Anthropic Launches Claude Duet AI Work Agent

Anthropic introduced Claude Duet on Wednesday, Oct. 21, launching an AI agent in an early test phase as a personal digital assistant for the office, not personal errands. The move pushes Claude beyond answering questions: it can now handle parts of users’ working lives. Anthropic is betting businesses and professionals will pay for software that does the work, not just talk about it.

A Morning Briefing and Constant Monitoring

Each morning, Claude Duet sends users a briefing drawn from their email, calendar and business software. It then checks those accounts every few hours. If it spots something that may need attention, it contacts the user directly. That’s a different proposition from a chatbot waiting for a prompt. The agent works in the background and connects to more than 6,000 outside apps, including Canva and DocuSign.

The competition is already crowded, and it’s getting crowded fast. Anthropic said Claude Duet competes with OpenAI’s dots, which debuted the previous month. It also faces Meta’s Muse, Google’s Gemini Spark, and agents from start-up Instinct, which raised $1 billion in its latest funding round. A billion dollars in venture money can buy a lot of engineering talent. The market has clearly decided agents are the next battleground.

Purchases, Limits and Guardrails

Claude Duet can make purchases through Link, Stripe’s digital wallet, or Shopify’s Shop Pay. Anthropic requires users to approve every purchase, and that’s the right instinct. Letting software spend money without a human signing off is a serious step. The approval requirement keeps responsibility where it belongs: with the person whose account is on the line.

Anthropic said the agent will show users exactly what it plans to do before sending a message or making a payment. Some actions remain off limits, even with a user’s approval. Trading stocks and moving money between accounts are blocked. A new filter aims to stop the agent from sharing credit-card numbers or government ID numbers. Those are hard lines, and they matter. A tool that can touch financial accounts is only as trustworthy as the boundaries around it.

Access and Cost

Subscribers to Anthropic’s Pro and Max plans in the United States can join a wait-list for Claude Duet at no added cost beyond their subscription fee. That’s a pricing choice worth noticing. Instead of charging a separate premium for the agent, Anthropic is folding it into plans customers already buy. That keeps the product accessible as the company learns how people actually use it. Whether the wait-list converts into paying customers will tell the market a good deal about demand.

Why This Matters: The agent race is becoming a test of whether AI can take real work off people’s plates without bringing real risk along with it. Anthropic’s design choices—user approval for every purchase, a ban on trading and money transfers, and a filter for sensitive identification numbers—reflect the view that individuals should keep control of their accounts and money. From a center-right vantage point, pressure from OpenAI, Meta, Google and Instinct shows that private companies, not regulators, are driving this market forward; firms that earn user trust while delivering measurable value will likely win. The early wait-list stage gives Anthropic room to refine its safeguards before the product reaches a wider audience. The next several months will show whether those guardrails hold up under real-world use.