Microsoft Develops New AI Agent Similar to OpenClaw
▼ Summary
– Microsoft is testing OpenClaw-like features for its Microsoft 365 Copilot, aiming at enterprise customers with enhanced security controls.
– The potential new agent would be a version of Copilot that is always active, capable of executing multi-step tasks over extended periods.
– This development follows other agentic tools from Microsoft, like the cloud-based Copilot Cowork and Copilot Tasks.
– Microsoft has partnered with Anthropic, using its Claude model to power the Cowork agent.
– The company may demonstrate this new tool at its Microsoft Build conference in June.
Microsoft is actively developing a new AI agent with capabilities similar to the popular open-source tool OpenClaw, integrating these features directly into its Microsoft 365 Copilot platform. According to a company confirmation, this initiative is specifically aimed at enterprise customers, with a strong emphasis on providing superior security controls compared to the inherently riskier open-source alternative. This move represents a significant step in Microsoft’s strategy to enhance its suite of agentic tools, which are designed to autonomously perform complex, multi-step tasks on a user’s behalf.
The project, often referred to internally as a “Claw,” would join a series of other agent announcements from the tech giant in recent months. While its exact architecture remains under wraps, a key feature described to reporters is an always-on version of 365 Copilot capable of taking actions at any time to complete lengthy workflows. This contrasts with the company’s other recently introduced agents, such as Copilot Cowork and Copilot Tasks, which operate primarily in the cloud. Cowork, for instance, leverages Anthropic’s Claude model and a proprietary Work IQ technology to personalize task automation within Microsoft 365 apps, but it does not run on local hardware.
The potential development of a local agent is particularly noteworthy. The open-source OpenClaw project runs directly on a user’s computer, a feature that has driven significant demand for hardware like the Mac Mini among its user base. By creating its own secured, locally-operating version, Microsoft could address both security concerns and competitive motivations, offering a tightly integrated solution within its own ecosystem. Industry observers anticipate a potential reveal of this technology or an upgraded existing tool at the upcoming Microsoft Build conference in June.
When pressed for specifics on how this new agent compares to prior announcements or whether it will indeed run locally, Microsoft maintained a cautious stance. A company spokesperson stated that they are continuously experimenting with broader orchestration and autonomy across both enterprise and consumer AI experiences. The core goal, they emphasized, is to reduce daily friction for users while remaining firmly anchored in principles of security, governance, and trust. This ongoing experimentation underscores the competitive and fast-evolving nature of the AI agent landscape.
(Source: TechCrunch)
