run in the cloud. This week, Pavan Davuluri showed what's possible with Portable Computer by Perplexity on Windows. Portable Computer runs a local model and agent workflows on compatible Windows PCs. It can work with permitted files, connect to local tools, and run recurring tasks without consuming Computer credits for on-device work. When a task needs current information or advanced reasoning, users can bring in Perplexity Search or frontier models, with permission before data is sent to the cloud. For developers building agents on Windows, Microsoft Execution Containers (MXC) provide an isolated environment to contain agent actions and help protect the host system. Try Portable Computer for Windows by Perplexity: https://lnkd.in/eKm2exvv
What if your coding workflow could intelligently choose between local and cloud AI? GitHub Copilot HydraFusion routes each task to the right model in the cloud. Now, we're extending that intelligence to Windows, enabling HydraFusion to tap into models running locally on your device. The result is a hybrid developer workflow that can use local intelligence when it makes sense while preserving access to cloud capability. Coming to GitHub Copilot app, GitHub Copilot CLI and Visual Studio Code in experimental preview later this month. Learn more: https://msft.it/6041aU29t
Windows Developer reposted this
An update for builders using Codex on Windows: We’ve built a new sandbox mode using @Microsoft’s Execution Containers (MXC) for faster setup, stronger network enforcement, and granular file access controls. Requires a compatible Windows 11 device. https://lnkd.in/ds2tCBRv
Building a more secure future for AI agents takes an ecosystem. 🤝 With Microsoft Execution Containers (MXC) now generally available, leading agents and frameworks are already bringing policy-driven containment into their workflows. From GitHub and OpenAI to NVIDIA, Replit, OpenClaw, LM Studio and Unsloth AI, partners are helping developers run agent workloads with clearer boundaries around what they can access. And the ecosystem continues to grow, with upcoming support from Anthropic, Box, Egnyte, Heidi Health, Hermes Agent, Manus AI, Perplexity, Raycast and Simular. Explore the growing MXC ecosystem: https://msft.it/6043aUjJJ
Local AI development is moving fast, and developers need the flexibility to experiment with the models that work best for them. That’s why we’re bringing llama.cpp support to Windows ML. The integration gives developers more open-source model choices and the ability to quickly experiment with emerging models as they land. And with Windows ML as a scalable runtime for deploying models across GPU, NPU, and CPU, developers have more options for building local AI experiences on Windows. Learn more: https://msft.it/6043aUwKc
What if frontier-class coding intelligence could run locally on your PC? MAI Code 1.1 Flash is designed for real-world coding workloads, with 137 billion total and 6.8 billion active parameters. On Windows, 3-bit precision reduces the model size by 33% while preserving coding quality and supporting a 256K context window locally. It joins other powerful models including NVIDIA Nemotron Bolt 74B and Deepseek v4 Pro Flash, a 280B parameter model, optimized to run locally on RTX Spark. See how Windows is bringing frontier-class capabilities to local PCs: https://msft.it/6042aUz3r