What Is Claude Duet, and Should You Join the Waitlist? (October 2026)

Update (October 23, 2026): Anthropic published its promised Duet safety, privacy, and security page. Key points are now in the safety section below: Duet never sees your passwords or sign-in tokens, a separate check blocks it from sharing card, bank, ID, or login details even if you ask, and disconnecting an app does not erase what Duet already learned. The help center also now says the one-time usage-limit reset for new users is available until November 30, 2026. Access is unchanged: US Pro and Max, by waitlist.

TL;DR: Claude Duet is a new, separate app from Anthropic, opened as a research preview on October 21, 2026, where Claude works as a proactive agent: it sends you a morning brief, checks your connected apps every few hours, takes on tasks you hand it, and keeps working in the cloud after you close the app. It runs on Claude Opus 5.5. Access is by waitlist for US Claude Pro and Max subscribers only, at no extra cost, but it shares your plan's usage limits. It asks before it sends, pays, signs, or deletes anything. Should you join the waitlist? If your day is mostly inbox, calendar, and follow-ups, yes, but start it with email and calendar only and approve every action yourself for the first week.

Until now, Claude mostly waited for you to type. Duet flips that: it comes to you. That is useful if it catches the thing you would have missed, and costly if it burns through your usage limits on check-ins you did not need. This post covers what Duet does, who gets it, and a sensible way to set it up.

What Claude Duet actually does

Anthropic describes Duet as an app "where Claude works as a proactive agent," aimed at professionals and small-business owners. Based on Anthropic's launch post, product page, and help center:

• A morning brief. Each morning it reviews your connected email and calendar, flags what matters, and can start first steps such as drafting replies for you to review.
• It checks in, then goes quiet. Through the day it checks your connected accounts at set intervals (Anthropic says every few hours, to save usage) and messages you only when something needs a decision. You can set what you want to hear about and how often.
• You can hand it work. Type or speak a task in one ongoing thread: set up a meeting, chase a stalled thread, build a sales forecast, a research brief, a slide deck, or a dashboard. Bigger results come back as editable files in a Make tab, and a Tasks tab tracks what is in progress. Only you can mark a task done.
• It works while you're away. Duet runs on its own cloud computer with its own browser, so your phone or laptop does not need to be on. For sites without a connector, it uses that browser.
• It uses your existing Claude setup. Your Claude memory, skills, and connected apps carry over. Anthropic says it works with more than 6,000 connectors and plugins, naming Gmail, Google Calendar, Google Drive, Slack, QuickBooks, HubSpot, DocuSign, Stripe, and Shopify, among others.

You cannot choose the model during the preview: Duet runs on Claude Opus 5.5. If you want background on that model, see our Claude Opus 5.5 guide.

Who can get Claude Duet, and what it costs

	Duet at launch?	Notes
Free	No
Pro	Yes, by waitlist	US accounts only
Max (5x and 20x)	Yes, by waitlist	US accounts only
Team, Enterprise	Not yet	Anthropic says it plans to add them later

Anthropic's help center adds a few details worth knowing before you sign up:

• Invitations go out in waves. Joining the waitlist does not guarantee a date.
• The email has to match. Access is tied to the email you give on the waitlist, which must match your Claude account or the Apple or Google account on your phone.
• Phones first. Duet is a standalone app on iPhone and Android. Anthropic's launch post says it is also available on claude.ai.

Cost: there is no separate price. Duet is included with Pro and Max during the preview. Pro is $20 a month ($17 a month billed annually) and Max starts at $100 a month, per Anthropic's pricing page.

The catch is usage. Duet "shares your plan's usage limits with the rest of Claude," including the five-hour and weekly limits. Its background check-ins count, and big jobs like a financial model or deck use more model turns. If you already hit your limits in regular Claude chats, expect to hit them sooner. Anthropic says new Duet users get one full usage-limit reset (covering both the five-hour and weekly limits) when admitted, available until November 30, 2026.

What it will and won't do on its own

This is the part to read before you connect anything.

It asks first for the risky stuff. Anthropic says Duet shows you the planned action and waits for approval before it sends a message, makes a payment, signs a document, or deletes anything.

Some things are off-limits even with approval. Anthropic names trading securities and moving money between accounts.

Purchases go through wallets, not your card number. Duet buys through Link (Stripe's wallet) or Shop Pay, each with your approval. Anthropic's safety page says purchases use a one-time virtual card limited to one merchant and amount, so your real card number is not shared with Duet or the merchant. Paying or sending money, deleting an order, and creating an account can never be set to "don't ask again."

It never sees your passwords. Per Anthropic's October 23 safety page, Duet "never sees your passwords, browser cookies, or the sign-in tokens for your connected apps." You type passwords into a form inside the Claude app, and they are passed to Duet's cloud browser without the model reading them. Anthropic says that sign-in form only appears inside the Claude app during a sign-in you started, never through a link in an email, text, or website. Treat any link that asks for your Claude or Duet login as a scam.

A second check watches what it sends. A separate classifier reviews each outgoing message and uploaded file. Duet is told never to share sign-in credentials, full card or bank account numbers, or government ID numbers, "even if you ask." Other private details, such as health or financial information, go out only with your consent (asking for it counts). Anthropic says this classifier will sometimes miss things and sometimes block harmless messages.

It can be fooled. Anthropic's help center says Duet treats content from emails and web pages as information, not instructions, but acknowledges that prompt injection (hidden instructions planted in an email or page) is still a risk. Anthropic says its approval check runs outside the model, so text hidden in a page cannot answer an approval prompt for you. That is the strongest reason to keep the approval step on and actually read what you approve.

Disconnecting is not deleting. Duet keeps its conversation history and memory until you delete them. Switching off an app stops new access, but what Duet already learned from it stays until you reset Duet's memory in the app's settings. Deleting your Claude account wipes everything Duet has stored. Anthropic's safety page does not give a retention period or say whether Duet data is used to train models.

Known limits in the preview, per Anthropic:

• It "still might misread a thread or miscalibrate what's urgent."
• Because it checks on a schedule, an urgent item can reach you late.
• It cannot make phone calls or send texts. It can send email from connected accounts, with approval.
• You cannot watch or take over its browser yet.
• It cannot see all your Claude projects or local files, and it can read but not edit your other Claude work.
• Microsoft 365 is not supported yet.
• It is weaker at personal errands, like booking tickets, because it was built around work.

How it compares with ChatGPT dots

Duet arrives three weeks after OpenAI's dots, and Reuters frames both as answers to Meta's Muse agent. The products overlap but differ in emphasis:

	Claude Duet	ChatGPT dots
Launched	October 21, 2026 (research preview)	September 29, 2026
Model	Claude Opus 5.5	GPT-6 Astra
Who gets it	US Claude Pro and Max, by waitlist	ChatGPT Pro (not EEA, Switzerland, UK) and Business Premium; Enterprise beta
Where you talk to it	Its own iPhone and Android app, plus claude.ai	ChatGPT, Slack, or Microsoft Teams
Built around	Daily brief, inbox and calendar triage, tasks you hand it	A standing goal you set
Extra cost	None during preview; shares plan limits	First dot included; OpenAI has not priced extra dots

If you already pay for one of these plans, the practical question is not which agent is "better" but which one sits on the apps you live in. Duet's lack of Microsoft 365 support today is a real gap for Outlook and Teams users.

A sensible way to start

If you get an invite:

1. Connect email and calendar only for the first week. Let it prove it reads your priorities correctly before you add accounting, CRM, or payment tools.
2. Approve every draft yourself. Read what it plans to send. The approval prompt is your main defense against both mistakes and prompt injection.
3. Know how to undo it. If you stop using Duet, disconnect the apps and reset its memory in settings; disconnecting alone leaves what it learned in place.
4. Watch your limits. Check your Claude usage after a few days. If the check-ins crowd out your regular work, narrow what it watches or how often it reports.
5. Check your employer's AI policy before connecting work accounts. Duet is offered on personal Pro and Max plans, not Team or Enterprise, so a company-managed account may not be the right place to try it.
6. Keep a human on anything urgent. Duet checks on a schedule; don't rely on it for time-critical alerts.

What we don't know yet

• When Team and Enterprise get Duet, and when it reaches countries outside the US.
• Whether pricing or usage rules change after the research preview.
• When Microsoft 365 support arrives.
• Whether Duet data is used for model training, and how long it is kept. Anthropic's October 23 safety page covers deletion but not either point.

We will update this post when Anthropic publishes those details.

Sources

• Anthropic, Introducing Claude Duet (October 21, 2026)
• Anthropic, Claude Duet product page
• Claude Help Center, Get started with Claude Duet
• Anthropic, Claude Duet Safety, Privacy, and Security (October 23, 2026)
• Anthropic, Claude pricing
• Reuters via WSAU, Anthropic unveils Claude Duet AI agent app (October 21, 2026)
• 9to5Mac, Anthropic takes on ChatGPT Dots and Meta Muse with Claude Duet (October 21, 2026)
• TechCrunch, Claude Duet app is Anthropic's bet on agents (October 22, 2026)
• The Verge, Anthropic takes on Muse with Claude Duet, its new AI agent (October 21, 2026)