Anthropic opened a research preview of Claude Duet on October 21, 2026: a standalone agent app aimed at professionals and small business owners. Claude Duet is meant to surface what needs attention, keep working when the user steps away, and ask for confirmation before acting on the user's behalf. Access starts with a waitlist for Pro and Max subscribers in the United States.
Key Facts
| Details | |
|---|---|
| Status | Research preview; waitlist open |
| Platforms | Standalone iOS and Android apps, plus claude.ai |
| Eligibility | US Pro and Max plans (Anthropic). The Decoder reports Team and Enterprise are not included yet |
| Usage | Included in Pro/Max; shares the same usage limits as the rest of Claude, including background checks |
| Model | Built around Claude Opus 5.5, per Anthropic's blog |
| Behavior | Morning brief from connected apps; messages when something needs attention; virtual machine with its own browser so work continues offline for the user |
| Approvals | Shows the planned action and waits before sending messages, making payments, signing documents, or deleting. Trading securities and moving money between accounts stay blocked even with approval |
| Connectors | Over 6,000 Claude Marketplace connectors and plugins (Gmail, Slack, QuickBooks, Stripe, Gusto, LinkedIn, DocuSign, HubSpot, Canva, Shopify, and others cited by Anthropic) |
| Payments | Can buy via Stripe Link with a virtual card, or Shop Pay; real card numbers are not shared. Sign-in with saved 1Password logins is listed as coming soon |
Background
Anthropic positions Claude Duet as a proactive partner rather than a chat window the user has to open first. Memory, skills, and connected apps carry over from the existing Claude account. Users talk to it in one ongoing thread by text or voice.
The company says Duet draws on its trustworthy-agents framework, Managed Agents credential handling (so the agent does not see passwords or sign-in tokens directly), the auto mode classifier used elsewhere in Claude and Claude Code, and browser-use research against prompt injection. Red teams and an external firm penetration-tested the product, and it is covered by Anthropic's public bug bounty. A new classifier and safety system prompt are meant to reject requests to share credit card numbers and government identification; health or personal finance details only go out when the user explicitly instructs Duet to send them.
TechCrunch and The Verge place the launch alongside other agent products such as Meta's Muse and OpenAI's Dots. Anthropic itself stresses that this early build is weaker at personal errands like booking tickets, cannot yet show its browser session, and lives in a separate app from the main Claude app.
Reactions and Impact
Anthropic published short quotes from early testers, including small-business owners who used Duet for morning rundowns, invoice work, and meeting briefs. Those accounts are Anthropic's own marketing quotes, not independent reviews.
Every.to published a "Vibe Check" review of Duet as an always-on personal agent (linked in Sources). Outside that, independent hands-on coverage was still thin at the time of writing.
What We Don't Know Yet
- When the research preview ends, or when Duet becomes generally available outside the US waitlist.
- Whether Team or Enterprise plans will get access, and on what timeline (official materials only name US Pro and Max).
- How usage burn from background checks compares with typical Claude chat use in practice.
- How Duet will eventually merge with, or stay separate from, the main Claude app.