Anthropic Launches Claude Duet: The Proactive AI Agent That Reaches Out Before You Ask
Key takeaway: Anthropic has launched Claude Duet, a proactive AI agent that checks your connected apps, sends a morning brief, and keeps working in the cloud while you are away. It is in a waitlisted research preview for US Claude Pro and Max subscribers, runs on the Opus 5.5 model, and asks your approval before sending messages, paying, signing, or deleting anything.
Anthropic unveiled Claude Duet on October 21, 2026, a standalone AI agent app built for professionals and small-business owners who, in Anthropic’s words, have too much to track. Unlike a chatbot that waits for your next prompt, Duet is designed to reach out before you ask — surfacing work that needs attention and continuing tasks in the background.
The launch lands in the middle of a race to build agents that do work, not just answer questions. OpenAI launched its always-on “Dots” agents last month, and Meta’s Muse agent has drawn millions of users since its launch. Duet is Anthropic’s answer: a work-first agent aimed squarely at busy professionals rather than general consumers.
What Claude Duet actually does
Duet starts each day with a short morning brief drawn from your connected apps — email, calendar, and other work tools — flagging what needs your attention. It then checks those apps every few hours and sends an alert when something new comes up or a task is finished.
When you hand it work, Duet runs it on its own cloud computer, with its own virtual machine and browser, so tasks continue even when your phone is off and without needing access to your local machine. It can handle simple administrative jobs, like scheduling and drafting replies, and larger projects, like building financial forecasts, dashboards, slide decks, and web apps. It carries over memory, skills, and connected apps from your existing Claude account, so setup is minimal, and all interaction happens in one ongoing thread rather than a string of separate chats.
Duet connects to more than 6,000 connectors in the Claude Marketplace, including Gmail and Slack, according to Reuters. It can also make purchases through Stripe’s Link digital wallet and Shopify’s Shop Pay.
How Duet compares with Muse and Dots
| Agent | Maker | Core idea | Availability |
|---|---|---|---|
| Claude Duet | Anthropic | Proactive work agent for professionals and small businesses; morning brief plus background tasks | Waitlisted research preview; US Pro and Max subscribers |
| Muse | Meta | Personal agent for everyday life, including shopping | General launch; over 10 million US and Canada users in its first weeks |
| Dots | OpenAI | Always-on agents that pursue goals across work apps like Slack and Microsoft Teams | Launched last month |
The comparison matters because the three products are converging on the same idea from different directions: Meta from consumers, OpenAI from workplace apps, and Anthropic from professional knowledge work. Duet’s bet is that the morning brief — a single, prioritised view of the day — is the habit that makes an agent stick.
Why it matters
Proactive agents change the relationship between people and AI assistants. A chatbot only helps when you remember to ask. An agent that watches your inbox and calendar can catch the meeting clash, the unpaid invoice, or the reply you forgot — but only if it understands what actually matters to you. That judgement is the hard part, and Anthropic is unusually open about it: the company says Duet can misjudge urgency and is less effective at personal tasks such as booking tickets.
For readers, the practical angle is simple. If you already live in Claude and juggle email, meetings, and documents, Duet is worth joining the waitlist for — but treat the preview as a trial, not a finished product. Check its briefs against your own priorities for the first few weeks, and keep approval switched on for anything it wants to send, pay, or sign.
What is confirmed, what is still unknown, and what to watch
Known: Duet launched October 21, 2026, as a waitlisted research preview for US Claude Pro and Max subscribers, built around the Opus 5.5 model, with a morning brief, periodic app checks, a cloud virtual machine, and approval gates for sensitive actions.
Unknown: Anthropic has not said when the preview will expand beyond the US, when Team and Enterprise plans will be included, or what Duet will cost once the preview ends. It has also not published performance data on how accurately Duet judges urgency.
What to watch: Whether Anthropic widens access before the end of the year, how it handles the first public mistakes by a proactive agent, and whether rivals match the morning-brief format. The first of those will tell you whether Duet is a real product line or a limited experiment.
Analysis: the approval gate is the product
Analysis: Every proactive agent faces the same trust problem — the more it does on your behalf, the more damage a misunderstanding can cause. Anthropic’s answer is to keep a human in the loop for irreversible actions and to run Duet in its own cloud machine, where it never sees your passwords or logins. A dedicated classifier and safety prompt check each message for sensitive-information leaks. That design is reassuring, but it is also a limit: an agent that must ask before acting is only as fast as your replies. The agents that win this race will be the ones that earn the right to act without asking — and that trust will take far longer to build than the technology did.
Duet also fits a wider Anthropic pattern this month. The company recently donated its Agent Skills format to the Linux Foundation’s Agentic AI Foundation, a move that makes Claude’s skills portable across platforms. Read more in our coverage of Anthropic’s Agent Skills donation. And Duet arrives in the same week Anthropic joined OpenAI in pulling frontier training off the live internet — a sign both labs are treating their own most capable systems as things to contain. See our report on OpenAI’s move to offline frontier training.
If you find USABite useful in Google Search, you can add it as a preferred source in the Google app so our stories show up more often in your results.
Frequently asked questions
What is Claude Duet?
Claude Duet is Anthropic’s proactive AI agent app. Instead of waiting for a prompt, it monitors your connected apps, sends a morning brief of items needing attention, and continues delegated tasks in the cloud while you are away.
How is Claude Duet different from regular Claude?
Regular Claude answers when you ask. Duet reaches out first: it checks your apps every few hours, flags what matters, and works on tasks in the background in one ongoing thread, using your existing Claude account’s memory, skills, and connected apps.
Who can use Claude Duet, and how much does it cost?
Duet is in a waitlisted research preview for Claude Pro and Max subscribers in the United States, on iOS, Android, and the web. It uses your existing plan’s usage limits, so background checks count toward them, and Anthropic has not announced separate pricing for the preview.
What can Claude Duet actually do?
Duet can draft replies, schedule meetings, prepare research briefs, financial models, sales forecasts, slide decks, and dashboards, and build prototypes and web apps. It connects to more than 6,000 apps and services, including Gmail and Slack, and can make purchases through Stripe Link and Shopify Shop Pay.
Is Claude Duet safe, and what does it need approval for?
Duet asks your approval before sending a message, making a payment, signing a document, or deleting a file, and it never sees your passwords or logins. Anthropic still warns it can misjudge urgency and is weaker at personal tasks such as booking tickets, so early users should review its work closely.
Sources
Reuters: Anthropic unveils Claude Duet AI agent app
MacRumors: Anthropic Launches Claude Duet, a Proactive AI Agent
Pulse 2: Anthropic Launches Claude Duet as Proactive AI Agent for Work and Small Businesses
