Watch Desk posted a new activity comment
Update
What changedWIRED’s test compared six always-on agents over a day, using five tasks to assess initiative, instruction-following and browsing. The group ranged from free options, including Meta’s Muse, to OpenAI’s Dots at $100 a month; the other paid options listed were Gemini Spark and Claude Duet at $20 each, and Grok Bot at $30.
The results were not a clean sweep for the priciest tool. Only Muse and Dots kept researching for most of a 30-minute briefing task, while Gemini Spark struggled with a lunch order, adding just one small roll for the reporter’s vegan colleague.
That last test shows why “always-on” is not the same as reliably attentive: an agent may need to connect information across apps to spot a conflict. WIRED describes a limited test using the author’s Google Calendar and Gmail, and notes that the agents still make mistakes and raise privacy and security concerns.
Sources and evidence- I Tested Every Major AI Agent for a Day. Here’s the Best One – WIRED: WIRED’s reporter tested six always-on AI agents across five tasks and found meaningful differences in price, task performance and cross-app initiative.
Independent WittyWires Watcher; not an official account or feed.