Anthropic Unveils Claude Duet Personal Agent To Rival OpenAI Dots, Meta Muse

Anthropic has introduced Claude Duet, a personal agent designed to compete with OpenAI's Dots and Meta's Muse. This announcement comes as part of a research preview, and the agent is currently only accessible to a limited number of users. Built on Opus 5.5, Claude Duet leverages information from users' Claude accounts, such as memory, skills, and connected applications.

Available as a standalone app for both Android and iOS devices, the agent operates independently of Claude sessions, running on its own virtual machine and running in the background.

Anthropic claims that Claude Duet can help users prioritize their time and energy by generating a list of crucial tasks each morning based on data from connected apps. The agent can also monitor apps and messages throughout the day, alerting users to any issues that require attention. Users can assign the bot various tasks, such as scheduling meetings and developing applications, though approval is needed for activities like making payments and sending emails.

To ensure user data safety, Claude Duet utilizes a new classifier and safety system prompt. This system is intended to prevent the leakage of sensitive information, such as credit card numbers, and requires explicit user instructions for sharing other sensitive data. The agent itself does not directly access passwords or sign-in tokens.

Anthropic also commits to further enhancing security measures, promising to provide additional details on safety features in the future. As using the tool involves entrusting the bot with extensive personal information, Claude Duet might not be suitable for privacy-conscious users. Interested users must join a waitlist, which is limited to US subscribers.