IT Home October 22 news — Anthropic announced yesterday (October 21) the launch of a brand-new app, Claude Duet,positioned as a proactively working AI agent, joining the 'around-the-clock proactive AI assistant' battle led by Meta's Muse and OpenAI's Dots.
As for skills, the app organizes important cross-app matters every day, proactively reminds users of issues that need handling, and continues completing delegated tasks in the background after the user leaves. IT Home includes a demo video below:
Claude Duet is built on Claude Opus 5.5 and is available as standalone iOS andAndroid apps and on the claude.ai web version. The first batch opens a waitlist to Pro and Max subscribers in the United States; during the preview phase it shares plan usage limits with other Claude features.
Unlike ordinary chatbots, Claude Duet has its own virtual machine and browser and can keep running while the user's device is off. Users can assign tasks by text or voice, such as scheduling meetings, organizing unread email, tracking invoices, making financial or sales forecasts, building dashboards, and developing web apps; when a task is completed, it notifies the user or requests review.
On the tool ecosystem front, Claude Duet can draw on more than 6,000 connectors and plugins in the Claude Marketplace, covering common office and business services such as Gmail, Slack, QuickBooks, Stripe, Gusto, LinkedIn, HubSpot, Canva, Shopify, and DocuSign. It can also complete online payments confirmed by the user via Link or Shop Pay, and use virtual cards to avoid exposing the real card number.
On the security design front, Anthropic says Claude Duet does not directly touch users' passwords or login tokens; sending messages, making payments, signing documents, or deleting content requires user approval first. The system also includes a dedicated classifier to block requests that would leak highly sensitive information such as credit card numbers and government ID documents.
However, Anthropic also admits this is still an early version: Claude Duet may misread conversations or misjudge how urgent matters are; its handling of personal matters is relatively weak, and it cannot yet show users its browser actions.