Anthropic has announced the launch of a research preview of Claude Duet — a personal AI agent that runs in the background and offers proactive actions based on connected applications.
How Claude Duet works as a proactive agent
Unlike standard chatbots, Duet creates a daily brief listing actions based on connected services and only reaches out to the user when direct involvement is required. The agent automatically analyses the context but does not communicate continuously.
Users will be able to adjust the frequency of notifications and the topics the agent should report on. To reduce quota consumption, Duet checks connected accounts every few hours instead of monitoring continuously.
Pricing and usage limits for Claude Duet
After full release the agent will be available within the standard Pro plan for $20 per month, while similar functionality in OpenAI Dots is available on plans starting from $100/month. The main advantage cited is its availability within a Pro subscription.
Duet shares usage limits with other Claude tools, including Claude Code, so the agent’s activity affects the overall quota. These settings let users control consumption of those limits by adjusting check frequency and topics.
Privacy, security and availability
The agent does not have direct access to passwords or authorization tokens, and health data or personal financial information is shared only with the user’s explicit consent. To prevent leaks, the company has implemented prompt protection and additional classifiers.
A detailed privacy protection report is promised to be published soon. For now, Claude Duet is available only to Claude Pro subscribers in the US who have signed up to the waiting list.
