Anthropic’s new AI agent Claude Duet can work while users are away

Anthropic has launched Claude Duet, a new artificial intelligence agent that can manage tasks on its own. The launch comes as the company competes in the growing market for AI assistants that do work for people.

Claude Duet is designed for professionals and small business owners who need help organizing their work, managing connected apps and completing tasks with few instructions. Unlike traditional chatbots, which mainly respond to questions and requests, Claude Duet can check connected apps, find problems that need attention and keep working in the background.

The agent runs on Anthropic’s Claude Opus 5.5 model. Every morning, it prepares a briefing about important updates from the accounts linked to it. It keeps checking those apps throughout the day and alerts users when they need to make a decision or take action.

Anthropic says the agent is designed to stop private information from being shared without permission. It will share details about health or personal finances only when users clearly tell Claude Duet to send them. The agent can also help users buy things online through Link, Stripe’s digital wallet, and Shopify’s Shop Pay.

“Claude Duet confirms the item with you, and Link asks you to approve the spend. It then pays with a virtual card, so your real card number isn’t shared,” Anthropic said in its blog post.

Users can give the agent many kinds of tasks. These include setting up meetings, writing follow-up emails, building dashboards, preparing financial forecasts and creating web applications. Claude Duet works inside its own virtual machine and browser, so it can keep working even when users leave their devices.

The agent can also connect to more than 6,000 connectors and plugins in the Claude Marketplace. These connections let it work with many different business apps.

Claude Duet is available as a research preview through a waitlist for subscribers to Anthropic’s Pro and Max plans in the U.S. The standalone app can be used on iOS, Android and the web through Claude.ai.

Anthropic says the product has safety measures to help users stay in control of important decisions. Claude Duet asks for permission before sending messages, making payments, signing documents or deleting information. By default, it refuses requests to share highly sensitive information, such as credit card numbers and government identification details. Some actions, including trading securities and moving money between accounts, are not allowed even if users approve them.

The launch comes as AI companies compete to build agents that can handle several tasks and reduce people's workloads, rather than simply write text or answer questions. Claude Duet joins a growing market for AI assistants that take action on their own, alongside products from other major technology companies.

However, as AI companies build more advanced agents, concerns remain about data breaches, actions taken without permission and security weaknesses. One reported incident involving Meta’s Muse AI raised questions about the risks of AI agents that act on their own. The system reportedly negotiated a deal on Facebook Marketplace, shared a user's address and arranged a late-night pickup without the user's approval.