The key facts
- Claude Duet runs on Claude Opus 5.5 and manages tasks: from scheduling meetings to financial forecasts and building web applications
- The agent operates through a virtual machine with a browser and keeps working even when the user's device is off
- Access via a waitlist, for now only for Pro and Max subscribers in the US
Anthropic Releases Claude Duet
American company Anthropic on Wednesday introduced Claude Duet — a proactive AI agent designed to independently manage tasks on behalf of users. The product runs on the Claude Opus 5.5 model and is aimed at professionals and small business owners. According to the company, the agent offers daily briefings and takes on tasks — from scheduling meetings to preparing financial forecasts and building web applications.
How the Agent Works
“Claude Duet proactively tells you what needs your attention, completes delegated tasks, keeps working even when you’ve stepped away, and checks in with you before acting on your behalf,” Anthropic’s announcement says. The agent operates through a virtual machine with a browser, allowing it to continue working even when the user’s device is off. The company states that before sending messages, making payments, signing documents, or deleting information, the agent requires user approval. Safeguards have also been introduced against unauthorized disclosure of sensitive data, including bank card numbers and government ID details.
Availability and Context
On Wednesday, Anthropic opened a waitlist for the research preview of Claude Duet, which at launch is limited to Pro and Max subscribers in the US. The company has not yet announced timelines for broader access. The launch came after Meta introduced the Muse agent in September, and OpenAI introduced Dots: major AI companies are competing to build proactive assistants capable of completing tasks on behalf of users.
