San Francisco: AI firm Anthropic has unveiled Claude Duet, a new proactive agent that does not wait for a prompt. Instead of a chatbot that sits idle, Duet flags things that need attention and keeps working in the background after the user steps away from their phone or laptop.
Duet runs on the Claude Opus 5.5 model and operates in its own virtual environment, complete with a browser. It sends a short morning summary of key items across connected apps, checks those accounts every few hours and alerts the user when something needs attention.
It can schedule meetings, draft emails, prepare sales forecasts, build dashboards and put together simple web apps. Anthropic says it connects to more than 6,000 connectors and plugins, including Gmail, Slack, QuickBooks, Stripe, HubSpot and Shopify. It can also prepare payroll in Gusto and draft job postings for LinkedIn.
Before sensitive actions, however, it asks permission. It shows the user what it plans to do and waits before sending messages, making payments, signing documents or deleting files. Trading securities and moving money between accounts are blocked outright. It also refuses requests to share credit card numbers or government IDs.
For now the feature is available to Claude Pro ($20 a month) and Max ($100) subscribers in the US through a waitlist, on iOS, Android and claude.ai.
The agent market already has OpenAI's Dots (September), Google's Gemini Agent (October 8) and Meta's Muse. Forrester analyst William McKeon-White said there is nothing hugely unique about Duet except the restraint with which it is being introduced. GAI Insights chief Paul Baier said enterprises are not yet ready for such agents.