Computer-using agents now deliver more secure UI automation at scale
Recent updates to computer-using agents (CUAs) in Microsoft Copilot Studio improve UI automation with more secure credentials, detailed monitoring, and enhanced scalability. The updates address user feedback regarding agent scalability, authentication issues, and lack of control before broad rollouts. CUAs now offer multiple foundation models, including Anthropic's Claude Sonnet 4.5, alongside OpenAI's Computer-Using Agent, providing flexibility in navigating dynamic interfaces. Authentication is streamlined using built-in credentials, eliminating manual login prompts and enabling unattended execution. Enhanced monitoring includes session replay, action logs, and audit logs, ensuring visibility into every agent action and improving compliance. Additionally, managed Cloud PCs powered by Windows 365 simplify infrastructure demands by offering scalable cloud-hosted machines for reliability in automation.
Article 3w
Login now to access my digest by 365.Training
Learn how my digest works
Features
- Articles, blogs, podcasts, training, and videos
- Quick read TL;DRs for each item
- Advanced filtering to prioritize what you care about
- Quick views to isolate what you are looking for right now
- Save your favorite items
- Share your favorites
- Snooze items you want to revisit when you have more time