Claude Duet turns the agent proactive: a morning brief, and your approval before it acts

Claude Duet reverses the direction of agent interaction: instead of waiting for a question, it brings items needing attention to the user and keeps working after the user steps away. Claude had been used conversationally, with the user initiating and watching the process; Claude Duet is positioned as a proactive agent for professionals and small business owners with more on their plates than they can track. Anthropic released it as a research preview, opening a waitlist to US Pro and Max users, and says it gave small business owners and entrepreneurs early access.

It works through a virtual machine with its own browser, and the user describes a task in one ongoing thread by typing or voice while it works out which apps to use and what steps to take. Tasks no longer require the user's laptop or phone to be on; a notification arrives when the task is complete or needs review. It can use the over 6,000 connectors and plugins in the Claude Marketplace, covering concrete settings such as Gmail, Slack, QuickBooks, Stripe, Gusto, LinkedIn, Docusign, HubSpot, Canva and Shopify.

Important actions stay with the user: before sending a message, making a payment, signing a document or deleting anything, it shows exactly what it is about to do and waits for approval. Some operations are off-limits even with approval, such as trading securities or moving money between accounts, and the user can disconnect any app. Credential handling follows the Claude Managed Agents design so it never directly sees passwords or sign-in tokens, and action approval uses the auto mode classifier that also protects Claude and Claude Code.

For sensitive information, a new dedicated classifier and safety system prompt check each message and reject requests to share the strictest tier, including credit card numbers and government identification. Other sensitive information, such as health or personal finances, is let through only when the user has explicitly instructed Claude Duet to send it. Anthropic says red teams and an external firm penetration tested Claude Duet's security infrastructure, and that it is covered by the public bug bounty program.

This is a research preview for professionals and small business owners, starting with a waitlist for Pro and Max users in the US, and it shares the plan's usage limits with the rest of Claude. Anthropic states that background checks count toward those limits and that larger tasks, such as building a financial model or a deck, require more model turns and background work.

Anthropic calls this an early version that might misread a thread or miscalibrate what is urgent; its core functionality is built around work, so personal errands such as booking tickets are not quite as strong, and it cannot yet show its browser. It currently lives in its own app, separate from the main Claude app, and checks connected accounts every few hours because checking nonstop would use up plan limits much faster; its judgment about when to message users is still being tested and tuned.