Anthropic introduces Claude Duet, a proactive AI agent for professionals
SiliconANGLE Kyt Dotson ● Covered by 7 sources
Anthropic launched Claude Duet, an AI agent that works through your apps and checks in on its own. It’s built to act like a secretary, but it’s still an early preview that can misread what matters.
Based on reporting by SiliconANGLE, Kyt Dotson — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Anthropic on Wednesday introduced Claude Duet, a proactive AI agent aimed at professionals and small business owners. It’s in research preview, and the pitch is simple: give it work, then let it keep going in the background instead of waiting for the next prompt.
Duet is available as a standalone app on the web, iOS and Android for Pro and Max subscribers. Anthropic says users can talk to it by text or voice without any special setup, and the system can pull from connected apps, calendar, email and other information it already has access to through Claude.
The company is pushing Duet as a kind of digital secretary. It can send a short brief each morning with the most important items across connected apps, watch messages only when something needs attention, and handle administrative jobs like scheduling meetings. Anthropic also says it can do heavier work, including building dashboards with interactive tools, creating financial or sales forecasts, and prototyping or building web apps.
It keeps a running history in one thread, remembers what users said earlier, and returns finished work as a doc, sheet or deck for review. Anthropic also says the assistant is built on its framework for trustworthy agents, the same approach it uses for systems like Claude Code and Claude in Chrome. An automatic classifier is meant to screen actions for sensitive information leaks and check with the user before important steps.
This is still an early product, and Anthropic is saying that plainly. The research preview is meant to show how people actually use Duet and where it fails. The company says it can still misread a thread or decide urgency badly, and that personal errands like booking tickets won’t be treated as urgently as work. It also can’t show its browser. Anthropic says the early feedback will shape how the assistant works next.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of AI ambition: boring, useful, and a little nosy. The real test isn’t whether it can build a dashboard, it’s whether it can avoid becoming one more overconfident assistant that treats every message like a fire drill. Anthropic gets points for admitting it’s not ready for the crown jewels yet.
Read more about this at: SiliconANGLE