Claude Duet: Anthropic's agent that messages you first
Anthropic opened a research preview of Claude Duet on October 21: a separate Claude app that watches the work apps you link to it, sends you a morning brief and works on tasks in the background on its own cloud computer. It asks before it sends, pays, signs or deletes anything. US Pro and Max subscribers can join the waitlist.
What happened
Anthropic announced Claude Duet on October 21, 2026, as a research preview. Most Claude products wait for you to type. Duet is built the other way round: it keeps an eye on the apps you connect, tells you what needs attention and does work while you are away. It runs on Claude Opus 5.5. Anthropic aims it at professionals and small business owners.
How it works
- Morning brief. Each morning it sends a short summary of the most important things in your email, calendar and other linked tools.
- Check-ins, not live monitoring. It looks at your accounts every few hours, not continuously, to save usage, and messages you only when something needs you. You choose what you want to hear about and how often.
- Its own computer. Tasks run on a cloud virtual machine with its own browser, so your laptop or phone can be off. You get a notification when a task is done or needs a look.
- One thread. You talk to it by text or voice in a single conversation. It inherits your Claude memory, skills and app connections.
- Connectors. Anthropic says it works with more than 6,000 connectors and plugins. The ones named include Gmail, Google Calendar, Slack, QuickBooks, Stripe, Gusto, Docusign, HubSpot, Canva, Shopify, Notion and Microsoft 365. For purchases it uses a virtual card through Stripe's Link, or Shop Pay, so your real card number is not handed over.
What it will and won't do on its own
Duet reads, drafts and prepares work by itself. Before it sends a message, makes a payment, signs a document or deletes anything, it shows you the exact action and waits for a yes. Some things are blocked even if you approve them: trading securities and moving money between accounts. It never sees your passwords or sign-in tokens directly, and a separate check refuses to share card numbers or government ID. Health and personal finance details are shared only when you tell it to.
Anthropic lists the weak spots itself: Duet can misread a thread or misjudge what is urgent, it can miss something that arrives between check-ins, it is better at work tasks than personal errands like booking tickets, and you can't yet watch its browser.
Who can use it
| Claude Duet | |
|---|---|
| Plans | Pro and Max, in the US, by waitlist. Team and Enterprise not yet. |
| Price | No extra charge. It uses the same limits as the rest of Claude. |
| Apps | Standalone iOS and Android apps; on a computer, claude.ai |
| Usage | Background check-ins count toward limits; big jobs like a financial model or a deck use more. |
What it means for you
- On Pro in the US? Join the waitlist, but start with one or two connected apps. Every check-in costs usage, and Pro limits are shared with your normal chats.
- Outside the US or on Team? Nothing to do yet. The same habit works in ordinary Claude today: one daily prompt that turns your inbox and calendar into a short list of what needs you. The first prompt below does that.
- Letting an agent near email and payments? Write down your own rules before you connect anything: who it may reply to, what it may never send, the spending limit. Duet asks about single actions; your rules decide which actions should happen at all.
Our Claude review covers the plans and their limits, and the Claude Skills catalog lists skills that Duet inherits from your account.
Try it
Below are today's unread emails, my calendar and my open to-do items. Write my morning brief: 1) Needs me today: at most five items, each with the deadline and the one action I should take. Put anything with money, a client or a deadline in the next 48 hours first. 2) Can wait: one line each. 3) Ignore: just count them. 4) Conflicts: double bookings, travel too close to a meeting, or a meeting with no agenda. Keep the whole brief under 200 words. Don't draft replies yet. Emails: [paste] Calendar: [paste] To-do: [paste]
I'm about to connect an AI agent to my work accounts: [list apps, e.g. Gmail, calendar, invoicing]. Interview me, one question at a time, to write a short rulebook for it. Cover: - who it may contact without asking me, and who never, - what it may send, pay, book, sign or delete, and the limits on each, - what counts as urgent enough to interrupt me, - what it should do when it isn't sure. When I've answered, write the rulebook as numbered rules under 250 words that I can paste into the agent's instructions.
Sources
- Anthropic, Introducing Claude Duet, October 21, 2026 (safety note updated October 23)
- Anthropic, Claude Duet product page and FAQ, checked October 27, 2026