The AI agent wars just got a third heavyweight. The Anthropic new AI agent, called Claude Duet, launched Wednesday as a research preview for paying U.S. subscribers — aimed squarely at Meta’s Muse and OpenAI’s dots, which beat it to market. Anthropic, which could file for its IPO as soon as Friday per Investopedia, is betting a proactive assistant that briefs you before you ask can win where chatbots stalled.
Anthropic New AI Agent: Meet Claude Duet
Claude Duet isn’t another chatbot window. Anthropic describes it as a proactive agent: each morning it prepares a short daily briefing that lists action items drawn from your connected apps, then keeps an eye on those accounts through the day. Find something that needs attention and Duet messages you about it — you can set how often it pings you and which topics trigger a message.
For now, access is tight. It’s a waitlisted research preview open only to U.S. subscribers on Claude’s Pro ($20 per month) and Max (starting at $100 per month) plans, with invitations going out in waves. Duet ships as standalone apps for iOS and Android, plus the web at claude.ai.
“It comes to you with what needs your attention, completes tasks you ask of it, keeps working even when you step away, and checks with you before it acts on your behalf,” Anthropic said.
What Duet Actually Does
Duet runs on Claude Opus 5.5 and operates on its own virtual machine with a browser — the key technical trick. Hand it a job by text or voice message, close your phone, and the work continues in the cloud.
A Briefing Before You Ask
The daily briefing is the centerpiece. Duet checks your connected accounts every few hours — nonstop checking would burn through plan limits too fast — and pulls together what matters from apps like Gmail and Slack. It can draw on more than 6,000 connectors and plugins in the Claude Marketplace, and take on anything from scheduling meetings to building dashboards, financial forecasts, and web apps.
Works While You’re Away
That background operation is what separates agents from chatbots: a research job or forecast model keeps cooking while you’re in meetings. Duet can even make purchases via Stripe’s Link or Shopify’s Shop Pay — though, crucially, it stops and asks first.
The Guardrails — and the Fine Print
Letting software read your inbox and spend money takes nerve, even with permission gates. Duet pauses for approval before sending messages, making payments, signing documents, or deleting files — and securities trading and moving money between accounts are prohibited outright, even with your approval.
On the safety side, the company says Duet includes a new classifier and system prompt designed to prevent leaks of sensitive information. It will never directly touch your passwords or sign-in tokens, and information related to health or personal finances gets shared only if you explicitly tell it to. More details on the safety protections are promised in the coming days.
Then the limits. Anthropic warns the preview may misread context or misjudge urgency, it’s weaker at personal errands like booking tickets, and you can’t yet see its browser. One more catch: background checks and big jobs count against your regular Claude limits, so heavy use burns your allowance faster than chat alone.
Muse, Dots, and the Agent Arms Race
Anthropic is late to a party that started without it. Meta rolled out its AI agent Muse last month to a warm reception, and OpenAI’s dots followed weeks later, according to Investopedia. All three are chasing the same idea: agents that act before being asked.
The dividing lines are ecosystem and price. Duet ties into your Claude account’s memory and thousands of connectors but is a U.S.-only preview for Pro and Max subscribers. OpenAI’s dots is limited to OpenAI Pro subscribers (plans start at $100/month), while Meta’s Muse plays the consumer angle inside its massive app family.
This launch is a product story, not a policy one: Anthropic’s lobbying on AI safety legislation is a separate fight. Duet is about the commercial race, and the timing — days before a possible IPO filing — is no coincidence. Anthropic clearly doesn’t intend to watch the agent wars from the sidelines.
Sources and Resources
- Investopedia — 5 Things to Know Before the Stock Market Opens on Thursday (Oct. 22, 2026)
- Dataconomy — Anthropic Unveils Claude Duet Personal AI Agent in Research Preview
- World Reach Daily — more technology news and Anthropic new AI agent coverage from our newsroom.
Related coverage from World Reach Daily: more technology news and Anthropic new AI agent coverage, Anthropic new AI agent availability details, how the Anthropic new AI agent compares to Meta’s Muse and OpenAI’s Dots, and what the Anthropic new AI agent launch means for the agent wars.
What is Claude Duet, exactly?
It’s Anthropic’s new proactive AI agent. Instead of waiting for you to open a chat, it sends a morning briefing of action items from your connected apps, monitors them during the day, and keeps working on tasks in the cloud even when your devices are off. It runs on Claude Opus 5.5.
How do I get access to Claude Duet?
Right now, you need to be a U.S.-based Claude Pro or Max subscriber, then join the waitlist — invitations are going out in waves. Team, Enterprise, and free accounts aren’t eligible for the preview, and it’s not available outside the U.S. yet.
Does Duet cost extra on top of my Claude subscription?
No separate price was announced. It’s included with eligible Pro and Max plans, but it shares your existing usage limits — background checks and big tasks count against the same allowance as your regular Claude chats.
Can Duet send emails or spend my money by itself?
No. It’s built to ask for your approval before sending messages, making payments, signing documents, or deleting files. Securities trading and moving money between accounts are prohibited entirely, even with your approval.
How is Duet different from Meta’s Muse or OpenAI’s Dots?
All three are proactive agents that work unprompted. The differences are ecosystem and access: Duet plugs into your Claude account’s memory and 6,000+ connectors but is a U.S.-only research preview for Pro/Max users; Dots is limited to OpenAI’s pricier Pro tier; Muse rides inside Meta’s consumer apps.
Conclusion
Claude Duet is Anthropic’s answer to the obvious question: what comes after the chatbot? A limited, U.S.-only research preview won’t decide the agent race by itself, but the direction is clear — morning briefings, background work, approval gates, and a fight with Meta and OpenAI over whose assistant earns your trust. With a possible IPO days away, Anthropic just made sure it has a flagship in that fight.