Imagine being stuck in traffic or simply not wanting to work, while your laptop finishes your spreadsheets and emails your pitch deck for you—all on its own.
Anthropic has officially pushed its Claude assistant to follow OpeClaw’s footsteps. Beyond mere text-based conversations, this is what they call agentic AI – newly introduced computer-use capabilities. Claude is being trained to take over your keyboard and screen to retrieve files, navigate the internet, and operate software autonomously.
This capability is managed through two distinct layers: “Dispatch,” a tool released last week for remote task assignment via smartphones, and “Orbit,” an experimental mobile-native layer currently under testing. As long as the computer remains awake and online, the Dispatch will execute the requested workflows locally in a secure sandbox. Anthropic is also reportedly testing a complementary feature called “Orbit,” which aims to deliver the same level of autonomous control to smartphones by allowing the AI to manage calendars, compose messages, and place calls on the user’s behalf.
Anthropic’s rollout is a direct competition to OpenClaw, the open-source framework that demonstrated how AI can control local devices via consumer apps like WhatsApp and Telegram. The strategic stakes were validated when OpenAI recruited OpenClaw’s creator, Peter Steinberger, and Nvidia CEO Jensen Huang labelled the technology “the next ChatGPT.”
While Anthropic has implemented safeguards in ‘Dispatch’ and ‘Orbit’, like automated vulnerability checks and requiring explicit permission before the AI accesses new apps, the technology is still in an early research phase. The next eighteen months wil be defined by the struggle to maintain human-in-the-loop control over a software that now possesses the keys to local machines, tasks, and data.
