A reviewer says Anthropic's new Claude Duet agent is a bit less capable than rivals, but its habit of checking in makes it easier to trust with real work.
Anthropic has a new AI agent called Claude Duet, and one reviewer says its best trait is knowing when to stop. An agent is a tool that can take actions for you, not just answer questions. According to the review in Inc., Duet seems slightly less capable than rival tools. That, oddly, made the reviewer willing to let it do more.
The reviewer's point is about trust. Duet is built to pause and check in rather than push ahead on its own. Because it asks first, the reviewer felt comfortable handing it more tasks and worrying less about what it might do without being told.
Anyone trying AI helpers for work tasks should care. A tool that charges ahead can save time, but it can also make mistakes you only find later. One that pauses and asks gives you a chance to catch problems early. For a small business owner, that trade may be worth a little less speed.
The material does not say what Claude Duet costs, when it ships or who can use it today. It also does not list which tasks it handles or how it compares with rivals in tests. This is one reviewer's experience, so it is worth trying the tool yourself on low-risk tasks before relying on it.
When testing an AI agent, favor one that checks in before acting, and start it on low-risk tasks.
Otto, a new customer-tracking app for field sales teams, lets reps speak their notes instead of typing them. Price and launch details are not yet known.
Read more: Otto launches a sales app you talk to instead of typingWorld models, AI that learns physics from video, are being built into home robots and driverless cars, and your own videos may help train them.
Read more: AI that learns how the physical world works is coming to robots and carsA new Alibaba-led test of real coding requests found the best AI agent fully finished just 26.4 percent of tasks, so human review still matters.
Read more: AI coding helpers fully finish only about 1 in 4 real tasks
