Claude Duet takes the initiative! Anthropic launches an agent that alerts you before you ask
Anthropic introduced Claude Duet, a standalone app that aims to change the usual relationship with artificial intelligence assistants: instead of waiting for instructions, it checks connected tools and alerts users when it detects relevant matters. The product is designed for professionals and small businesses, uses Claude Opus 5.5 and can continue tasks in a remote environment. For now, it is a research preview with a waitlist for Pro and Max subscribers in the United States.
From reactive chatbot to proactive assistant. On 21 October, Anthropic announced a preview of Claude Duet, which can prepare a morning summary of connected applications and periodically check whether anything needs attention. Users can also assign it tasks such as organizing meetings, preparing forecasts, building dashboards or developing prototypes. Unlike a conversation that depends on the device being switched on, the agent has a virtual machine with its own browser so it can keep working in the background.
Autonomy has deliberate limits. Duet asks for authorization before sending messages, making payments, signing documents or deleting files; it also does not allow certain financial operations, even with approval. Anthropic explains that it includes controls to reduce information leaks and attacks through hidden instructions. However, it acknowledges that this version may misinterpret the urgency of a task or the context of a conversation. Initial access is limited to Pro and Max users in the United States through a waitlist, and background work counts toward the plan's usage limits.


