Claude Duet wants a place in your workday. Anthropic opened a research preview on October 21, offering an AI agent that monitors work, prioritises tasks and handles business activities in the background. The pitch is familiar; the attempt to make it useful without turning it into an unsupervised liability is the more interesting part.
Access is limited to users in the US with Claude Pro or Max subscriptions, and entry comes through a waitlist. Pro costs $20 monthly or $200 annually, while Max starts at $100 monthly. Anthropic says it is letting people in from the waitlist as quickly as possible, which is the standard way of saying demand is being measured one queue at a time.
Claude Duet runs on Claude Opus 5.5, released in late September, and is available through claude.ai as well as standalone apps for iOS and Android. It carries over a user’s existing memory, skills and connected apps from their Claude account, so setup does not begin with the agent pretending it has never met the user.
From Morning Briefings to Business Tasks
The agent prepares a morning briefing from connected applications and sends notifications when it finds something requiring attention. It checks connected accounts every few hours, then sends updates when tasks finish or need review. Users can step away from their devices because Claude Duet keeps working inside a virtual machine with its own browser.
Its task list covers scheduling meetings, drafting emails, building dashboards, preparing financial and sales forecasts, developing prototypes and creating web applications. Claude Duet can connect to more than 6,000 tools through the Claude Marketplace, including Gmail, Slack, QuickBooks and HubSpot.
Anthropic says early testers have used the agent for accounting, customer support and marketing. The company described its target audience in plain terms: “We built Claude Duet for professionals and small business owners with more on their plates than they can track.”
Karan Ahuja led Duet’s development after starting at Anthropic in August. He led product management for Meta Superintelligence Labs until July and has been using Duet to build Duet, describing it as built for professionals and small business owners “juggling a lot at work and at home.” The product is aimed at people with overflowing task lists, which is most of the working population and every small-business owner with an inbox.
Automation With Approval Gates
Claude Duet prepares work on its own but still needs approval before sending messages, making payments, signing documents or deleting information. Securities trading and transfers between accounts remain prohibited even when a user gives approval.
The agent does not access users’ passwords or sign-in tokens. Its safeguards detect prompt injection and block requests to share highly sensitive information, while users can share health and personal financial information when they instruct Claude Duet to send it.
Mike Krieger, co-lead of Anthropic’s Labs team, said the agent “can have the discernment and judgment around when to be proactive, when to act on your behalf.” That judgment is the product’s central promise—and the part users will test every time a connected inbox, payment system or financial account enters the workflow.
Claude Duet also arrives as Anthropic competes for attention in a crowded AI assistant market. Meta has Muse, while OpenAI has Dots; Apptopia estimates that Muse had more than 14 million installations as of Wednesday.
Claude users who installed Muse were using Claude more than before, but for fewer minutes per day than they spent on Meta’s product. Anthropic’s company valuation could reach as much as $2 trillion in an initial public offering, placing this preview inside a much larger contest over users, work software and the companies that control it.
Claude Duet is still a research preview, and its waitlist limits access to US Pro and Max subscribers. But its design points beyond chat: an agent that watches connected work, prepares the next move and stops at the decisions that carry real consequences.
Based on
- Anthropic Opens Claude Duet AI Agent Preview for US Pro and Max Subscribers — analyticsindiamag.com
- Anthropic launches Claude Duet, its answer to Meta and OpenAI’s personal agents — africa.businessinsider.com
- Claude Duet app is Anthropic’s bet on agents | TechCrunch — techcrunch.com
- Anthropic opening research preview of ‘proactive agent’ called Claude Duet — cnbc.com
- Anthropic’s answer to Muse is here: Claude Duet, a mobile and desktop AI agent that takes actions proactively | VentureBeat — venturebeat.com
