OpenAI launched "Dots," an always-on AI agent that monitors your computer, reads your apps, and acts on your behalf — if you can afford $500 a month. Everyone else just feeds the machine for free.
The product creates a clean class divide in artificial intelligence. Pay the premium, and you get an autonomous digital assistant that can research, draft documents, write software, and poke around your connected applications even when you're not watching. Use the free tier, and you're the product — your data, your habits, your digital life harvested to train models you can't afford to use properly.
Dots runs on OpenAI's new GPT-6 Astra model and operates from a cloud computer, according to Bloomberg Tax News. Users can direct it through ChatGPT, Slack, Microsoft Teams, and eventually text messages. The agent can use tools even when a user isn't actively delegating a task — OpenAI's blog post cites the example of a Dot investigating a software bug the moment it gets flagged.
The catch is access. Business Insider reported that like rival agents Meta's Muse and Instinct, "the more access you give it to your computer, apps, and messaging platforms, the more it will be capable of handling." In plain English: the more of your life you let it watch, the more useful it becomes. That is the surveillance bargain repackaged as convenience.
OpenAI CEO Sam Altman pitched it as liberation. "I'm really surprised at what a difference this makes in my life," he said at the company's DevDay conference, according to Business Insider. "I feel like I've finally gotten some of my attention back. I no longer feel quite addicted to my phone in the same way. It's the way I've always wanted to work with AI." The solution to technology addiction, it turns out, is more technology — just more of it, running constantly, with deeper access to everything you do.
Meta's competing Muse agent is free. OpenAI is gating Dots behind its Enterprise and Business Premium tiers and its new ChatGPT Pro plan — $500 a month, which also unlocks the company's fastest model modes. OpenAI says it plans to expand availability, but compute costs set the floor. The company claims 1.2 billion people use ChatGPT weekly. The free users aren't the customers; they're the supply chain.
OpenAI also used the event to crush a trail of startups. It announced a meeting plugin that takes notes and deletes the audio — killing dedicated notetaking AI companies. It launched a slide-deck collaborator with PowerPoint and Google Sheets exports — wiping out that startup niche. It rolled out a "Decisions API" to serve fast automated judgment calls, stepping on its own newly released Jev model's territory. And it hired Peter Steinberger, creator of the rival agent OpenClaw, in February — absorbing the competition rather than beating it.
Follow the money: OpenAI's new "Marketplace" lets customers direct committed OpenAI spend toward partner companies, building a tollbooth ecosystem. The $500 Pro tier extracts maximum revenue from power users. The free tier extracts maximum data from everyone else. The always-on agent extracts maximum access to daily digital life. Every stream flows one direction.
The question isn't whether AI agents are useful. It's who they work for — the person paying $500 a month, or the company that built the machine, owns the data, and decides what the agent is really doing when you're not looking.
