Microsoft is adding more AI agents to its workplace tools, alongside updates for Copilot, Windows, Foundry, and new device prototypes. The announcements, made at the Build developer conference in San Francisco, covered Copilot agents, Project Solara prototypes, Nvidia-powered developer machines, OpenClaw controls for Windows, and new models from Microsoft’s AI unit.
The Copilot updates included Scout, an agent designed to track items that need a user decision. Microsoft said Scout can collect emails, messages, or other items that require a user’s response before work can move forward.
Scout is part of a wider group of autonomous agents that Microsoft calls Autopilots. Satya Nadella, CEO of Microsoft, described them as enterprise-grade versions of tools such as OpenClaw, with more agents expected over time.
Microsoft also announced Microsoft IQ, a context layer for agents across GitHub Copilot, Microsoft Foundry, and Copilot Studio. One component, Work IQ, gives agents access to Microsoft 365 data, including emails, documents, and meetings. Work IQ is scheduled to become generally available on June 16.
Microsoft’s 2025 Work Trend Index said 81% of leaders expect AI agents to be moderately or extensively integrated into their company’s AI strategy within 12 to 18 months.
OpenClaw gets Windows controls
Microsoft is working on tools to support OpenClaw on Windows. OpenClaw is open-source software that can coordinate groups of AI agents to complete everyday tasks for users.
The company introduced Microsoft Execution Containers, or MXC, a Windows system designed to run AI agents inside restricted environments. OpenClaw ran inside MXC during a stage demo.
In the demonstration, Windows prevented the agent from deleting files from a user’s desktop even when OpenClaw’s own safety layers were disabled. The controls are designed for enterprise environments where computers handle sensitive corporate data, according to Microsoft.
Microsoft also showed how an IT department could prevent users from accidentally deleting desktop files.
Peter Steinberger, the software engineer who created OpenClaw, said during the event that companies can now run OpenClaw inside their organisations.
Steinberger said OpenClaw has worked with companies including Microsoft to add security controls that avoid an “all or nothing” access model. The controls are aimed at enterprise environments where IT teams need to limit what autonomous agents can access or change.
Microsoft expands its MAI model lineup
Microsoft gave updates on models developed by its AI unit. They include MAI Thinking-1, the company’s first reasoning model.
MAI Thinking-1 matched Anthropic’s Claude Opus 4.6 on coding abilities on SWE Bench Pro, according to Microsoft. Anthropic announced Opus 4.8 last week.
MAI Thinking-1 is available in private preview on Foundry. Microsoft described it as a 35-billion-active-parameter model with a 256K context window.
The model is built for complex multi-step instructions, long-context reasoning, and code generation, according to Microsoft. The MAI models are separate from the OpenAI models Microsoft also offers through its products and cloud services.
Microsoft’s AI unit also released a transcription model it described as the most efficient among cloud hyperscalers. The company announced an image model that it compared with Google’s image tools.
Microsoft announced several MAI models alongside MAI Thinking-1. MAI-Code-1 is available in Copilot and VS Code, while MAI-Transcribe-1.5 and MAI-Voice-2 update the company’s speech recognition and voice generation models.
CNET reported that MAI-Transcribe-1.5 can transcribe 43 languages, while MAI-Voice-2 adds 15 languages and new voice options.
The company also introduced MAI-Image-2.5 and a faster variant for image generation. MAI-Image-2.5 is available in PowerPoint and Foundry, with a planned rollout to OneDrive.
Project Solara targets agent devices
The company showed Project Solara, a group of prototype devices designed to run AI agents instead of conventional apps. The prototypes include small devices similar in size to a smart speaker or keycard badge.
The devices use chips from Qualcomm and MediaTek. They include screens and microphones and are designed to connect with cloud systems for specific tasks.
Reuters reported that one example shown by Microsoft involved documenting a medical visit with a nurse.
Microsoft described Project Solara as a chip-to-cloud platform for agent-first devices. One prototype, an Access Badge built with Qualcomm silicon, supports fingerprint unlocking and voice commands.
Companies including AccuWeather, Best Buy, CVS Health, and Target are planning pilots for the Project Solara hardware, according to Microsoft. The company also showed a desk concept based on Project Solara.
A working prototype of the MediaTek device was not shown, and Microsoft did not give a launch date.
The Verge reported that Project Solara is built on a version of Android known as the Microsoft Device Ecosystem Platform. It also reported that Microsoft does not plan to sell the concept devices itself and is using them as reference designs for hardware partners.
During the keynote, CEO Satya Nadella said Project Solara would let developers and enterprises design different form factors for AI agents. He said new computing platforms allow agents to be used across different form factors.
Microsoft pushes AI workloads onto PCs
Microsoft introduced the Surface RTX Spark Dev Box, a compact computer fitted with an Nvidia chip. Nadella described the device on stage as a “dream machine” and said he was on the waitlist to buy it.
Nvidia has said PCs using its RTX Spark chip are designed to bring AI workloads directly onto personal computers. Microsoft showed the Surface RTX Spark Dev Box running a 120-billion-parameter AI model, which most PCs cannot load locally.
The device follows another Microsoft and Nvidia laptop announcement made during the same week. The new machines are priced against premium PCs, though analysts said business adoption may take time.
Discovery brings agents to research
Microsoft introduced Microsoft Discovery, an Azure-based AI platform for scientists and researchers. Organisations including BHP, GSK, and Syensqo are using Discovery for research, according to the company.
BHP is using it for mining-related research, GSK for drug development, and Syensqo for semiconductor-related work.
Discovery can explore hypotheses and run simulations over long periods, according to Microsoft. The platform draws on public scientific literature and internal knowledge.
David Carmona, vice president of Microsoft Discovery and Quantum, said the system is designed to give researchers visibility into how results are produced.
Microsoft also extended its AI work in healthcare through an agreement with Mayo Clinic to develop frontier AI for medical diagnostics. The partnership will combine Microsoft’s reasoning models and compute infrastructure with Mayo Clinic’s clinical expertise and data. Farrugia said the goal is to help medical staff reach diagnoses faster and more accurately.
(Photo by Rubaitul Azad)
See also: Microsoft moves engineers from Claude Code to GitHub Copilot CLI
Want to dive deeper into the tools and frameworks shaping modern development? Check out the AI & Big Data Expo, taking place in Amsterdam, California, and London. Explore cutting-edge sessions on machine learning, data pipelines, and next-gen AI applications. The event is part of TechEx and co-located with other leading technology events. Click here for more information.
Developer Tech News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.