Claude Duet is a proactive AI agent app from Anthropic, opened as a research preview on October 21, 2026. Built around Claude Opus 5.5, it sends a morning brief of what needs your attention across connected apps, watches those apps during the day, takes on tasks from scheduling meetings to building dashboards and web apps, keeps working in its own cloud virtual machine when your devices are off, and asks for approval before it sends messages, pays, signs or deletes anything.

Who can use Claude Duet and how much does it cost?

During the research preview Claude Duet is included in Claude Pro and Max plans for users in the United States, with access through a waitlist. There is no separate price, but it shares your plan's usage limits with the rest of Claude, and its background checks and larger tasks such as financial models or decks use more of those limits than a normal chat.

What can Claude Duet not do?

Anthropic says some actions are off-limits even with your approval, such as trading securities or moving money between accounts. In the preview it cannot yet show you its browser, it is weaker at personal errands such as booking tickets, it may misread a thread or misjudge urgency, and it lives in a separate app from the main Claude app.

How does Claude Duet handle security and sensitive data?

Anthropic says Duet never directly sees your passwords or sign-in tokens, following the credential design of Claude Managed Agents; actions pass its auto mode classifier; prompt-injection defences come from its browser-use research; and a new classifier and safety system prompt reject requests to share the strictest tier of data, such as credit card numbers and government ID. Purchases go through Stripe's Link with a virtual card, so your real card number is not shared.

Quick Answer

What Is Claude Duet? Anthropic's Proactive Agent Explained

Published: October 25, 2026

The short answer

Claude Duet is Anthropic’s always-on assistant app: instead of waiting for prompts, it comes to you with what needs attention, works on tasks in the background on its own cloud computer, and waits for your approval before anything consequential. It launched as a research preview on October 21, 2026, runs on Claude Opus 5.5, and is available by waitlist to Pro and Max subscribers in the US, on iOS, Android and claude.ai.

What it does

According to Anthropic’s announcement:

Morning brief. Each morning it summarises the most important items across your connected apps, then messages you during the day only when something needs you.
One thread for everything. You message it by text or voice; a quick reply and a week-long project start in the same conversation. It carries over your memory, skills and connected apps from your Claude account.
Background work. Duet has a virtual machine with its own browser, so tasks continue with your laptop and phone off. It checks connected accounts every few hours rather than constantly, to save your plan’s limits.
Real tasks. Anthropic’s examples: replying to missed Gmail and Slack messages, chasing unpaid invoices in QuickBooks or Stripe, preparing payroll in Gusto, posting jobs to LinkedIn, sending agreements through Docusign, drafting follow-ups from HubSpot, social campaigns in Canva and managing a Shopify store — using over 6,000 connectors and plugins in the Claude Marketplace.
Purchases. Through Stripe’s Link wallet with a virtual card, and Shopify’s Shop Pay. Signing in with logins saved in 1Password is “coming soon”.

The approval rules

Duet reads, drafts and prepares on its own. Before it sends a message, makes a payment, signs a document or deletes anything, it shows you exactly what it will do and waits. Some actions are blocked even with approval — Anthropic names trading securities and moving money between accounts. You can disconnect any app at any time.

Safety design

Anthropic lists five layers: its framework for trustworthy agents (shared with Claude Code and Claude in Chrome); the Claude Managed Agents credential design, so Duet never directly sees passwords or tokens; the auto mode classifier that approves actions; prompt-injection defences from its browser-use research; and red-teaming plus an external penetration test, with Duet covered by Anthropic’s bug bounty. New for Duet is a dedicated classifier and safety system prompt that rejects sharing the strictest data tier (card numbers, government ID); health and personal-finance details go out only when you explicitly instruct it.

Access, cost and limits (as of October 25, 2026)

	Detail
Status	Research preview, launched October 21, 2026
Who	Claude Pro and Max users in the US, via waitlist
Price	Included in Pro and Max; no separate fee
Usage	Shares your plan’s limits; background checks and big tasks count
Model	Claude Opus 5.5
Apps	Standalone iOS and Android apps, plus claude.ai
Known gaps	Can’t show its browser yet; weaker on personal errands; may misjudge urgency

Is it for you?

Duet is aimed at professionals and small-business owners with more threads than they can track. If you mostly want a coding agent, Claude Code remains the tool; if you want an agent you can script and deploy for customers, look at the API and Managed Agents. If your limits are already tight, expect Duet to consume more of them — see our notes on Claude Code weekly limits.

Duet enters a field that already includes always-on agents from OpenAI and Meta; for how those compare, see OpenAI Dots vs Meta Muse vs Microsoft Autopilot vs Manus Cue.

Last verified: October 25, 2026.