Anthropic is piloting a feature that lets users message Claude from a phone to autonomously complete tasks on their computer, such as opening apps, browsing the web and updating spreadsheets. The move intensifies the industry push toward “agentic” AI, following the viral rise of OpenClaw—whose creator was recently hired by OpenAI—and Nvidia’s unveiling of an enterprise-focused NemoClaw. Anthropic said Claude will request permission before accessing new apps and cautioned that the capability remains early and error-prone. The company also highlighted Dispatch in Claude Cowork for continuous conversations across devices, underscoring a race among AI firms to build assistants that act, not just chat.





























