Anthropic said it has opened a research preview, an early test release, of Claude Duet. The Verge called the app a rival to Meta's Muse and OpenAI's Dots. Duet targets professionals and small business owners. It sends a morning summary, watches connected apps and takes on tasks. It is included in Pro and Max plans, and US users can join a waitlist.
Anthropic describes two safeguards. The first is a classifier, a screening model that checks each message. Anthropic says it is designed to refuse requests to share the strictest data, such as card numbers. That states an aim, and the sources give no measure of how well it works. The second is a rule: before Duet sends, pays, signs or deletes, it shows the action and waits for approval. Trading securities and moving money between accounts stay off-limits even then. Duet runs on a virtual machine with its own browser, so it keeps working when the user's devices are off. Anthropic says it never directly sees passwords or sign-in tokens. More safety details are promised.
Anthropic states the approval step as a firm rule but the classifier only as a design aim. The preview sits on individual plans, and Anthropic describes no admin, approval-routing or logging controls, so teams weighing wider use should ask whether they exist.