Microsoft Eyes OpenClaw-Like AI Agent for Enterprise Customers
Microsoft is secretly working on integrating OpenClaw-like features into its existing Microsoft 365 Copilot tool, aimed at enterprise customers. This new feature would provide enhanced security controls compared to the open-source OpenClaw agent, which has raised concerns about its security risks.
"The idea is an agent that can complete multistep tasks over long periods," a source within Microsoft revealed to The Information.
Microsoft's AI Agent Expansion
This development is part of Microsoft's ongoing efforts to expand its AI agent capabilities. In March, the company announced Copilot Cowork, designed to take actions in Microsoft 365 apps, not just provide search results or chat in a separate work pane. Cowork is powered by its own 'Work IQ' technology, an intelligence layer that tries to personalize Cowork for the user across Microsoft 365 apps.
A New Local AI Agent?
It's unclear whether this new agent would be local or if it would simply adopt some of the other features that OpenClaw advocates love. Microsoft did tell The Information that one of the main features of the agent is that it would essentially be a version of 365 Copilot that is always working, able to take actions at any time.
Security Concerns and Microsoft's Motivations
OpenClaw is a tool that runs locally on a user's computer and can create agents to perform tasks on behalf of the person. While the open-source project OpenClaw can run on Windows machines, the Mac Mini has been the go-to platform for OpenClaw users. So much so that the small, affordable, cube-shaped Mac Mini desktops have suddenly been selling like hotcakes. Beyond security, Microsoft may have a number of motivations for wanting its own version.
A June Debut?
The company is expected to show off this new agent (or an upgraded version of one of its existing agents) at its Microsoft Build conference in June, The Verge reports. We've asked Microsoft how this new agent fits in with its other agents and will update the story when we receive comment.