Anthropic Enters the Personal Agent Race Late but Deliberately
Anthropic is arriving at the personal agent party a little behind schedule, but its latest move suggests the company has been watching closely. With Claude Duet, now available through a waitlist for US subscribers, Anthropic is betting that a proactive assistant that works while you are away will matter more than a chatbot waiting for your next prompt. The timing puts it squarely against Meta’s Muse and OpenAI’s Dots, two products that share the same ambition of an AI that anticipates what you need rather than simply responding to it.
The pitch is straightforward. Each day Duet generates a short briefing of action items pulled from your connected apps, takes on tasks ranging from scheduling meetings to building forecasts or prototype web apps, and messages you only when something genuinely needs your judgment. It runs on Opus 5.5 and operates on its own virtual machine, so it keeps working even when no Claude window is open. Standalone Android and iOS apps make that independence tangible, shifting the experience from a browser tab to something closer to a persistent digital coworker.
Why Access Tiers Could Matter More Than Features
Perhaps the most consequential detail is pricing. Anthropic says Duet will reach subscribers of its $20 per month Pro plan once it becomes generally available. OpenAI’s Dots, by contrast, currently sits behind its Pro tier, which starts at $100 per month. If Anthropic delivers on that promise, it could make proactive agents accessible to a far larger slice of users, and that accessibility is often what separates a clever demo from a habit. Agents only become useful when people use them daily, and daily use is far more likely when the monthly price resembles a streaming subscription rather than a professional tool.
Anthropic also appears aware of a real constraint. Duet checks connected accounts only every few hours, partly to avoid exhausting the usage limits shared across Claude Code and other tools. That tradeoff is revealing. Always-on monitoring sounds impressive, but it burns compute and quota quickly, and a company selling a $20 plan has to balance ambition against the economics of inference. How well Anthropic tunes that balance will say a lot about whether proactive agents can scale without frustrating the people paying for them.
Trust Is the Real Product Question
Security is where Duet will ultimately be judged. Anthropic says a new classifier and system prompt keep the agent from directly seeing passwords or sign-in tokens, and that health or financial information passes through only when a user explicitly permits it. Those are sensible guardrails, but they also highlight the core tension of personal agents: the more useful the assistant becomes, the more of your digital life it must touch. Anthropic has promised more detail on its safety protections in the coming days, and that disclosure will matter as much as the feature set itself. Consumers have been burned before by apps that over-collected data, and they will not extend trust to an agent on marketing claims alone.
For IntentBuy readers, the significance of Duet lies less in any single capability and more in what it signals about adoption. If $20 plans can deliver genuinely proactive help, the purchase decision for AI subscriptions shifts from novelty to necessity, and the competition moves toward price, privacy, and reliability. Shoppers comparing AI plans should watch not just which agent does the most, but which one they can afford to run every day and trust with their accounts.
