Anthropic opens preview of Claude Duet research to US Pro and Max users – Unite.AI

Anthropic on October 21, 2026 opened a preview of Claude Duet Search, an app that the company says turns Claude into a proactive agent that highlights items that need attention, completes required tasks, continues to work when users walk away, and checks before taking action on their behalf. Access is through a waitlist, starting with Pro and Max users in the US.

Anthropic said it created Claude Duet for professionals and small business owners with more things to do than they can track. The app is available as a standalone app on iOS and Android and on claude.ai; during the search preview it is included in the Pro and Max plans and uses the same usage limits as the rest of Claude. Anthropic said it will email users on the waitlist as soon as their access is ready.

How Claude Duet works

Anthropic said it designed Claude Duet around Claude Opus 5.5 and built it to decide what the user needs, what they can wait for, and when to check first. Every morning the app sends a short brief on the most important topics on the user’s connected apps; during the day it keeps tabs on apps and messages only when something needs attention. Tasks range from scheduling meetings to creating dashboards with dozens of tools, creating financial or sales forecasts, prototyping, and building web apps.

Claude Duet transfers memory, skills, and connected apps from a user’s Claude account, and users submit it via typing or voice in an ongoing thread. The app runs on a virtual machine with its own browser, so the user’s laptop or phone doesn’t need to be turned on; Sends a notification when a task is completed or needs review. Anthropic said the app checks linked accounts every few hours because continuously checking would use up plan limits much faster; these background checks count towards shared use limits, and larger tasks like building a financial model or deck require multiple model shifts.

Approval gates, connectors and payments

Claude Duet reads, drafts and prepares the work himself, but before sending a message, making a payment, signing a document or deleting anything, he shows the user exactly what he is going to do and waits for approval. Some actions are prohibited even with approval, including trading securities and moving money between accounts, and users can log out of any app Claude Duet has access to.

The app can use the more than 6,000 connectors and plugins in the Claude Marketplace, with Anthropic’s listed uses ranging from helping you find and respond to missed Gmail and Slack messages, to checking on unpaid invoices in Intuit QuickBooks or Stripe with redacted reminders, to preparing and managing payroll in Gusto with schedules, bonuses and reimbursements filled out, and LinkedIn job ads posted after the user edits and approves them. Other uses listed include routing deals through Docusign for review and signature, drafting follow-up emails from HubSpot contacts and deal history, turning a product launch into a Canva social campaign, and setting up, designing, and managing a Shopify store, with a richer visual storefront editor on the way.

For business purchases, Claude Duet works with Link, Stripe’s digital wallet: the app confirms the item to the user, Link asks the user to approve the spend, and payment is made via a virtual card so the real card number isn’t shared. It can also use Shopify’s Shop Pay to make purchases at online stores, and Anthropic said that, with user permission, it will soon be able to log in to sites with logins saved in 1Password.

Safety design and testing

Anthropic said Claude Duet is based on its trusted agent framework, which also drives Claude Code and Claude in Chrome, and that credential management follows the design used for Claude Managed Agents, so the app never directly sees users’ passwords or access tokens. The company said the actions are approved by the automatic mode classifier, the same protection that protects claude.ai and Claude Code from taking actions unsolicited by users, and that the app’s defenses against hidden instructions and rapid injection come from its research into browser usage. Anthropic said Red Teams and an outside firm have been penetration testing Claude Duet and that the app is covered by its public bug bounty program on HackerOne.

Anthropic has released Claude Duet with a new dedicated classifier and a security system that checks every message, which the company says is designed to reject requests to share the strictest level of information, including credit card numbers and government identification. Other sensitive information, such as health or personal financial details, is transmitted only when the user has explicitly instructed Claude Duet to send it, Anthropic said. The company said it will share more details about the app’s security design in the coming days.

First testers and limitations stated

Anthropic said its employees use the app. Boris Cherny, who leads the Claude Code team, talks about his mental list of things to do with Claude Duet during walks; Chelsea Larsson, who leads the product model and behavior design team, gives him her to-do list before she leaves to be with her kids, and saves the results for an evening debrief; and Karan Ahuja, who leads the Claude Duet team, uses it to schedule user interviews, collaborate on prototypes, and ship code.

Anthropic also offered small business owners and entrepreneurs early access. Jess Provencher, a broker and owner of Pro Homes, said the app marked, without being asked, a booked property on her calendar for the wrong day. Amin Ebrahimi, owner of Shahs of Kabob in South Florida, which has about 110 employees, said he has used it for accounting, customer service and marketing, working on invoices on a store-by-store basis. Mya Dalton, a brand strategist, said Claude Duet prepared a brief for an upcoming call with a potential client and, seeing the call drop shortly before her flight, offered to move it.

Anthropic described the preview as an early version that can still misread a thread or miscalibrate what’s urgent, and said personal errands like booking tickets aren’t as strong because the app’s core functionality was built around work. Claude Duet can’t show users its own browser yet, and for now, it lives in its own app separate from the main Claude app; Anthropic said that what it learns from early users will determine how the two work together, and that it will continue to tune in as the app sends messages to users based on tester feedback.