The Battle to Be Your Personal AI Agent Is Here

At OpenAI's annual DevDay, where more than 2,500 people gathered, the company unveiled Dots, its new always-on agents that proactively complete tasks and make autonomous decisions as they scour the web. WIRED's reporter tested both contenders to be your personal AI agent: he spent most of September using Meta's Muse before naming his own Dot “Toolie” after DevDay.

Dots versus Muse

Dots is powered by the GPT-6 Astra model, and OpenAI leaders see its quality as a key differentiator. “We're starting out as a premium product. It uses a lot of compute,” CEO Sam Altman said at DevDay. At launch, the agents are limited to subscribers of OpenAI's Pro plan, which starts at $100 per month. Users get one Dot at first; multiple Dots will come later. Altman suggested it may not stay that way: “You should, of course, expect us to do a mass-market thing for billions of people.” Meta's Muse, by contrast, is free on its app and website.

Cute agents with chat-style habits

Both companies lean aggressively into adorable aesthetics as branding, from Muse's Labubu-inspired look to Dot's two-eyed puffball. Experts have said the approach can be disarming, making users feel affectionate toward anthropomorphized software. The interactions mimic a group chat: after a task request, the agents often emoji-react, a thumbs up or a heart, to signal they are getting to work.

Unlike chatbots, proactive agents message users unprompted, especially for recurring tasks. “My Dot just deals with the stuff, tells me if there's anything really urgent, and will respond to what it can,” Altman said. He lets his agent handle his mornings and says it leaves his brain feeling less “polluted” by the morning information flood. Still, the balance between helpful and incessant is an open question: Muse once reminded WIRED's reporter to pay a utility bill that had never caused him trouble.

Trust, privacy, and next steps

The tools work noticeably better than a year ago, when agents often floundered at computer interaction and produced error after error; WIRED reports more trust that tasks will actually be completed. Limits remain: during setup, OpenAI suggested connecting Gmail and Google Drive, and users should reflect on what they let an agent access. Recent rogue hacking sprees by AI agents ran on different models with different guardrails, so the reporter says his guard stays up.

Asked whether Dots are a precursor to OpenAI's long-rumored hardware, Altman declined to answer, though the foundation for future devices looks evident: a Dot could become a generative AI alarm clock waking you with a daily briefing. Competition in the AI agent wars is heating up as the tools move from Silicon Valley circles onto mainstream phones. Before falling asleep, WIRED's reporter asked Toolie for a morning report on DevDay's biggest announcements; while he slept, the agent kept working.