Anthropic launches Claude Duet: a proactive agent on Opus 5.5 that messages you before you ask

• Who: Anthropic, launching Claude Duet, a standalone proactive-agent app.
• What: Claude as a proactive agent — morning briefs, monitoring of connected apps, and tasks from scheduling to web-app building — running on its own cloud virtual machine with a browser.
• When: announced Wednesday, October 21, 2026, as a research preview with waitlist access.
• Availability: iOS, Android, and claude.ai; initially US Pro and Max subscribers; included in existing plan limits — background checks count toward them.
• Scale: built around Claude Opus 5.5; 6,000+ connectors in the Claude Marketplace (Gmail, Slack, QuickBooks, Shopify); purchases via Stripe Link and Shopify Shop Pay.
• Guardrails: approval required before sending messages, making payments, signing documents, or deleting files; a dedicated classifier and safety system prompt check each message; Anthropic acknowledges the agent can misjudge urgency.

Anthropic on Wednesday opened a research preview of Claude Duet, a standalone app where Claude works as a proactive agent — one that comes to you with what needs your attention, keeps working after you step away, and asks before acting on your behalf. The launch, announced October 21, 2026, puts Anthropic directly into the proactive-agent race that Meta's Muse and OpenAI's Dots opened this year. (MacRumors; Reuters)

What Duet actually does

Each morning, Duet sends a short brief on the most important items across your connected apps, then checks in every few hours and messages you only when something needs a decision. Tasks range from scheduling meetings to building dashboards, creating financial or sales forecasts, and prototyping web apps. You talk to it in one ongoing thread, by typing or voice, and it decides what needs you, what can wait, and when to check first.

A virtual machine that works while you're away

Duet runs on its own virtual machine with its own browser, so it doesn't need your laptop or phone to stay on — it sends a notification when a task completes or needs review. It carries over the memory, skills, and connected apps from your Claude account, which means there is little to set up. Anthropic designed Duet around Claude Opus 5.5, its flagship model family, aiming it squarely at professionals and small-business owners with more to track than they can handle. (Digital Trends; Unite.AI)

Connected to 6,000+ apps

Duet plugs into more than 6,000 connectors in the Claude Marketplace, including Gmail, Slack, QuickBooks, and Shopify — and it can make purchases through Stripe's Link digital wallet and Shopify's Shop Pay. Anthropic says Duet does not see passwords or logins when given account access, and that a dedicated classifier and safety system prompt check each message to guard against sensitive information leaks. (Reuters)

The guardrails

Anthropic is explicit about what Duet won't do unsupervised: it asks for approval before sending a message, making a payment, signing a document, or deleting a file. The company also acknowledges the agent can misjudge urgency — and that it is less effective at personal tasks like booking tickets. Background checks count toward your existing usage limits; Anthropic says Duet checks connected apps every few hours rather than nonstop, because continuous monitoring would burn through plan limits fast. (MacRumors)

Why it matters

The launch completes the trio: Meta's Muse drew millions of downloads, OpenAI launched its always-on Dots agents last month, and now Anthropic's entry is built directly on top of its most capable model with the whole Claude account — memory, skills, apps — riding along. For now the preview is limited to Pro and Max subscribers in the US through a waitlist, included in existing plan limits. If proactive agents become the default interface, the race won't be won by the smartest model alone — it will be won by whichever agent you trust to act while you're not watching.