Claude Duet Enters The Always-On AI Race With The Switchboard In Play

🔍 Read the full analysis: Claude Duet Enters The Always-On AI Race With The Switchboard In Play on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

Anthropic launched Claude Duet on October 21 as a waitlisted research preview for U.S. Pro and Max subscribers, adding a cloud-based agent to a growing market for AI assistants that can keep working between conversations. The separate app can monitor connected services, prepare tasks and send updates, while Anthropic says actions such as sending messages or making payments require user approval.

Duet is available through a separate experience on iOS, Android and claude.ai. Anthropic says it runs on a virtual computer with its own browser, allowing it to continue working while a user’s personal devices are offline. People can assign tasks by text or voice in an ongoing thread. The company’s examples include scheduling meetings and preparing dashboards, sales forecasts or small web applications.

Anthropic says Duet can draw on memory, skills and connected apps associated with a user’s Claude account, including more than 6,000 connectors and plugins in the Claude Marketplace. Its examples include checking invoices in QuickBooks or Stripe, payroll in Gusto, agreements in DocuSign and storefront tasks in Shopify. The system is designed to provide a morning brief and contact users during the day when it says an item needs attention.

In a safety post dated October 23, Anthropic said Duet presents potentially sensitive actions for review, including the proposed action, destination and content, and waits for approval before proceeding. The company says trading securities, moving money between accounts and accessing another person’s account are blocked, even with approval. For purchases, Anthropic says Duet uses Stripe Link with a one-time card limited to a single merchant and amount.

Control of the Always-On Assistant

Duet’s launch adds another major technology company to a competition over AI agents that act across services, rather than only answering questions in a chat. If these systems become routine, users and businesses will need to weigh their convenience against the permissions granted, the reliability of their decisions and the time required to review proposed actions.

The source material also points to a second competition: who controls the agent’s interface, connected accounts and model routing. It reports that Grok Bot may route tasks to models from other companies, including Claude Opus 5.5, while xAI’s security FAQ says users cannot restrict that model choice through a team allowlist. That raises a practical question for customers: whether they choose a particular model directly or rely on an assistant platform to select one. The longer-term commercial effects are not yet established.

For organizations, the launch is a reason to test carefully rather than grant broad access at once. The source recommends small pilots, limited permissions and an exit plan. For individual users, starting with low-risk tasks and retaining approval for sending, paying and deleting would limit the consequences of an error.

How Duet Fits the Launch Wave

Duet follows a sequence of launches described in the source article: SpaceXAI opened a Grok Bot beta on August 11, Meta launched Muse on September 8, OpenAI announced Dots at DevDay on September 29, and Google released Gemini Agent on October 8, according to TechTarget. Anthropic’s October 21 announcement came after those developments.

The products are presented as variations on a shared idea: an assistant can work in a cloud environment, use context and connected tools, then report back or ask the user for a decision. The source characterizes Duet as built on Claude Opus 5.5, but its availability is limited to U.S. Pro and Max subscribers selected through a waitlist. Details such as launch prices for competing products can change; the source identifies its comparison figures as launch figures rather than current pricing.

Anthropic’s October 23 safety post gives more detail about the product’s controls than the initial launch description. It says a separate check, based on the auto-mode classifier used in Claude Code, evaluates whether a proposed action matches the user’s intent. These are company descriptions of the system, not results from independent testing.

“Approval fatigue is a serious issue, because too many requests teach people to tap through them.”

— Anthropic, in its October 23 safety post

Testing and Access Still Limited

Duet is a research preview, and the source says no independent test of its safeguards or performance is available yet. Anthropic’s statements about approvals, prohibited actions, credentials and data handling should therefore be understood as the company’s descriptions, not independently verified findings.

Anthropic says the early version may misunderstand message threads or misjudge urgency. The source also reports that Duet is less capable with personal errands such as booking tickets, cannot yet show users its browser, and checks connected accounts only every few hours. That polling schedule could mean it misses changes between checks. Background activity counts toward the same usage limits as other Claude work, while large jobs may require additional model turns.

It is also unclear when the waitlist will open more broadly, whether Duet will expand beyond U.S. Pro and Max subscribers, or how its capabilities and limits may change. Anthropic says disconnecting an app stops future access, but information Duet has already learned remains until the user resets its memory. The source does not provide independent evidence about how well that memory behavior or the approval checks work in practice.

Waitlist, Pilots and Safety Checks

The next immediate step is for waitlisted U.S. Pro and Max subscribers to gain access as Anthropic expands the research preview. The source gives no date for wider availability or for a move from preview to general release. Further product updates, if announced, will clarify whether the current limits on account checks, browser visibility and personal errands change.

Companies considering Duet can begin with a narrow, written pilot that uses only the app connections needed for a defined task, records what the agent is allowed to do and sets a process for stopping the trial. Individual users can test low-stakes work first, connect only a few services and keep approvals switched on for actions that send, pay or delete.

Users and organizations should also watch for evidence beyond launch statements: how often the agent makes mistakes, how reliably approval screens show the intended recipient and content, and what happens when connected accounts are removed or memory is reset. Until those points are tested, Duet’s real-world reliability and the reach of its safeguards remain open questions.

Key Questions

What is Claude Duet?

Claude Duet is a cloud-based AI agent from Anthropic, offered as a separate app experience on iOS, Android and claude.ai. Anthropic says it can work across connected apps, prepare tasks and contact users with briefings or requests for input.

Who can use Claude Duet now?

The source describes Duet as a waitlisted research preview for U.S. subscribers to Claude Pro and Max. Anthropic has not provided a broader release date in the supplied material.

Can Duet send messages or make payments without approval?

Anthropic says Duet must wait for approval before sending potentially sensitive messages, paying or deleting items. The company says trading securities and moving money between accounts are blocked, even if the user approves.

What are Duet’s known limitations?

Anthropic says the preview may misread a thread or misjudge urgency. The source also lists checks every few hours, limited support for personal errands, no browser view yet and shared Claude usage limits as current constraints.

Has Duet’s safety been independently verified?

Not according to the source material. The safeguards and data-handling details cited here come from Anthropic, and no independent test is reported.

Source: ThorstenMeyerAI.com