Los AngelesNew YorkChicagoHoustonPhoenixPhiladelphiaSan FranciscoSeattleAtlantaBoston
Tech · Elon Musk

6,000: The Connector Count Behind Anthropic's New Business Agent

Anthropic launches Claude Duet, a specialized agent for professionals that distinguishes itself from personal assistants by focusing on background automation and enterprise integrations.

By Marcus Johnson ·

The artificial intelligence landscape is currently saturated with personal and productivity-focused assistants, with new entrants appearing weekly. Amidst this crowded field, Anthropic has introduced Claude Duet, a strategic move designed to capture user attention by competing directly with established players like Meta Muse, OpenAI Dots, Hark, Instinct, Grok Bot, and Fo. The core tension in this release lies in the company’s deliberate positioning of the new tool not as a general-purpose personal companion, but as a specialized instrument for professional workflows.

The scale of Anthropic’s integration strategy is evident in the numbers. Claude Duet is built upon the Claude Opus 5.5 model and connects to more than 6,000 connectors and plug-ins available in the Claude Marketplace. This extensive network allows the agent to interact with a wide array of professional tools, a significant step beyond the capabilities of many earlier iterations of AI assistants that operated in more isolated environments.

Anthropic has made a clear distinction between Claude Duet and personal agents such as Instinct. The company notes that Duet is not optimized for handling everyday errands, such as booking tickets, nor does it provide a live view of its browser activity. Instead, the agent is designed to cater to professionals and small businesses by automating and managing complex tasks. It can construct daily briefings, generate dashboards, and suggest actions based on user data, all while operating in the background.

The operational mechanics of Duet reflect a shift in the broader market toward agents that work on behalf of users rather than simply conversing with them. Much like other agents in the space, Duet executes tasks within its own virtual machine. This allows users to continue with their own work while the agent processes information in the background. It will only return for user approval when necessary, ensuring that human oversight is maintained for critical steps.

This launch coincides with a wider industry pivot toward business-focused applications. While Meta’s Muse remains primarily a personal AI assistant, the company has also released a version tailored for small businesses with app connections. Similarly, OpenAI’s Dots leans into business use cases through its integration with professional tools. Anthropic appears to be aligning its strategy with this trend, emphasizing utility in professional settings over casual, personal interaction.

To maintain trust, Anthropic has implemented specific protocols for sensitive operations. While Duet can make online purchases on behalf of users via Stripe’s Link digital wallet or Shopify’s Shop Pay, these actions require explicit user approval. The company has clarified that the agent will not trade securities or move money between accounts, setting clear boundaries on its financial authority. Additionally, a new classifier has been introduced to scan every message, preventing the leakage of sensitive information. By default, the agent is programmed to decline requests involving credit card numbers or government identification.

Currently, Claude Duet is available as an application on iOS and Android devices. It is also accessible via the web through Claude.ai, specifically for Pro and Max customers. The service is currently in research preview, meaning users must join a waitlist to gain access. This staged rollout suggests Anthropic is monitoring performance and user feedback before a broader general availability launch.

Source: TechCrunch

More from Elon Musk