Claude Duet Takes Its Place In The Race For Always-On AI

🔍 Read the full analysis: Claude Duet Takes Its Place In The Race For Always-On AI 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, adding a cloud-based AI agent to a growing field of assistants designed to work beyond a user’s active chat session. The separate app is available to US Pro and Max subscribers through iOS, Android and claude.ai, according to the company.

Anthropic describes Duet as an agent for professionals and small business owners. It can send a morning brief, accept tasks by text or voice in an ongoing thread, and work on a virtual computer with its own browser while the user’s devices are off. Company examples include scheduling meetings, preparing dashboards and sales forecasts, and creating small web apps. Duet is built on Claude Opus 5.5 and can use connected apps and services through Claude Marketplace integrations, Anthropic says.

The company says Duet can read, draft and prepare work on its own, but requires user approval before sending sensitive messages, making payments or deleting items. Anthropic says approval screens show the action, destination and content, including message recipients. Trading securities, moving money between accounts and accessing another person’s account are listed as prohibited actions, even with user approval.

Anthropic says the agent does not see users’ passwords, browser cookies or sign-in tokens, and that a separate classifier checks whether a proposed action matches the user’s intent. These are company descriptions of the system, not findings from an independent test. The launch material also identifies limits: Duet may misread conversations or urgency, cannot yet show users its browser, and checks connected accounts every few hours rather than continuously.

The Agent Race Moves to Workflow Control

Duet’s launch adds Anthropic to a competition over assistants that can act across services, rather than only answer prompts. The appeal is practical: an agent could monitor work, assemble information and prepare routine tasks without a person keeping a chat window open. For businesses, that could reduce manual coordination, but it also puts permissions, oversight and reliability at the center of deployment decisions.

The wider contest may depend on more than which company offers the most capable model. The source report points to a shift toward the “switchboard”: the interface that holds a user’s context, connects apps and routes tasks to models. It says SpaceXAI’s Grok Bot may route work to models from other providers, including Claude Opus 5.5, and cites xAI’s FAQ saying customers cannot restrict Grok Bot’s model choice. If that account is accurate, control over the user relationship and task routing could matter alongside the underlying model. The commercial implications remain uncertain, especially while Duet is a limited preview and product terms may change.

For users, the immediate stakes are less about replacing human judgment than deciding what information an agent can access and what actions it may take. A mistaken draft is different from a sent message or completed payment. Anthropic’s approval rules are intended to separate preparation from consequential action, but the company itself warns that frequent prompts can lead people to approve requests without careful review.

A Rapid Run of Agent Launches

The source report describes a series of launches over roughly two months. It says SpaceXAI opened a Grok Bot beta on August 11, Meta launched Muse on September 8, OpenAI introduced Dots at DevDay on September 29, and Google released Gemini Agent on October 8, citing TechTarget for the Google date. Anthropic’s October 21 launch followed those developments.

The products share a broad premise: an assistant can operate in a cloud environment, retain context and return with work completed or a request for attention. Their features, access rules and commercial terms are not necessarily equivalent. The source material provides limited comparative detail on Dots, Muse and Gemini Agent, so it does not support a detailed product ranking or a reliable price comparison.

Anthropic’s launch is a research preview, not evidence that the service is broadly available or that its stated performance has been established independently. Access is described as limited to US Pro and Max subscribers selected through a waitlist. The company says Duet carries over memory, skills and connected apps from a user’s Claude account and supports more than 6,000 connectors and plugins through Claude Marketplace.

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

— Anthropic, on approval prompts

Preview Limits and Safeguards Await Testing

There has been no independent test of Duet’s capabilities or safety measures in the source material. It is not clear how often the agent will complete tasks accurately in everyday use, how the approval process performs under pressure, or how the service will change as Anthropic expands access. The waitlist size, rollout pace and any broader availability date are also not given.

Anthropic says Duet checks connected accounts every few hours, which means it may miss events between checks. Background monitoring counts against the usage limits shared with other Claude activity, and larger jobs may require more model turns. The company also says that disconnecting an app stops new access, but information Duet has already learned remains until its memory is reset. The source material does not specify in detail how long different kinds of retained information persist or how users can inspect all of it.

It is also unclear how Duet’s functions and restrictions compare directly with competing agents. The source report describes Grok Bot’s model routing, but does not provide enough matched testing or product detail to establish which assistant is more capable, safer or less costly. Launch features and prices may change, and the material gives no common benchmark for comparing them.

Watch for Broader Access and User Controls

The next immediate development is the expansion of Anthropic’s waitlisted preview. The company has not provided a public date for general availability in the source material. Users and businesses considering access can first confirm which apps are connected, what the service can read, and which actions require approval.

For organizations, the source report recommends small, written pilots with read-mostly permissions and an exit plan. That approach would let teams test task accuracy and monitoring frequency before granting broader access. Individual users can start with low-stakes tasks, connect only a few apps and keep approvals enabled for actions that send, pay or delete. Those precautions do not establish that the system is safe for every use; they limit exposure while its performance is assessed.

Further information to watch for includes changes to the waitlist, detailed retention and memory controls, and evidence from real-world use. Until those details and independent evaluations are available, Duet’s announced functions and safety boundaries should be treated as Anthropic’s stated design, not verified outcomes.

Key Questions

What is Claude Duet?

Claude Duet is a separate Anthropic app that can run tasks on a cloud computer and send users briefings or messages. Anthropic says it is built on Claude Opus 5.5 and uses connected apps and services.

Who can use Claude Duet?

At launch, Anthropic described Duet as a waitlisted research preview for US subscribers to its Pro and Max plans. The company has not announced a general-release date in the source material.

Can Duet send messages or make payments without approval?

Anthropic says Duet asks for approval before sending sensitive messages, making payments or deleting items. The company also says trading securities and moving money between accounts are prohibited, even if a user approves.

Has Claude Duet’s safety been independently verified?

No independent test is included in the source material. The safeguards and capabilities described so far are Anthropic’s statements; real-world performance remains to be evaluated.

Source: ThorstenMeyerAI.com