Anthropic's Claude Duet keeps working while you're away

Anthropic just told you where chat is going. You stop asking. It starts reporting.

Anthropic has opened a waitlist for Claude Duet, a research preview of a proactive AI agent that sends you a morning brief, drafts next steps and keeps working on tasks while you are away. It runs on Opus 5.5 and is limited, for now, to Pro and Max subscribers in the US.

What Claude Duet does, and who gets it

According to Anthropic's product page, Duet watches what needs your attention, takes on tasks large and small, and comes back with finished work such as docs, sheets and decks. It can set up meetings and build dashboards. It uses your Claude memory, skills and connected apps in one message thread.

It works on its own computer in the cloud, using the apps you connect.

The page says Duet "works with more than 6,000 connectors and plugins."

The guardrail Anthropic leads with: Duet "asks you before sensitive operations like sending messages or paying." Logins and approvals run on that cloud computer, and you can see what it has done and steer it.

Availability is narrow:

• A waitlist, labeled research preview
• Pro and Max plans, US only
• iOS and Android apps, and claude.ai on the web
• "Team and Enterprise plans are not included yet"

No price beyond the plan and no launch date are listed.

The cost is your usage limit

Here's the thing. Duet shares your plan's usage limits with the rest of Claude, and its background checks count against them, per Anthropic. Bigger jobs, like a financial model or a deck, burn more turns. Engadget reports that Duet checks your connected accounts every few hours rather than constantly, for exactly that reason.

So an always-on assistant and your afternoon coding session now draw from the same pool.

Anthropic is also upfront about the weak spots. The page warns that "it may misread a thread or miscalibrate what's urgent," and that it can miss things that arrive between check-ins.

Late, and cheaper to get into

Duet is not first. Engadget frames it as Anthropic's answer to Meta's Muse and OpenAI's Dots, and reports that Dots is limited to OpenAI's Pro tier, which starts at $100 a month. Engadget also reports Anthropic says Pro subscribers, at $20 a month, will get Duet when it reaches general availability.

$20 versus $100.

That is the real competitive move. OpenAI, Meta and now Anthropic are building toward the same idea: an assistant that comes to you. Anthropic is betting on the entry price and the size of its connector library to make up for arriving later.

Skeptical? Fair. A morning brief is easy to demo and hard to trust. The test is whether it flags the email that matters and leaves the other forty alone, and a waitlist preview tells you nothing about that yet.

What to do and what to watch

If you are on Pro or Max in the US, join the waitlist, then connect the fewest apps that cover your real work. Every connection is something the agent reads, and every check-in spends limits you may want for other work.

If you run a team, note that Team and Enterprise are out for now. Your people may try it on personal plans first, with work accounts attached. Decide your policy before they do.

Anthropic says more detail on Duet's safety protections is coming in the days ahead, per Engadget. Read it before you hand over your inbox.

Then watch the general availability date and the price. That is when this stops being a preview.