Claude Duet Is Anthropic’s Reply To Meta Muse And OpenAI Dots
Anthropic is late to the celebration, however as of at the moment the corporate has a private agent to match the likes of Meta Muse and OpenAI Dots. As a part of a brand new “analysis” preview, the corporate on Wednesday announced Claude Duet. “It involves you with what wants your consideration, completes duties you ask of it, retains working even if you step away, and checks with you earlier than it acts in your behalf,” Anthropic says of Duet. For now, you will must join a waitlist that is solely open to Claude subscribers within the US to strive the brand new instrument.
Like OpenAI, Anthropic is billing Duet as a proactive agent. Each day, it’ll generate a brief briefing with a listing of motion objects out of your related apps. “You may hand it duties of all shapes — from easy administrative duties of scheduling conferences to extra advanced ones, akin to constructing dashboards with dozens of instruments, creating monetary or gross sales forecasts, prototyping, and constructing net apps,” the corporate writes, promising Duet will solely message you when one thing requires your consideration.
While Duet runs on Opus 5.5, one of many fashions Anthropic gives by Claude, the corporate is releasing standalone Android and iOS apps for folks to make use of Duet. The agent does not require you to have a Claude session open always. It runs by itself digital machine the place it might probably full duties within the background.
Notably, Anthropic says subscribers of its $20 per 30 days Pro plan will get entry as soon as the agent rolls out generally availability. That sets Duet apart from Dots, that are at present solely accessible to OpenAI’s Pro tier subscribers. Those plans begin at $100 per 30 days. Duet shares utilization with all of your different Claude instruments, together with Code, which feels like a recipe for the brand new agent to blow by your charge limits, however Anthropic not less than seems to be considering forward.
“Duet checks your related accounts each few hours, as a result of checking nonstop would burn up your plan’s limits a lot sooner. We’re persevering with to check its judgment about when to message you, and we’ll hold tuning it primarily based on what testers inform us,” the corporate says, including customers will be capable of set how steadily Duet messages them and in relation to what subjects.
On the safety aspect of issues, Anthropic says Duet incorporates a brand new classifier and system immediate designed to forestall it from leaking delicate info. According to Anthropic, Duet won’t ever instantly see your passwords or sign-in tokens, and data which may relate to your well being or private funds is barely let by should you explicitly inform Duet to share it. Anthropic mentioned it might have extra details about Duet’s security protections to share within the coming days.
Of course, how a lot you will get out of Duet will depend upon simply how trusting you’re of Anthropic. Context is every little thing to AI fashions, and for some, sharing their total digital life with an AI agent, is simply an excessive amount of to ask.
