Claude Duet
claude.com
October 21, 2026
3 min read
62/100
Summary
Claude Duet is Anthropic’s proactive AI agent that handles work through messages, operates on its own cloud computer, and contacts users when a task is complete or requires a decision. Powered by Opus 5.5, it is available as a research preview through a waitlist for Claude Pro and Max subscribers in the United States; Team and Enterprise plans are not included. Claude Duet is offered as a standalone iOS and Android app and is accessible on the web at claude.ai. The agent uses a customer’s Claude account, including its memory, skills, and connected apps, and retains conversation history in one thread. It can produce morning briefs, identify inbox and calendar items requiring attention, draft replies, schedule meetings, build dashboards, and return work such as documents, spreadsheets, and slide decks. It can also find unpaid invoices and draft reminders, prepare payroll for review, and ready contracts for sending. Claude Duet requires approval before sensitive actions such as sending messages or making payments. Approvals, logins, and safety checks run in the user’s cloud computer, and users can review and steer its activity. Its background checks count against a subscriber’s Claude usage limits, with larger tasks consuming more model turns and background work. Anthropic says the preview may misread threads, misjudge urgency, or miss items that arrive between check-ins.
Key Takeaways
- Claude Duet is a proactive Claude agent that performs tasks in a cloud computer and alerts users when work is finished or approval is needed.
- The research preview is on a waitlist for Claude Pro and Max subscribers in the United States and is not yet available for Team or Enterprise plans.
- Claude Duet can generate morning briefs, draft communications, schedule meetings, build dashboards, and help with business tasks including invoice reminders, payroll preparation, and contracts.
- Sensitive actions, including sending messages and making payments, require user approval, while Duet’s background activity counts toward the user’s Claude plan limits.
What the discussion said
Commenters spent less time on Claude Duet’s announcement than on whether persistent, delegated AI assistants have earned a place in ordinary life. Enthusiasts argue that pairing a frontier model with long-lived context and browser access changes the experience from isolated prompting into genuine delegation: agents can monitor inboxes, catch costly mistakes, manage household chores, navigate opaque websites, and coordinate complicated cross-language logistics. Several readers say the value compounds as users discover more routine work to hand off, especially when the agent can proactively surface issues rather than wait for a request. The skepticism is equally practical. Coding and other complex tasks reportedly remain brittle, web automation is slowed or defeated by blocked pages and CAPTCHAs, and a slow agent feels detached rather than helpful. Readers also question whether polished demos serve executives rather than typical households, and whether another proprietary assistant should be trusted with passwords, communications, and highly personal data. Duet’s clearest appeal is its claimed use of Opus 5.5: some see a stronger underlying model as decisive after disappointing results from rivals. Others think model lock-in, fragmented products, and awkward onboarding undermine that promise. The thread also reflects fatigue with AI dominating tech discussion, though many treat that attention as inevitable while the technology is rapidly reshaping products and work.
Where opinion split
The sharpest dispute is whether always-on agents are a breakthrough personal tool or an overbuilt, risky convenience for a narrow professional class. Supporters point to proactive monitoring and successful real-world delegation that recovers time and prevents expensive errors; skeptics say unreliable execution, limited website access, privacy exposure, and unclear everyday use cases make on-demand assistance safer and more sensible.
Coverage & discussions
Hacker News · claude.com
Community discussions
Community Sentiment
Positives
- Frontier-grade reasoning is seen as the differentiator that can turn a persistent agent from a flaky novelty into something users trust with consequential, multi-step work.
- Proactive monitoring of email, calendars, bills, and reservations has already caught buried messages and preventable charges, giving agents value before the user notices a problem.
- Long-running household agents can collapse meal planning, grocery ordering, chores, and calendar upkeep into a shared assistant that frees up family time.
- A persistent conversation and retained context reduce the constant re-explaining that makes ordinary chatbots feel like disposable tools rather than assistants.
Concerns
- Giving an autonomous model access to password managers, communications, and sensitive life logistics concentrates an alarming amount of leverage in a corporate service and its supply chain.
- Agent reliability still falls apart on coding and complex web tasks, while blocked pages, CAPTCHAs, and missing official sources can push the model into guessing.
- Slow execution undermines the point of delegation: even a smart agent feels disconnected when a task takes long enough that users cannot comfortably collaborate with it.
- A model-tied assistant risks trapping users in one vendor’s ecosystem precisely when the underlying models are becoming easier to swap elsewhere.
