What is Claude Duet? Anthropic's proactive AI agent, explained

Most AI assistants wait for you to ask. Duet comes to you with what needs your attention, then helps you deal with it.

• A morning brief: a short summary of the important items across your connected apps, like email and calendar.
• Check-ins through the day: it looks at your connected accounts every few hours and messages you only when something needs you. You choose what you want to hear about and how often.
• Tasks you hand it: from scheduling meetings to building a dashboard, a forecast or a simple web app.
• Work while you're away: it runs on its own virtual machine with its own browser, so your laptop or phone doesn't need to be on.

It carries over your memory, skills and connected apps from your Claude account, and can use the connectors and plugins in the Claude Marketplace, such as Gmail, Slack, QuickBooks, HubSpot, DocuSign, Canva and Shopify.

This is the part that matters most for any team. Anthropic built Duet so you stay the one who decides.

• Before it sends a message, makes a payment, signs a document or deletes anything, it shows you exactly what it's about to do and waits for your approval.
• Some actions are off-limits even with approval, such as trading securities or moving money between accounts.
• It never directly sees your passwords or sign-in tokens.
• You start every connection yourself and can disconnect any app at any time.
• Highly sensitive details like credit card and government ID numbers are designed to be rejected, and health or personal finance information is shared only when you explicitly ask.

An agent is only as good as the data it can see. If your inbox, calendar and files are messy or out of date, Duet's brief will be too. A little tidying first pays off quickly.

• Availability: a research preview by waitlist, starting with US Claude Pro and Max subscribers. People are being let in gradually.
• Where: a standalone app for iPhone and Android, and on claude.ai.
• Cost: no extra charge. It's included in Pro and Max and shares the same usage limits as the rest of Claude.
• Watch your limits: the morning brief, background check-ins and big tasks all count, so you may reach your limits sooner than with chat alone.
• Not yet: Team and Enterprise plans aren't included.

Anthropic is upfront that this is early. It can misread a thread or misjudge what's urgent, personal errands like booking tickets are less developed, and you can't watch its browser yet. Because it checks every few hours, something urgent can also arrive between checks. Early reviews are mixed: people like the design and the morning brief, and some find the follow-through slow.

1. 1Start with your own accountsTry it on your own email and calendar for two weeks before connecting anything shared. Notice what it flags well and what it misses.
2. 2Give it one recurring jobPick something repetitive with low stakes: a daily brief, drafting replies to routine questions, or preparing for tomorrow's meetings.
3. 3Keep approvals on for anything outwardRead every message before it goes out. Trust is earned one week at a time.
4. 4Check your AI policy firstBecause Team and Enterprise plans aren't supported yet, staff would be using personal Pro or Max accounts. Decide what organization data, if any, may be connected that way.
5. 5Keep sensitive conversations outDon't connect inboxes that hold pastoral care, counseling, HR, donor financial or health information until you have clear rules for it.