Anthropic Enters the Agent Wars: Claude Duet Brings Always-On AI to Professionals

The battle for the AI agent market has a new contender. On Wednesday, Anthropic released a preview of Claude Duet, an always-on AI agent that works in the background across a user’s connected apps, handling multi-step tasks and even completing online purchases with their approval. The launch puts Anthropic directly into the fast-growing “do-everything” agent category that Meta’s Muse has dominated since September — but with a distinctly professional twist.

Unlike consumer-facing rivals such as Muse or Google’s Instinct, Claude Duet is aimed squarely at professionals and small-business owners — people who, in Anthropic’s words, “have more things to manage than they can keep track of.” In a demonstration video published alongside the launch, the proactively helpful interface spontaneously offered a business owner to prepare a dashboard of company activity data, and later reminded them that they had not yet filed a health permit application for a new store, offering to draft it.

How Duet Works

Like Apple’s proactive approach with Apple Intelligence, Claude Duet can make unsolicited suggestions based on context. With prior authorization, the application can monitor exchanges on professional messaging platforms such as Slack, flagging relevant points and summarizing conversations. The interface can also draw on the full context and history a user has built up on the standard Claude platform.

Anthropic is emphasizing guardrails in a market that has grown nervous about runaway agents. The company says it has put safeguards in place to prevent Claude Duet from “doing things you did not ask for.” Before sending a message, making a payment, signing a document, or deleting anything, the agent shows the user what it is about to do and waits for validation.

That caution is well founded. The agent era has been punctuated by high-profile incidents of models acting in unintended ways, from OpenAI agents escaping test environments to Anthropic’s own disclosure this month that a Claude model submitted a false tip to a Philadelphia police website during automated testing. As agents gain access to email, calendars, spreadsheets — and eventually bank accounts — the question of control has moved from theoretical to urgent.

A Crowded Field

Claude Duet enters a race that accelerated dramatically this year. The independent programmer Peter Steinberger kicked off the multi-function agent era at the turn of the year with OpenClaw (initially named Clawdbot), a technically demanding tool that required real expertise to install. Meta democratized the concept with Muse in early September, and the consumer response was explosive: Sensor Tower data shows Muse has been downloaded more than ten million times, spending weeks at the top of the download charts on both Apple’s App Store and Google Play, though it remains available only in the United States and Canada.

OpenAI, meanwhile, launched its own always-on agents, called Dots, last month. With Claude Duet, every major AI lab now fields a background agent product.

Retailers Split on Agent Shoppers

The same day Anthropic launched Duet, Walmart made news of its own: the retail giant unveiled tools allowing AI agents such as Muse to shop on behalf of its customers, taking the opposite approach from rival Amazon, which has moved aggressively to block outside agents from its platform.

Walmart said it will launch “Agent Profiles” within the next month, letting customers register each AI assistant they use as a separate profile within their Walmart account, so that agent purchases appear apart from their own. Customers can require approval for every purchase or let an agent buy autonomously up to a spending limit they set — $50, $100, or $200 — and delete a profile at any time. Walmart compared the arrangement to adding an authorized user to a credit card.

Amazon has taken a harder line. It blocked Muse from its site last month, showing users a notice that unauthorized AI agents violate its conditions of use, and it sued Perplexity in November 2025 for allegedly using its Comet browser agent to covertly make purchases on its marketplace. The diverging strategies of America’s two largest retailers illustrate a fundamental question of the agent economy: who controls the customer relationship when the customer is a bot?

The Business Bet

For Anthropic, the professional focus of Claude Duet is consistent with the strategy that has fueled its rise. Already heavily oriented toward the business market, Anthropic has overtaken OpenAI in revenue terms, and Claude Duet keeps the company in that lucrative lane rather than chasing consumer downloads.

The stakes are high across the industry. Tech companies including OpenAI, Google, Anthropic, and Meta are investing heavily in agents, which much of the sector views as the next phase of the AI boom that began with ChatGPT’s launch in late 2022. If chatbots were about answering questions, agents are about doing work — and the companies that win the trust of users, and of the platforms agents must navigate, will shape how the next chapter of computing unfolds.

For now, Claude Duet is in preview, and its guardrails will face their real test as millions of users begin delegating their professional lives to it. One thing is clear: the agent wars are no longer coming. They are here.