Daily AI CatchupAnthropic Launches Claude Duet, a Proactive Agent
Anthropic has released Claude Duet, a research preview of a proactive AI agent built on Opus 5.5. It monitors connected apps, decides what needs attention, and works autonomously, requesting approval before certain actions rather than waiting to be asked. It is starting as a waitlisted preview for US Pro and Max subscribers.
Why it matters
💻 Developer · A new agent pattern worth studying: autonomous monitoring plus approval gates, rather than a chat window you have to prompt.
📦 Product · Proactive agents that decide what to surface could reshape how notification and task-management features get designed.
🎨 Design · The 'ask before acting' approval flow is the interesting UX bet here: how do you signal trust without annoying users?
📈 Business · If this works, it points toward AI assistants that manage your inbox and tools without constant prompting, a real productivity shift if reliable.
🤔 Just Curious · Think of it as an AI assistant that checks your apps and decides what needs your attention before you even ask.
Sources: Claude Duet