Microsoft has unveiled Scout, its first AI agent built on the open-source OpenClaw platform. The launch was one of the key announcements at the Build conference and highlights the company’s growing focus on using agent-based systems to automate work across the Microsoft 365 ecosystem.
According to Microsoft, Scout is integrated across Microsoft 365 services and operates in the cloud, on desktop devices, and through web interfaces. Users can interact with the agent through Teams and extend its capabilities using a desktop companion app that provides access to web browsers, local resources, and Model Context Protocol (MCP) servers.
Microsoft says Scout is designed to work proactively. The agent can independently schedule meetings, prepare supporting materials, carry out various work-related tasks, and gradually adapt to the needs and preferences of individual users.
“You interact with it in Teams and extend its reach through a desktop app to your browser, local resources, and Model Context Protocol servers,” Microsoft said.
Microsoft also placed a strong emphasis on security. Alongside the launch of Scout, the company shared new details about Microsoft Execution Containers (MXC), an isolated execution technology designed to improve the security of AI agent systems in enterprise environments.
For business customers, agents operate under their own managed Entra identities rather than shared service accounts. Microsoft says this approach provides more precise control over agent actions and access permissions within corporate environments.
“Identity shows who is acting; access control determines what they can do,” the company said.
At the moment, Scout is available only to a limited group of customers. Microsoft has not yet announced when the AI agent will become generally available.
The company also revealed that OpenClaw on Windows uses MXC containers, allowing nodes and gateways to run in isolated environments that help protect user systems from potential security risks.
“You can use the new Windows companion app to easily set up your own Claws or connect to existing ones,” Microsoft said.
In addition to Scout, Microsoft announced new capabilities for its Microsoft Security platform aimed at testing and securing AI agents. The company’s new multi-model agentic scanning system is now available in extended preview for eligible organizations and has been integrated with Microsoft Defender.
According to Microsoft, the system orchestrates more than 100 specialized AI agents and multiple AI models to identify, validate, and prove exploitable vulnerabilities in codebases written in popular programming languages.
“This new agentic security system orchestrates more than 100 specialized AI agents, using an ensemble of models to discover, validate, and demonstrate exploitability in codebases,” Microsoft said.
The company concluded that the future success of agentic AI will largely depend on trust. As organizations deploy increasingly autonomous systems, security remains one of Microsoft’s key priorities.
