Skip to content

An editorial team and a community building with AI

← All articles

News · Anthropic · 2 min read

Event date: · Published:

Anthropic opens a research preview of Claude Duet, a proactive agent app

On October 21, 2026, Anthropic opened a waitlist research preview of Claude Duet, an app in which Claude works as a proactive agent that surfaces what needs attention and checks in before acting on the user's behalf.

Archive entry: this covers an event from October 21, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On October 21, 2026, Anthropic opened a waitlist research preview of Claude Duet, an app in which Claude works as a proactive agent. Anthropic described it in a post on the Claude blog.

What Duet does

  • Duet surfaced what needed the user's attention, completed the tasks the user asked for, kept working when the user stepped away and checked with the user before acting on their behalf.
  • Anthropic designed it around Claude Opus 5.5. Each morning it sent a short brief on the most important items across the user's connected apps, and during the day it messaged the user only when something needed attention.
  • Duet was available on iOS, Android and claude.ai.
  • During the research preview, Duet used the same usage limits as the rest of Claude.

What it means for people building with AI

Duet was Anthropic's first app built around Claude acting proactively for users. That matters for anyone building or evaluating always-on personal agents, because the design Anthropic described (a morning brief, messages only when something needed attention, and a check-in before acting on the user's behalf) shows one concrete approach to how such an agent can behave.

Availability and safeguards

On eligibility, the Claude Help Center release notes named the Pro and Max plans, while the blog post, as published, said the preview started with US Pro, Max and Ultra subscribers.

Anthropic said Duet's handling of credentials followed the design it used for Claude Managed Agents, so Duet never directly saw passwords or sign-in tokens, and that its actions were approved by Anthropic's auto mode classifier. It also said red teams and an external firm penetration tested Duet, which was covered by Anthropic's public bug bounty program. An October 23, 2026 update on the blog post links to a separate article on Duet's safety, privacy and security.


Sources

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.

Anthropic opens a research preview of Claude Duet, a proactive agent app | vibecoding.inc