Anthropic Unveils Claude Duet AI Agent to Compete With OpenAI Dots and Meta Muse

Anthropic Unveils Claude Duet AI Agent to Compete With OpenAI Dots and Meta Muse

Anthropic has introduced Claude Duet, a proactive AI agent designed to manage tasks, monitor connected applications and assist professionals and small business owners with ongoing work. The new service runs on Claude Opus 5.5 and is launching as a research preview through a waitlist for U.S. subscribers on Claude Pro and Max plans. It is accessible through standalone iOS and Android apps and Claude on the web.

Claude Duet is designed to work beyond individual chatbot conversations, allowing users to delegate assignments that continue running without constant supervision. It can organize schedules, prepare research reports, develop financial forecasts, build dashboards and create web applications. The agent determines which connected tools are needed for each assignment and can deliver completed work as editable files.

One of its central features is proactive task management. Each morning, Duet reviews connected email and calendar accounts to identify priorities and prepare a brief covering important developments. It can also draft responses, track unfinished conversations and notify users when an assignment requires their attention.

Throughout the day, the agent periodically checks connected services and provides updates on ongoing work. Rather than sending every development into a single conversation, Claude Duet organizes assignments in a dedicated Tasks tab, where users can review progress and respond to approval requests. A separate Make tab stores completed materials, including presentations, financial models and research documents.

The service draws on existing Claude account information, including saved memories, skills and application connections. Users can communicate with Duet through typed or recorded voice messages, while separate conversation channels allow different projects to share the same underlying memory.

Claude Duet also operates through its own cloud-based computer and browser, enabling it to continue assignments after a user closes the app. For websites without supported integrations, the agent can navigate pages directly rather than relying exclusively on application connectors.

Anthropic supports connections with services including Gmail, Google Calendar, Google Drive, Slack, QuickBooks, HubSpot, DocuSign and Calendly. The agent can also use plugins tailored to particular workflows, including a small business plugin, and recommend additional plugins based on a user's assignments.

Payment-related capabilities are available through integrations with Stripe, Link and Shopify's Shop Pay. For example, Duet can identify overdue invoices in Stripe and prepare reminder emails for review. It can also complete approved purchases through Link or Shop Pay, with each transaction using a payment card restricted to that specific purchase.

Because the agent can access business information and perform actions across multiple services, Anthropic has incorporated approval requirements for consequential activities. Duet is designed to request permission before sending communications, authorizing payments, signing documents or deleting information. If approval is delayed, the agent can continue other portions of an assignment and return to the pending decision later.

Anthropic has also introduced additional safeguards intended to prevent the disclosure of highly sensitive information, including credit card details. The company acknowledges that content retrieved from websites and emails could contain malicious instructions designed to manipulate the agent, a security risk known as prompt injection.

Some activities remain prohibited even when users provide permission. Claude Duet cannot execute securities trades, transfer funds between accounts or access another person's account. It also cannot place telephone calls or send text messages to other people, although it can send emails after receiving approval.

The research preview includes several operational limitations. Duet checks connected accounts periodically rather than continuously, meaning urgent developments may not be detected immediately. Users also cannot observe or manually control its cloud browser, and the agent lacks access to some local files and Claude projects.

Anthropic cautions that Duet can misunderstand conversations or incorrectly assess which tasks deserve priority. Some integrations, including Slack, have restricted functionality, while Microsoft 365 support is not yet available. The company also acknowledges that the agent is less capable at certain personal errands, such as booking tickets.

Access is initially restricted to U.S. users with Claude Pro or Max subscriptions. Anthropic is distributing invitations gradually through a waitlist, with no guaranteed admission date. Team and Enterprise customers are excluded from the initial preview, although the company plans to extend availability to those subscriptions later.

Claude Duet carries no additional charge during the preview, but its activity counts against existing Claude subscription limits. Background operations, including morning briefings and periodic account checks, consume usage allowances alongside ordinary conversations. Users admitted to the preview receive a one-time reset of their session and weekly limits.

The release places Anthropic in more direct competition with Meta's Muse and OpenAI's Dots, which also focus on proactive assistance and completing work across connected services. Claude Duet's approach combines existing Claude account context with independent task execution, ongoing monitoring and user approval controls.

For now, Anthropic is positioning Duet as an early-stage product rather than a fully developed autonomous assistant. Its broader availability and capabilities remain subject to changes during the research preview, as the company evaluates how reliably the agent can handle work involving multiple applications, sensitive information and decisions requiring human oversight.

See more here:

This analysis is based on reporting from Claude & The Verge.

Image courtesy of Anthropic.

This article was generated with AI assistance and reviewed for accuracy and quality.

Updated Oct 21, 2026

About this article: This article was generated with AI assistance and reviewed by our editorial team to ensure it follows our editorial standards for accuracy and independence. We maintain strict fact-checking protocols and cite all sources.

Word count: 856Reading time: 0 minutes

AI News Daily

Breaking Intelligence • Since 2023

Join hundreds of thousands of AI professionals who start their day with our curated newsletter. Get breaking news, expert analysis, and exclusive insights.

Stay Ahead of AI

Get the latest AI breakthroughs, tools, and insights delivered to your inbox every week.

✓ Free forever✓ Unsubscribe anytime✓ No spam guarantee

Go Premium

Unlock unlimited AI tools and an ad-free reading experience designed for AI professionals.

• Ad-free experience• Premium AI tools
Start Free Trial

14-day free trial • Cancel anytime
Plus $9/mo • Pro $90/yr (2 months free)

Follow Our Community

ChatAI

Breaking Intelligence

Your daily briefing on what matters in AI. Trusted by developers, researchers, executives, and AI enthusiasts worldwide.

© 2026 ChatAI. All rights reserved.