Every Claude product so far waits for you to type. Claude Duet does not. On October 21, 2026, Anthropic opened a research preview of a new app where Claude comes to you first, and it keeps working while your phone and laptop are off. Access is by waitlist for now.
What Claude Duet is
Anthropic calls it "a new app where Claude works as a proactive agent." It is built around Claude Opus 5.5, and Anthropic says it was made for professionals and small business owners who have more on their plates than they can track. It is a standalone app on iOS and Android, and it is also on claude.ai.
Setup is light because it carries over your memory, skills and connected apps from your Claude account. You talk to it by typing or by voice in one ongoing thread, so a quick reply and a weeklong project start in the same place.
How it works
- Morning brief. Each morning it sends you a short brief on the most important items across your connected apps.
- It reaches out first. It checks your connected accounts every few hours and messages you only when something needs you. Anthropic says it does not check nonstop because that would use up your plan's limits much faster. You can also tell it what you want to hear about and how often.
- It keeps working when you step away. Duet has a virtual machine with its own browser, so your laptop or phone does not need to be on. You describe a task in a short message and it works out which apps and steps to use in the background.
- Connectors. It can use the over 6,000 connectors and plugins in the Claude Marketplace. Anthropic's examples include finding missed Gmail and Slack messages, checking QuickBooks or Stripe for unpaid invoices, preparing payroll in Gusto and drafting a LinkedIn job post.
In the launch video, Duet notices a health permit that was never filed and drafts it for you. That is the whole idea: it spots the thing you forgot before you have to ask.
The part I like most: it asks first
Duet reads, drafts and prepares work on its own. But before it sends a message, makes a payment, signs a document or deletes anything, it shows you exactly what it is about to do and waits for your approval. In the app that is a simple choice: Send, or Not now.
Some things stay off limits even if you approve, like trading securities or moving money between accounts. When it buys something for your business through Link (Stripe's digital wallet), it pays with a virtual card, so your real card number is not shared.
This is the pattern I would copy into any agent you build yourself: act alone on reads and drafts, ask before anything you cannot undo. Claude Code uses the same kind of guard, and I wrote about another one in Claude Code Now Has a Second Agent Watching Your Back.
The limits
- Access. It is a research preview by waitlist, starting with Pro and Max subscribers in the US. So it is not free.
- Usage. Duet shares your plan's usage limits with the rest of Claude. Its background checks count toward those limits, and bigger tasks like a financial model or a deck use more.
- Judgment. Anthropic says it can still misread a thread or miscalibrate what is urgent.
- Personal errands. It was built around work, so personal errands like booking tickets are not as strong yet.
- Visibility. It cannot yet show you its browser, and for now it lives in its own app, separate from the main Claude app.
My take
A proactive agent is only useful if it is quiet most of the time and right when it speaks. Checking every few hours instead of nonstop is an honest tradeoff, and the approval step for anything irreversible is the right default. The real test is whether its idea of "urgent" matches yours after a few weeks. If you are on Pro or Max in the US, join the waitlist and see.
Source: Anthropic: Introducing Claude Duet
Watch the short above for the quick version. If you want the Claude Code skills and automations I use every day, join the AI Academy community on Skool: https://www.skool.com/ai-academy-with-robby-6849/about