Claude Duet: Anthropic’s Powerful AI Agent That Keeps Working After You Close the App

• Claude Duet keeps working in the cloud after you close the app, handling connected tasks in the background.
• The AI agent can work with connected email and calendar apps and sends a brief each morning.
• Claude Duet is a US research preview for Pro and Max users, with India availability not yet offered.

Claude Duet is Anthropic’s powerful AI agent designed to keep working after you close the app. You can message it by text or voice, then let it continue tasks in the background using its own computer in the cloud.

The agent works inside apps you connect, including email and calendar services, and can use its own browser for websites without a connector. Anthropic says the agent is powered by Claude Opus 5.5 and will initially be available as a research preview through a waitlist.

How Claude Duet Works

The agent is designed to take on tasks that may continue beyond a single chat session. Users can give it instructions in natural language, either by typing or speaking, rather than managing each step manually.

It operates on its own cloud computer, which lets work continue even when the app is closed. This background setup is central to the service: the agent can keep moving through a task without requiring the user to leave the app open.

The agent can work with apps that users connect, such as email and calendar tools. For websites that do not offer a connector, it can use its own browser to access the site and carry out supported steps.

Duet also sends a short brief every morning highlighting what needs attention. This daily summary is intended to help users see important items and pending work without having to ask the agent for a fresh update each time.

Approval Required for Important Actions

Although the agent can work in the background, it does not have unrestricted permission to complete every action. It asks the user first before sending a message, making a payment, signing a document or deleting anything.

These approval prompts create a checkpoint before actions that could have a direct impact. Users should still review what the agent proposes and make sure the details are correct before approving a message, payment, signature or deletion.

Availability and Plan Limits

The service is available as a research preview on iOS, Android and claude.ai for Pro and Max subscribers in the United States. Access is through a waitlist, so eligible users may need to join it before they can try the agent. Details are available on the official Claude website.

Anthropic says there is no additional charge, but background work counts towards the limits of the user’s existing plan. Users should keep those plan limits in mind when assigning longer-running tasks.

The research preview is not available in India yet. Availability may expand later, but no wider launch timing is included in the information announced so far.

How the Agent Uses Claude Opus 5.5

The agent is powered by Claude Opus 5.5. Its main distinction is the way the agent combines conversational instructions with background work, connected apps, browser access and a daily summary. The service is designed to keep tasks moving after the initial request rather than ending when a chat is closed.

For context, MobileTelco has covered AI developments and tools as well as technology updates on MobileTelco. Its initial availability is limited to a research preview for Pro and Max users in the US.

Conclusion

The agent brings background task execution to Anthropic’s assistant, with connected-app access, browser use and a morning brief. It requires approval for important actions and is currently limited to a US research preview for Pro and Max users, with background activity counting towards plan limits. It is not available in India yet.

Frequently Asked Questions