Particle.news
Get it on Google Play
Download on the App Store
22 ARTICLES
·
2h ago

Anthropic Launches Claude Duet as a Proactive AI Agent

Built on Opus 5.5 with a cloud virtual machine; the research preview automates business workflows, requiring user approval for sensitive actions.

Image
Anthropic / Wikimedia Commons (Public domain)
Claude logo and a keyboard in this illustration taken October 19, 2026. REUTERS/Dado Ruvic/Illustration
Ami Vora, Anthropic's chief product officer, and Daniela Amodei, its president.

Overview

  • Anthropic opened a waitlisted research preview of Claude Duet for US Claude Pro and Max subscribers on Oct. 21, 2026, with access via standalone iOS and Android apps and on the web.
  • Duet runs on Claude Opus 5.5 in its own cloud virtual machine so it can keep working when your device is off and it reuses a user’s existing Claude memories, skills, and connectors.
  • The agent checks connected apps, sends a daily morning brief, surfaces items that need attention, and can perform tasks from scheduling and drafting replies to building dashboards and financial forecasts.
  • Anthropic requires Duet to ask users before taking consequential actions such as sending messages, making payments, signing documents, or deleting data and adds a classifier to block the strictest sensitive data.
  • The preview is limited: background checks and work count against existing plan usage limits, Team and Enterprise customers are excluded for now, and Anthropic warns Duet may misread threads or misjudge urgency in this early version.