Anthropic previews Claude Duet personal AI agent for subscribers

Anthropic has unveiled Claude Duet, a new AI agent designed to operate autonomously on behalf of its human subscribers. The feature competes with other AI agents like Meta's Muse and OpenAI's Dots, which have already gained popularity among consumers. Currently, Claude Duet is only accessible to U.S. Pro and Max subscribers via a waitlist.

The AI agent functions as a proactive assistant, automatically addressing tasks assigned to it. It takes actions independently, completes tasks as requested, remains active even when the user is absent, and consults with the user before acting on their behalf. Anthropic highlights Claude Duet's utility for professionals and small business owners, noting its superior judgment and ability to prioritize tasks.

Each morning, Claude Duet sends its user a brief highlighting the most important daily tasks. To function effectively, the agent must be connected to the user's apps, such as email accounts and calendars. Anthropic acknowledges security concerns raised by some professionals regarding AI agents' access to sensitive information and reassures users that Claude Duet's security measures prevent it from directly viewing passwords or sign-in tokens.

The agent can manage various tasks, including scheduling meetings, creating financial or sales forecasts, building web applications, drafting job postings, and even making purchases on behalf of the user. Anthropic researchers have tested Claude Duet for managing personal to-do lists, building applications, scheduling interviews, and more.

Interested users can sign up for a waitlist on Anthropic's Claude Duet product page. Invitations are issued in waves, and the product is currently only available in the U.S. during its research preview.