Can you trust Meta’s Muse or OpenAI’s Dots to run your life?
The Verge Nilay Patel ● Covered by 30 sources
Meta and OpenAI just launched cute AI agents for everyday tasks. Meta’s is free; OpenAI’s sits behind pricey subscriptions and aims at work.
Based on reporting by The Verge, Nilay Patel — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
The new agent fight is less about raw AI brilliance than about packaging. Meta has Muse. OpenAI has Dots. Both are trying to sell the same idea: software that can sit in the background, use a browser, and finish multi-step chores without constant hand-holding.
The pitch is broad and a little unsettling. Book a flight. Make a dinner reservation. Buy a gift. Triage an inbox. OpenAI is pushing further, with specialist Dots aimed at marketing, legal work, and accounting. Sam Altman even called it a “chief of staff,” which tells you exactly how ambitious this thing is supposed to feel.
What makes this round different is how consumer-friendly the wrappers are. Meta is leaning hard into Muse as a free, easy product it can splash across its own services, including Instagram. OpenAI is taking the opposite route: Dots sit behind higher-paid ChatGPT tiers, and the company is openly selling them to businesses and knowledge workers. One side wants scale first. The other wants revenue first.
Underneath the shiny mascots, the technical model is familiar. These agents are basically a model inside a harness that can control a computer, usually through a browser. The lineage runs through OpenClaw, the tool that showed one person could build something useful enough to make the bigger companies scramble. Meta says Muse was built from scratch but inspired by that approach, while OpenAI hired OpenClaw’s creator. Either way, the industry seems to have settled on the same basic recipe.
The real question is not whether the demos look neat. It’s whether people want to hand over their inboxes, card details, and other sensitive data to something that’s supposed to work on its own. Meta and OpenAI are betting that convenience will win. They’re also betting that once these agents get into daily workflows, users won’t leave easily.
My take — AI-written commentary, not fact-checked reporting
This is mostly a lock-in race dressed up as a helpful assistant story. Free from Meta, paid and enterprise-flavored from OpenAI, and both asking for trust before they’ve earned it. The mascot is cute; the moat is the point.
Read more about this at: The Verge