Microsoft has unveiled a significant update for Copilot on Windows 11, fulfilling earlier cryptic hints about an upcoming feature that would allow users to “rest their fingers”. This major enhancement introduces voice interaction capabilities, allowing users to converse directly with the AI assistant, marking a crucial step in transforming Windows into a fully integrated AI-native platform.
The core addition is Copilot Voice, which will be generally available in countries supporting Copilot services on Windows 11. Users can initiate conversations with the command “Hey Copilot” and conclude them with “Goodbye”. This opt-in feature, which must be enabled within Copilot, aims for more natural interaction. It will enable tasks such as summarising webpages, setting up meetings, drafting emails or editing documents in Word, all without typing, and it supports continuous, cross-application commands, functioning much like a personal secretary.
Complementing the voice feature is Copilot Vision, which grants the AI the ability to “see” what is on the user’s screen, including tables in images, scanned documents, or PowerPoint slides. When a user shares their desktop or an application, Copilot analyses queries and offers real-time guidance through a “Highlights” feature. This can involve visually indicating where to click or demonstrating usage steps within an application. For users less inclined towards voice, Microsoft is also preparing a Text-in Text-out mode for the Windows Insider build.
The Windows 11 interface is being updated with an Ask Copilot button on the taskbar, resembling a search box, alongside new shortcuts to quickly access the Voice and Vision features, integrating Copilot into daily workflows. Microsoft has reiterated that while search responsiveness is improving, Copilot does not access files or applications directly, instead relying only on the same Windows APIs as Windows Search and requiring explicit user permission for every access request.
Further development is underway with Copilot Actions, currently being tested with Windows Insiders, and Copilot Labs. These initiatives are focused on enabling Copilot to directly manage local files – such as organising images, correcting skewed photos, deleting duplicates, or extracting data from PDFs. These background processes will allow users to monitor progress at any time, with the scope of these local file capabilities expected to expand as the feature matures.
Copilot Connectors on Windows will facilitate data linkage with services across both Microsoft and Google ecosystems, including OneDrive, Outlook, Gmail, Google Calendar, and Contacts. Once authorised, Copilot can search for and retrieve personal data based on verbal commands – for instance, locating a calendar appointment or pulling an old cloud document – and can then immediately pass the results to Word, Excel, or PowerPoint for further work.
In a move to simplify system management, Copilot will also integrate fully with Windows Settings. Users can now type or speak requests like “make the screen easier to read” or “reduce distractions during work”, and Copilot will directly open the relevant settings panel. Microsoft confirms that all new Copilot features operate on an explicit consent basis, meaning the AI only accesses data when permission is granted, and this access can be revoked at any time. While the taskbar experience does not require an NPU chip (the chip designed for AI processing) for the Voice and Vision features on Windows 11, the presence of one will ensure faster and more secure audio and visual processing.
This comprehensive update represents a significant shift, evolving Windows from a standard operating system into a more intuitive personal assistant capable of listening, seeing the screen, and acting upon request. Sound, sight, and AI are positioned to become the new cornerstones of the modern user experience, with Microsoft aiming to establish Windows as the central hub for AI-native experiences for both consumers and enterprises, promising further enterprise details at the next Microsoft Ignite event.