OpenAI used its DevDay event on Tuesday to introduce Dots, a new class of personal AI agent that it says can pursue goals continuously rather than wait for the next prompt. The company’s own launch framing was simple: it was “Introducing dots.”¹

The product is powered by GPT-6 Astra and aimed first at Pro, Business and Enterprise customers. Each user starts with one Dot, an agent assigned its own cloud computer and able to work across connected apps, Codex and ChatGPT Work. OpenAI says users set a goal and boundaries, then let the agent research, analyze data, prepare documents or build software in the background.²

At the event, OpenAI cast that autonomy as the next step beyond the chatbot. “Today, you can start with your primary dot, give it a name, and make it your own,” the company said, while holding out a longer-term vision of “teams of Dots working together on your behalf.”³ Sam Altman made the broader ambition even plainer, calling Dots “the real deal version of AI we’ve always imagined.”²

The timing is no accident. Meta’s Muse has already gone broad, reaching a huge consumer audience and moving quickly toward workplace integrations. OpenAI, by contrast, is starting with high-end subscribers and developers—an approach that may offer tighter control, but also leaves it playing catch-up on mass adoption.²

Dots also arrives with a deliberately softer face: a floating, cartoon-like avatar intended to make independent software feel less abstract. But beneath the bubbly branding lies a serious wager. OpenAI is asking people to hand over not just questions, but ongoing tasks—and to accept that the most important AI work may happen out of sight.³