Anthropic’s Claude Duet adds always-on agent to Pro, Max, Ultra plans
Anthropic’s Claude Duet is an always-on AI agent that proactively checks connected apps—like Gmail, Slack, and Google Drive—and sends morning briefs or alerts without manual input. Unlike traditional chatbots, Duet runs in the background on a virtual machine, using OAuth for security and sharing Claude’s usage limits. It’s currently in research preview for U.S. subscribers on Pro, Max, and Ultra…
Key points
- Claude Duet is Anthropic’s always-on agent for Pro/Max/Ultra users, checking apps and sending briefs without manual prompts
- Uses Claude’s usage limits but drains them faster due to background checks and tasks, per tester reports
- Zapier integration adds 9,000+ app connections and automates repetitive tasks while Duet handles variable workflows
Duet’s core features include proactive task suggestions, a Tasks tab for background work, and a Make tab for project templates like email triage. It asks for approval before sending, paying, signing, or deleting anything, though sensitive actions (e.g., trading securities) remain restricted. Users report quick setup but note rapid token consumption—background checks and tasks drain Claude’s session limits faster than expected. Anthropic emphasizes safety measures like prompt injection screening and red-team testing, but the agent’s accuracy depends on user oversight, especially for nuanced priorities. Zapier integration expands connectivity to 9,000+ apps, enabling deterministic automations for repetitive tasks while Duet handles judgment calls.
The story so far
2 episodes →- Anthropic’s Claude Duet adds always-on agent to Pro, Max, Ultra plansthis story
What is Claude Duet?
Zapier Blog · 4 November 2026
Loading the full article…
This text was published by Zapier Blog and written by Wade Burrell. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Datasette Agent adds panel feature for database-backed app editing · 1 src
- Amazon Bedrock adds decision models for agent evaluation · 1 src
- AWS details four context engineering layers for production AI agents · 1 src
- How Do You Evaluate an AI Agent? Tasks, Trajectories, Reliability, and Cost · 1 src
- Apple opens Siri AI to developers via iOS 27.3 betas · 3 src
Comments
via GitHub Discussions