Anthropic has launched Claude Duet, an AI agent that works on its own schedule instead of waiting for you to type a prompt.
The company is calling it a "research preview", a label that usually means rough edges are expected. Duet runs as a standalone app on iOS and Android and through Claude on the web. It is built around Claude Opus 5.5 and aimed at professionals and small business owners.
Anthropic described the routine this way. Each morning, Duet sends a short brief on the most important items across your connected apps. During the day it keeps tabs on those apps and messages you only when something needs your attention. You can also hand it tasks directly.
The pitch is a change in who starts the conversation. A chatbot answers when asked. Duet watches and speaks up, which makes it closer to an assistant than a search box. Anthropic is entering a field where Meta's Muse and OpenAI's Dots already compete, so the rivalry is now over agents, not just models.
That shifts what counts as the hard part. Raw model quality matters, but the product lives or dies on a judgment call: what "needs your attention" means for a given person. Too many alerts and the brief becomes noise. Too few and a missed message costs real money for a small business. That is also why a research preview is a sensible label here.
The source leaves several practical questions open, including pricing, which apps Duet connects to, and how much access it needs to your accounts. For a tool meant to read across your work life, those answers may matter more than the model underneath.