Anthropic launches Claude Duet agent for US Pro and Max subscribers
Anthropic announced Claude Duet on 21 October, a proactive agent on Claude Opus 5.5 for US Pro and Max subscribers at $20 per month, with approval before payments or deletions.
Photo: The VergeKey points
Anthropic announced Claude Duet on 21 October, a proactive agent on Claude Opus 5.5 available as a research preview to US Pro and Max subscribers via waitlist.
Anthropic announced Claude Duet on 21 October, a proactive AI agent built around Claude Opus 5.5 and aimed at professionals and small business owners. The company is releasing it as a research preview, with access granted through a waitlist open to Pro and Max subscribers in the United States. Duet is available as a standalone app on iOS and Android and through Claude on the web.
Duet is designed to reach out before a user asks. Each morning it sends a short brief on the most important items across connected apps, then keeps tabs on those apps and messages the user only when something needs attention. Anthropic says it built Duet for people with too much to track, the group most likely to find it helpful.
What Duet does each day
The agent runs on its own cloud-based computer with a browser, so it keeps working after the user closes the app. It pulls in memory, skills and connected apps from a Claude account, so there is little setup. Interactions happen in one ongoing thread rather than across separate conversations, and users can start separate channels for different topics that share the same memory.
Anthropic says users can hand Duet tasks of all shapes, from scheduling meetings to building dashboards with dozens of tools, creating financial or sales forecasts, prototyping and building web apps. Larger results come back as files a user can review and edit, such as a research brief, a financial model, a slide deck or a dashboard. Finished work appears in a Make tab.
Before any big step, Duet messages the user for approval. Anthropic lists sending a message, making a payment, signing a document and deleting a file among the actions that require a check-in. The agent also has a Tasks tab where users track ongoing work and see which tasks need attention, and only users can mark tasks as finished.
Approval before sensitive actions
Anthropic says Duet has a new dedicated classifier and safety designed to reject requests to share the strictest tier of information, such as credit card numbers. The company says Duet does not see passwords or sign-in tokens when given access to accounts, and information that might relate to health or personal finances is let through only if a user explicitly tells Duet to share it.
Duet supports plugins, including a small business plugin, and can suggest ones that fit a user's work. It also offers integrations with Stripe to check for unpaid invoices and draft payment reminders, and with Link and Shop Pay to make purchases with user approval. For sites without a connector, Duet uses its own browser in the cloud.
Duet uses a user's existing Claude plan limits, and work performed in the background counts toward them. Anthropic says checking connected accounts every few hours avoids using up plan limits as fast as checking nonstop would, and that it is continuing to test the agent's judgment about when to message users. Users can set how frequently Duet messages them and on what topics.
Limits and availability
Anthropic says subscribers of its $20 per month Pro plan will get access once Duet rolls out in general availability. That differs from OpenAI's Dots, which are currently available only to OpenAI's Pro tier subscribers, whose plans start at $100 per month. Duet is not available on Team or Enterprise plans yet, though Anthropic plans to expand availability to those subscriptions in the future.
Anthropic describes Duet as an early version that still might misread a thread or miscalibrate what is urgent, and says it is not as strong at personal errands such as booking tickets. The company says it will have more information about Duet's safety protections to share in the coming days. Invitations go out in waves to users who join the waitlist.
Duet is the third major proactive agent to reach users, following Meta's Muse and OpenAI's Dots. Anthropic is billing Duet as a proactive agent that comes to users with what needs their attention, completes tasks, keeps working when they step away and checks with them before acting on their behalf. The company announced it on 21 October.
Anthropic says it will keep tuning Duet's judgment about when to message users based on what testers tell it, and that it plans to expand availability beyond Pro and Max subscribers in the future. The company says it will share more information about Duet's safety protections in the coming days.
Frequently asked questions
What is Claude Duet?
Claude Duet is a proactive AI agent Anthropic announced on 21 October. Built around Claude Opus 5.5, it sends a morning brief, checks connected apps every few hours, completes tasks and asks for approval before actions such as sending a payment or deleting a file.
Who can use Claude Duet?
It is a research preview available through a waitlist to Pro and Max subscribers in the United States, as a standalone app on iOS and Android and through Claude on the web. It is not available on Team or Enterprise plans yet.
How does Claude Duet protect sensitive information?
Anthropic says Duet has a dedicated classifier and safety designed to reject requests to share the strictest tier of information, such as credit card numbers. It does not see passwords or sign-in tokens, and health or personal finance information is shared only if a user explicitly allows it.


