Microsoft tests OpenClaw-style AI agents for autonomous Copilot
What's the story
Microsoft is testing OpenClaw-style features in its 365 Copilot, The Information reported. The move is part of a larger effort to make the AI assistant work independently around the clock, performing tasks on behalf of users. Omar Shahine, Microsoft's corporate vice president, confirmed that they are "exploring the potential of technologies like OpenClaw in an enterprise context."
Tech details
What is OpenClaw?
OpenClaw is an open-source platform that lets users create AI-powered agents running locally on their devices. The platform gained popularity earlier this year but has since raised some serious security concerns. Despite these issues, Microsoft remains confident in its ability to implement "safer" versions of the tool within 365 Copilot.
Future plans
The always-on version of Copilot could suggest daily tasks
The always-on version of 365 Copilot could do things like monitor a user's Outlook inbox and calendar, suggesting tasks daily. Microsoft is also looking into OpenClaw-like agents for specific roles like marketing, sales, and accounting. This would limit the permissions needed by these agents and keep them isolated from other parts of a business.
Event showcase
Competing with Claude Cowork
Microsoft plans to showcase some of these features at its Build conference starting June 2. The move comes after Anthropic integrated its Claude AI chatbot into Microsoft 365 services and introduced the Claude Cowork tool to Copilot for completing "long-running, multi-step tasks." Integrating OpenClaw-like capabilities into Copilot could help Microsoft win back customers who switched to rival services.