Analyst(s): Keith Kirkpatrick
Publication Date: November 4, 2025
Microsoft has launched Researcher with Computer Use in Microsoft 365 Copilot, a capability that enables the AI agent to perform hands-on tasks through a secure Windows 365 virtual machine. It enhances Copilot’s ability to navigate, log in, and interact with authenticated data sources safely.
What is Covered in this Article:
- Launch of Researcher with Computer Use in Microsoft 365 Copilot.
- Functionality through a Windows 365 virtual machine for authenticated access.
- Use of sandboxed environments, visual browsers, terminals, and text browsers.
- Benchmarked performance improvements on GAIA and BrowseComp.
- Security measures, user safeguards, and administrative controls.
The News: Microsoft has rolled out Researcher with Computer Use in Microsoft 365 Copilot, marking a move toward autonomous AI that can work on behalf of users. This new feature lets Researcher run inside a secure Windows 365 virtual machine, where it can browse public and restricted websites, sign in to verified sources, and create research materials like reports and presentations.
The system combines reasoning with hands-on execution, using tools like a virtual browser, terminal, and text browser, all within a closed sandbox. Microsoft says these environments are temporary, isolated from networks, and monitored by safety filters to protect users and company data. The feature is first being made available to Microsoft 365 Copilot customers in the Frontier program.
Microsoft 365 Copilot Expands Researcher Agent with Secure Computer Use
Analyst Take: With Researcher with Computer Use, Microsoft is moving Copilot from just reasoning to actually taking action safely and independently. It’s a careful step toward AI agents that can interact with real systems while keeping enterprise-level security intact, which is a key element to driving significant ROI from agents.
Secure Autonomy Through Sandboxed Execution
Researcher with Computer Use runs entirely inside a sandboxed Windows 365 setup, keeping all its actions separate from the user’s device and network. This sandbox runs a full operating system with a browser and terminal, allowing it to browse, fill out forms, run code, and log into secure sites. Every action inside the sandbox is checked by a network proxy using built-in safety filters. These filters make sure all outgoing activity matches the user’s task, protecting against harmful or unrelated web actions. This setup keeps Copilot’s new autonomy safe, contained, and trackable at every step.
Enhanced Research Depth and Benchmark Performance
The Computer Use feature lets Researcher reach beyond open web content to access secure or subscription-based data, when permission is given. Microsoft tested the upgraded model with GAIA and BrowseComp benchmarks, which measure multi-step reasoning and browsing ability. On BrowseComp, it performed 44% better than before, and GAIA showed a 6% improvement. These results show that Researcher’s mix of reasoning and execution boosts both accuracy and completeness in research work.
User Transparency and Control
A key part of the new setup is the Visual Chain of Thought, which shows live screenshots and terminal outputs as Researcher operates. This gives users a clear look at what’s happening and lets them step in when credentials or approvals are needed. The model never keeps or shares personal data, and users can take over the sandbox safely through screen-sharing when prompted. This transparency means users stay aware, in charge, and protected while still benefiting from the AI’s autonomy.
Governance and Administrative Oversight
Microsoft has built several layers of governance into this release. Admins can control when Computer Use is active, choose approved security groups, and allow or block certain web domains. Any browser activity or files created during a chat are auditable to meet company policy. By default, access to internal organizational data is turned off while Computer Use is running, reducing risk but still letting users enable data sources as needed. This balance of flexibility and oversight keeps Microsoft’s focus on enterprise trust and accountability intact.
What to Watch:
- Adoption trends among Frontier program enterprises as Computer Use expands.
- Potential integration with broader Copilot features, such as Copilot Actions.
- Performance consistency and latency during extended research sessions.
- How enterprises balance sandbox restrictions with productivity demands.
- Administrator feedback on auditability and compliance management.
See the complete blog post on the introduction of Researcher with Computer Use on the Microsoft 365 Copilot blog.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other insights from Futurum:
Microsoft Q1 FY 2026: Cloud and AI Fuel Broad-Based Growth
Microsoft Introduces New Agents and Benchmarks for Dynamics 365
Microsoft Agent Framework: Advancing The Agentic DevOps Vision
Author Information
Keith Kirkpatrick is VP & Research Director, Enterprise Software & Digital Workflows for The Futurum Group. Keith has over 25 years of experience in research, marketing, and consulting-based fields.
He has authored in-depth reports and market forecast studies covering artificial intelligence, biometrics, data analytics, robotics, high performance computing, and quantum computing, with a specific focus on the use of these technologies within large enterprise organizations and SMBs. He has also established strong working relationships with the international technology vendor community and is a frequent speaker at industry conferences and events.
In his career as a financial and technology journalist he has written for national and trade publications, including BusinessWeek, CNBC.com, Investment Dealers’ Digest, The Red Herring, The Communications of the ACM, and Mobile Computing & Communications, among others.
He is a member of the Association of Independent Information Professionals (AIIP).
Keith holds dual Bachelor of Arts degrees in Magazine Journalism and Sociology from Syracuse University.
