Claude Duet writes first and asks before it acts: the right order

My take: A Claude that writes first is what I want, as long as it asks before sending. The real test will be how much of my quota its checks eat.
What to know
- On October 21, 2026, Anthropic opened a research preview of Claude Duet, a separate app on iOS, Android and claude.ai in which Claude acts as a proactive agent: an AI that takes the first step instead of waiting for your question. Each morning it sends a short brief on what matters in your connected apps, then checks them every few hours and messages you only when something needs you.
- Anthropic's examples: find the Gmail and Slack messages you missed, check QuickBooks or Stripe for unpaid invoices and draft the reminders, flag a call that clashes with a flight. It works on its own virtual machine, so your laptop and phone can stay off. Before it sends a message, pays, signs or deletes anything, it shows you the action and waits for your approval; trading securities and moving money between accounts stay off-limits even then.
- Access goes through a waitlist, starting with Pro, Max and Ultra subscribers in the US. During the preview, Duet shares your plan's usage limits with the rest of Claude, and its background checks count toward them.
Spotted on @believemyhq's reel, in French.
What changes
For freelancers and small business owners, the people Anthropic built it for, the promise is fewer things that slip: the unpaid invoice, the reply you forgot. The reel says Duet can even send the reminder. It can, but only once you approve it, and that is what makes it usable. Anthropic also warns that it can misread a thread or misjudge what's urgent. The October 21, 2026 announcement covers the US only: if you are on a US plan, join the waitlist on the product page and connect one account first, not all of them; elsewhere, keep an eye on that page.
Source: Anthropic, “Introducing Claude Duet”, October 21, 2026
Drafted with Claude, Anthropic's AI, from my take. Reviewed and owned by me.