Microsoft Debuts 'Scout': The Agentic AI Autopilot Revolutionizing M365

Published 5 hours ago3 minute read
Uche Emeka
Uche Emeka
Microsoft Debuts 'Scout': The Agentic AI Autopilot Revolutionizing M365

Microsoft has officially announced the wider testing of its innovative Autopilot feature at the recent Microsoft Build event, detailed further in a company blog post. Autopilots are defined as a new category of intelligent agents designed to operate autonomously on behalf of users. A key characteristic is that each Autopilot possesses its own distinct identity, enabling multiple agents to coexist and function within different rule sets. This allows users to deploy Autopilots for various contexts, such as at home or work, with tailored governance and stipulations that either limit or permit specific activities based on the operational environment.

Microsoft's inaugural Autopilot project is named Scout. Following initial internal beta testing, Scout is now being rolled out to a select group of customers and 'Frontier organizations'. Scout's primary role will be as an agentic assistant within Microsoft 365 applications, integrating seamlessly across Outlook, OneDrive, SharePoint, and Teams. Its capabilities include coordinating data from these platforms to facilitate tasks such as scheduling meetings, flagging important messages, and generating calendar events, all aimed at ensuring workers stay on track with their responsibilities. Over time, Microsoft Scout is designed to learn individual user preferences and work patterns, continuously aligning its activities and priorities to enhance efficiency and provide a more personalized experience.

Underpinning Scout's technology is OpenClaw, a project created by Peter Steinberger. Microsoft has also stated its intention to contribute upstream to the open-source OpenClaw project, fostering collaborative development. A significant focus for Microsoft is enterprise-grade security and robust controls for Scout, ensuring it can be trusted within organizations from its initial deployment. Administrators will have the ability to validate that any Scout implementations operate securely within the established IT and security policies, and agent identities can be validated via dedicated Entra entries.

Microsoft asserts that its agentic platform will be managed with the same rigorous standards expected from any first-party Microsoft service. Data protection policies for the algorithm are derived from Microsoft Purview, and credentials associated with machine identities are redacted from logs and diagnostics to safeguard anonymity. Crucially, human approval is required for any actions deemed sensitive by the algorithm, adding a layer of oversight. Early internal trials have allowed Microsoft to identify and mitigate potential risks for testers using Scout on the desktop, enabling the company to fine-tune the agent to strike a balance between security concerns and the need to maintain work momentum without constant user prompting.

The benefits of Autopilots include alleviating the burden of low-level tasks, thereby keeping work in motion even when a user's attention is elsewhere. One specific feature highlighted is Scout's ability to identify deadlines, proactively block-book a user’s calendar to prevent conflicting activities in the lead-up to a critical deadline, and provide necessary materials to overcome identified bottlenecks for important, focused projects. The announcement on the Microsoft website was penned by Omar Shahine, Corporate Vice President of Microsoft Scout, a long-standing Microsoft executive with previous experience in divisions such as Windows Live, OneDrive, and Mac Office.

For early adopters interested in trying Scout, enrollment in Microsoft’s Frontier programme is required, along with an Intune policy configuration, an 'opt-in attestation', and an active GitHub Copilot licence.

Loading...
Loading...

You may also like...