Anthropic Launches Claude Duet as Proactive Agent for Professionals and Small Businesses

Anthropic has opened a research preview of Claude Duet, a standalone AI agent designed to work proactively for professionals and small-business owners. The app joins a crowded market that includes Meta Muse and OpenAI Dots, both of which are pursuing assistants that handle work beyond a traditional chat window.

Duet is available through a waitlist for eligible Claude Pro and Max subscribers in the United States, according to TechCrunch and Engadget. Anthropic’s launch positions Claude as an assistant that initiates useful work, rather than simply responding when a user submits a prompt.

Claude Duet delivers a daily briefing, identifies tasks that need attention, and continues working while the user is away. Its proposed workload includes scheduling meetings, preparing financial forecasts, building web apps, and handling other professional tasks, according to Reuters coverage carried by The Star and Central Oregon Daily.

The agent is designed to communicate through a conversational interface while connecting to cloud services and work tools. Anthropic says Duet checks with users before taking actions on their behalf, creating an approval step for consequential activity such as sending messages or making changes. That design gives the product a clear operating model: monitor, prepare, ask, then act.

Duet’s audience separates it from several consumer-focused personal assistants. Anthropic is pitching the product toward people who manage calendars, projects, finances, client work, and small operations, making reliability and task continuity central to its appeal.

The product’s positioning also reflects Anthropic’s broader identity as a provider of work-oriented AI. Every described capability ties the agent to an outcome, from a morning briefing to a working forecast or web application. Every, an early reviewer, characterized Duet as closer to a professional tool than a novelty, while also pointing to unfinished infrastructure around the experience.

Claude Duet arrives as Meta and OpenAI push personal agents that can maintain context, take initiative, and complete multi-step tasks. Meta has said Muse has more than 10 million users, while OpenAI’s Dots has helped establish always-on agents as a visible product category, according to CNBC and TechCrunch.

Anthropic is entering after those rivals gained attention, but Duet brings a distinct combination of Claude’s professional brand, background execution, and explicit approval controls. The competitive test will be whether users trust it with enough access to make proactive behavior valuable. If the agent asks for approval too often, it becomes a chatbot with extra steps. If it acts too freely, users face the familiar risks of automation errors and unwanted changes.

The approval requirement is the key safety feature in Duet’s initial design. It keeps users involved before external actions occur while allowing the agent to spend time preparing work in the background. That balance matters for scheduling, finance-related tasks, communications, and software creation, where a small mistake can create real costs.

The research-preview label also signals that Anthropic is testing the product’s behavior, integrations, and workflow before a broad release. Early descriptions point to a capable system, but access remains limited and the experience’s underlying connections are still being built out. Duet’s success will depend as much on dependable permissions and integrations as on Claude’s model performance.

Anthropic’s next challenge is turning Duet from a compelling preview into a dependable daily service. That requires consistent briefings, accurate task execution, useful prioritization, and an approval flow that feels quick rather than obstructive.

The launch also gives professionals a reason to compare agents by workflow instead of benchmark scores. Meta has reach, OpenAI has an established productivity ecosystem, and Anthropic now has a product aimed at persistent professional assistance. As access expands, user retention and the number of tasks people trust Duet to prepare will show whether proactive agents can become regular work infrastructure.