OpenAI Launches Always-On Agents Called Dots to Rival Meta’s Muse

OpenAI has decided the chatbot shouldn’t just wait around for prompts anymore. At DevDay in San Francisco on Tuesday, it showed off “dots,” a new kind of agent that keeps working in the background instead of sitting idle between messages.

The pitch is simple enough: give each dot its own cloud computer, let it use a browser and connected apps, and let it juggle more than one task at a time. OpenAI says the agents are powered by GPT-6 Astra and can reach more than 4,000 apps through plugins. They can be reached through ChatGPT, Slack and Teams, by voice call, and soon by text message.

The examples lean hard into office drudgery. A dot can review customer feedback, turn it into finished pull requests with videos attached, revise launch materials when scope changes, or keep a scientific analysis current when new data lands. One early tester had a dot prepare an invoice he had forgotten to send, then send it after approval. Another dot, called Iggy, helped with a release and rebuilt a missing iOS beta checklist.

There’s a catch, and it’s a big one for Europe. Pro users in the European Economic Area, Switzerland and the United Kingdom are not getting the launch for now. Business Premium is available in supported regions, and enterprise customers can try the beta once their admins switch it on. OpenAI says access will roll out gradually, setup starts on desktop, and sensitive actions like password changes stay with the user. It also says users can set rules for when a dot acts on its own, asks first, or hands the job back.

The timing is impossible to miss. Meta put out its own agent, Muse, about three weeks ago in the U.S., and OpenAI is clearly happy to frame this as a race. At the same time, the company is still dealing with safety questions: on Monday it canceled GPT-6.1 Astra because it missed internal requirements for authorization and scope. So yes, the future is an agent that never sleeps. The paperwork around it still does.