Microsoft launches Scout, an OpenClaw-powered always-on AI agent for Microsoft 365
Microsoft unveiled Scout at its Build 2026 developer conference on June 2, an always-on AI agent built on the open-source OpenClaw framework. Scout is designed to operate continuously within Microsoft 365 apps like Teams, Outlook, OneDrive, and SharePoint, proactively handling workplace tasks without waiting for user prompts. The agent can schedule meetings, manage calendars, draft responses, and identify workflow issues. Unlike Copilot, which responds to direct questions, Scout acts autonomously with its own identity and persistent memory. Microsoft positions Scout as an enterprise-grade alternative to uncontrolled OpenClaw deployments, emphasizing security controls including policy conformance systems, sandboxed execution, and audit trails. Over three thousand Microsoft employees currently use the tool internally. Scout launches in limited preview via Microsoft's Frontier program, with broader availability expected later in 2026. Pricing remains undisclosed, though the company is building Scout as an add-on to Microsoft 365 subscriptions.
Key facts
- Scout launched June 2, 2026 at Microsoft Build conference as an always-on AI agent built on OpenClaw framework
- Over 3,000 Microsoft employees currently use Scout internally for tasks including scheduling, travel booking, and form filling
- Scout integrates with Teams, Outlook, OneDrive, SharePoint and can access calendar, email, chat, and contact data
- Microsoft operates OpenClaw in a sandboxed environment treating it as untrusted code, with policy conformance system that continuously audits agent actions
- Scout available initially through Frontier program to limited enterprise customers; requires GitHub Copilot subscription and Intune policy configuration
- Pricing not yet disclosed; analyst Jeff Pollard warns Scout amplifies existing data governance problems but cannot introduce entirely new risks to organizations with Copilot already deployed
How different outlets framed it
Left-leaning coverage: Privacy advocates and security analysts highlight amplified data risks from autonomous agents accessing emails, calendars, and files. Sources note OpenClaw has previously demonstrated security flaws, and even Microsoft's sandboxing cannot eliminate risks of data exposure, agent manipulation through prompt injection, or unexpected autonomous actions. Some analysts express concern about always-on surveillance of workplace activity and question whether data governance problems existing in current systems will be magnified when agents can act on sensitive information.
Right-leaning coverage: Business and enterprise-focused outlets emphasize productivity gains, competitive advantage, and Microsoft's strategic positioning. Sources highlight how Scout reduces coordination overhead for knowledge workers, allows meeting scheduling across time zones, and helps prevent missed deadlines. Commentary stresses Microsoft's existing enterprise infrastructure advantages and how Scout drives adoption of Microsoft 365 by increasing switching costs through customization and learned preferences over time.
Limits of this briefing
Coverage reflects early post-announcement perspectives from tech media and analysts. Limited real-world usage data available beyond Microsoft's internal 3,000-person pilot. Some security concerns raised by Microsoft's own February 2026 guidance on OpenClaw risks, which recommended against deploying it on standard workstations. Broader enterprise adoption patterns and actual security incident data will emerge only after wider rollout.
Sources
- Computerworld: Microsoft unveils Scout, an autonomous AI agent built on OpenClaw Centre
- TechCrunch: Microsoft launches Scout, an OpenClaw-inspired personal assistant Centre
- Bloomberg: Microsoft Unveils Scout AI Assistant to Automate Workplace Tasks Centre
- Windows News: Microsoft Scout Turns AI Into a Governed Workplace Worker Centre
- Semafor: Microsoft launches AI assistant powered by OpenClaw Centre