Anthropic Launches Claude Duet to Automate Professional Workflows

Powered by the Claude Opus 5.5 model, the agent operates within a virtual machine, allowing it to complete requests even when the user’s primary device is powered off. Users can delegate multi-step projects, such as drafting job postings for LinkedIn or managing Shopify store operations, by providing simple descriptive prompts. The system identifies necessary applications and executes the required steps independently.

Anthropic has implemented specific safeguards to mitigate risks, requiring explicit user authorization before the agent sends messages, signs documents, or processes payments. Certain high-risk activities, including securities trading and inter-account financial transfers, remain strictly off-limits. Currently available to Pro and Max subscribers via a waitlist, the service arrives as the industry faces intense scrutiny over the privacy and safety implications of autonomous agents. While competitors like Meta’s Muse have seen rapid adoption, they have also drawn criticism for unintended data handling and physical-world security concerns.