🔍 Read the full analysis: The Always-On AI Race Gains A Contender In Claude Duet on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Anthropic launched Claude Duet as a waitlisted research preview on October 21, 2026, for U.S. Pro and Max subscribers. The separate app can work in the cloud across connected services, but Anthropic says it requires user approval for sensitive actions and blocks some financial activity; independent testing has not been reported in the source material.
Anthropic launched Claude Duet on October 21 as a waitlisted research preview for U.S. Pro and Max subscribers, adding its entry to a growing race to build AI assistants that keep working on a cloud computer between user prompts. Anthropic says Duet can monitor connected apps, prepare work and ask for approval before certain actions, but its capabilities and safeguards have not been independently tested in the source material. Read more about Claude Duet working offline.
Duet is a separate app available on iOS, Android and claude.ai. Anthropic says it runs tasks on a virtual machine with its own browser, allowing it to continue working when a user’s phone or computer is off. Users can give it instructions by text or voice in an ongoing thread. The company describes features including a morning brief, reminders when something needs attention, and work such as scheduling meetings or preparing dashboards, sales forecasts and small web apps. More details are available in this report on Claude Duet.
The agent can use the memory, skills and connected apps associated with a Claude account. Anthropic says it has access to more than 6,000 connectors and plugins in the Claude Marketplace, with examples including QuickBooks, Stripe, Gusto, DocuSign and Shopify. For purchases, the company says Duet uses Stripe’s Link wallet and a one-time card limited to a specific merchant and amount.
Anthropic’s October 23 safety post says Duet must show the proposed action, destination and content—and wait for approval—before it sends a potentially sensitive message, pays or deletes something. Some actions remain blocked even with approval, including trading securities and moving money between accounts. The company also says users cannot set payments, sending money, deleting an order or creating an account to run without asking each time.
Infographic 01 · The permission ladder
What Claude Duet does alone, what it asks first, what it never does.
Anthropic’s three tiers for an agent that keeps working while you are away. Health or financial details go out only with your explicit consent.
Does it alone
no approval
- Reads connected apps: mail, calendar, Slack, invoices
- Sends a morning brief, then flags what needs you
- Drafts replies, reminders, decks, dashboards
- Prepares payroll or contracts for your review
- Checks accounts every few hours, not nonstop
Asks you first
approval card shows action, destination, content, recipients
- Sending a potentially sensitive message
- Paying (one-time card via Link, limited to merchant and amount)
- Signing a document
- Deleting anything
- Never on “don’t ask me again”: paying or sending money, deleting an order, creating an account
Never, even if you approve
blocked
- Trading securities
- Moving money between accounts
- Accessing someone else’s account
- Sharing sign-in credentials, full card or bank numbers, government ID numbers
- Seeing your passwords, cookies or sign-in tokens (kept off the model’s machines)
“Approval fatigue is a serious issue, because too many requests teach people to tap through them.”Anthropic, Claude Duet safety post
These are Anthropic’s stated rules for the research preview. No independent test has been published yet.
Sources: Anthropic, Introducing Claude Duet (21 Oct 2026); Anthropic, Duet safety post (23 Oct 2026); Claude Duet FAQ.
Powered by Thorsten Meyer AI
A New Bid for Persistent AI Work
Duet extends the assistant model from answering a prompt to handling work across connected services over time. If it performs reliably, that could change how professionals manage routine monitoring, follow-ups and preparation. It also puts more weight on questions of access: an agent that reads business systems and prepares actions can save attention, but mistakes or poorly judged permissions could have consequences beyond an incorrect chat response.
The launch also adds another competitor to an increasingly crowded category. The source account describes Grok Bot, Meta Muse, OpenAI Dots and Google Gemini Agent as other efforts to make assistants more persistent. Their availability, capabilities and terms differ, and the launch dates and feature descriptions in the source are not independently verified here. The broad competitive issue is whether users will choose a model, an interface that coordinates models, or the services and permissions that make an agent useful.
For companies, Duet’s preview status argues for cautious evaluation rather than broad deployment. A small pilot can reveal whether the agent’s monitoring and task preparation are useful while limiting exposure to connected data. For individuals, starting with low-stakes work and a small number of connected apps can help keep errors and access risks contained.
Products Worth Considering
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
As an affiliate, we earn on qualifying purchases.
The Agent Race Before Duet
Anthropic’s announcement follows a series of launches described in the source material: 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. Anthropic announced Duet on October 21. These dates frame the product as part of a rapidly developing market, not an established category with settled performance standards.
The products share a premise: an assistant can work in a cloud environment, use context from connected services and return with a result or request for input. Their feature sets and access rules are not necessarily equivalent. Anthropic says Duet inherits account memory and connections, checks accounts every few hours rather than continuously, and uses the same usage limits as the rest of Claude. The company says those periodic checks can miss developments between them and that larger tasks may require more model turns.
One wider issue raised by the source is who controls model selection when an agent routes work to other providers. It reports that xAI has said Grok Bot may route tasks to external models, including Claude Opus 5.5, and that customers cannot choose its model through a customer-facing picker. That detail is attributed to the source’s account of company statements; it does not establish how every agent in the market will handle routing.
“for professionals and small business owners with more on their plates than they can track.”
— Anthropic, describing the intended audience for Claude Duet
Preview Limits and Open Questions
There is no independent test in the supplied reporting to confirm Duet’s performance, safety checks or ability to complete the tasks Anthropic describes. The product is a waitlisted research preview, and the source does not provide a general release date, a full account of regional availability beyond the stated U.S. subscriber eligibility, or separate pricing.
Anthropic says the early version may misread message threads or misjudge urgency. It also says personal errands such as booking tickets are weaker, Duet cannot yet show users its browser, and account checks happen every few hours. The source does not quantify how often the agent misses updates, how its approval prompts perform in practice, or how much usage typical monitoring and larger tasks consume.
Data retention and access also warrant attention. Anthropic says Duet does not see passwords, browser cookies or sign-in tokens, and that disconnecting an app stops new access. However, information the agent has already learned remains until its memory is reset, according to the company. The source does not describe independent audits or provide further technical detail on that retention process.
What Users and Companies Can Test
The immediate next step is access through Anthropic’s waitlist for eligible U.S. Pro and Max subscribers. The source gives no timeline for a wider release. As users try the preview, the key questions will be whether Duet’s briefs and background work are accurate enough to trust, whether its checks surface relevant changes in time, and whether approval prompts remain useful rather than becoming routine clicks.
Companies considering a pilot can limit the initial test to a small group, a defined set of low-risk tasks and read-mostly permissions. They can document what data is connected, which actions require approval, how to stop access and how to reset agent memory. Users can take a similar approach by connecting only a few apps and keeping approvals enabled for anything that sends, pays or deletes.
Anthropic’s statements set out the controls it says are in place; real-world performance remains to be established. Until independent evaluation and broader rollout details emerge, the preview should be treated as a limited trial, not proof that an always-on agent can safely manage consequential work without close supervision.
Key Questions
What is Claude Duet?
Claude Duet is Anthropic’s cloud-based AI agent, offered as a separate app on iOS, Android and claude.ai. Anthropic says it can work across connected apps, prepare tasks and send users a morning brief or a message when attention is needed.
Who can access Claude Duet?
Anthropic launched Duet as a waitlisted research preview for U.S. Pro and Max subscribers on October 21, 2026. The source material does not give a date for wider availability.
Can Duet send messages or spend money without approval?
Anthropic says Duet must ask before sending potentially sensitive messages, making payments or deleting items. It says some actions, including securities trading and moving money between accounts, are blocked even if a user approves.
Has Claude Duet been independently tested?
No independent test is cited in the source material. The available capability and safety descriptions come from Anthropic, and the product remains a research preview.
What should users do before connecting accounts?
For an initial trial, users can choose low-stakes tasks, connect few apps and keep approval prompts on for actions that send, pay or delete. Anthropic says disconnecting an app stops new access, while information already learned remains until memory is reset.
Source: ThorstenMeyerAI.com
