Anthropic launches Claude Duet, a background AI agent, for US users

Anthropic, the US company behind the Claude family of artificial intelligence models, launched Claude Duet on Wednesday, October 21, 2026. Claude Duet is a personal AI agent that, according to the company, works in the background on a user's email, calendar and other connected apps, and alerts the user when something needs attention. It is a research preview, an early test version, and for now is open only to paying Pro and Max subscribers in the United States through a waitlist.

Most people use AI chatbots by typing a question and waiting for an answer. An AI agent is different: it is software that can take steps on its own, such as reading new emails, checking a calendar or drafting a reply, without being asked each time. Claude Duet is what the industry calls a proactive agent. Instead of waiting for a prompt, it checks the user's connected accounts from time to time, prepares a morning briefing, and flags items that need a decision. The company says it can also take on larger jobs such as building a financial model or a slide deck.

The launch puts Anthropic in direct competition with Meta and OpenAI, both of which released personal agent products in September 2026. The race matters because the company whose agent sits on a person's phone and inbox gains a daily habit, a steady subscription and a view of how people work. It also raises the stakes on safety. An agent that can send messages or make payments can do real harm if it misunderstands an instruction. Anthropic's answer is to require the user's approval before any such action, and its support pages state that Duet does not see the passwords or logins of the accounts it connects to. These are the company's own descriptions; independent testing of how well the safeguards work is not yet available.

For India, the product is not yet on offer. Indian users of Claude cannot join, because access is limited to the US. Indian companies and developers will still watch closely, because agents of this kind are expected to change office work such as scheduling, reporting and customer follow-up, which employs large numbers of people in India's technology and services industry. Questions about how such agents handle personal data will also matter under India's Digital Personal Data Protection Act once products like this arrive here.

Short-term: US subscribers admitted from the waitlist can try Duet as a test product. Early users will show whether its briefings and alerts are useful or noisy, and whether the background checks use up their plan limits too quickly.

Long-term: If proactive agents prove reliable, much routine office coordination could shift from people to software, with humans approving rather than doing each step. That would raise new questions for regulators about consent, data access and who is responsible when an agent makes a mistake.

Who is affected: Claude subscribers in the US, rival AI companies such as Meta and OpenAI, businesses that may later deploy such agents for staff, and, in time, Indian users and companies once the service is offered outside the US.

Questions and Answers