Anthropic launches Claude Duet: proactive AI assistant

Anthropic has presented the launch of Claude Duet, an innovative AI assistant capable of autonomously identifying and executing important tasks when the user is absent. This new application generates daily morning briefs, monitors connected applications, and executes commands. The first beta version is available for US Pro and Max subscribers through a waitlist.

Claude Duet is specifically designed for professionals and small entrepreneurs dealing with an abundance of tasks. The official introduction took place on October 21, 2026. According to Anthropic, the assistant employs a proactive approach, where it not only waits for instructions but also identifies issues that require attention and completes tasks based on pre-given commands.

The functionalities of Claude Duet are versatile; users can schedule appointments, make sales forecasts, build dashboards, and develop web applications. It is possible to issue a command and later view the progress or outcome. The assistant provides a summary of urgent customer queries in the morning while also preparing reports. The reliance on connected applications determines the effectiveness of these briefs.

Claude Duet is developed based on Claude Opus 5.5 and is available via a separate app on iOS and Android, as well as through the platform claude.ai. Importantly, the assistant requests permission before acting on behalf of the user. Additionally, Anthropic has taken measures to protect sensitive information by ensuring that passwords and login tokens are not directly accessible to the assistant. The outcome of the research trial will further clarify the reliability of these security measures.

For Dutch users, no specific introduction date for Claude Duet has been established yet. Access is currently limited to US Pro and Max subscribers who can sign up via a waitlist for the beta version. This means that Dutch IT decision-makers need to stay alert for further developments surrounding this advanced AI solution.

Read the full article from AI Insider.