CONFIRMED: wrong name, real product
Our strong-signal story two days ago pointed to converging technical evidence of an always-on assistant inside ChatGPT Pro, internally called "o". OpenAI confirmed at its September 29 DevDay in San Francisco that the product is real and ships today, but its official name is Dots. The company describes Dots as remarkably capable, always-on agents built to handle entire tasks on their own while the user does something else.
How the agent actually works
Each Dot runs on GPT-6 Astra, gets its own cloud computer, and connects to more than 4,000 apps through OpenAI's plugin ecosystem. You can talk to it by text or voice call inside ChatGPT, in Slack, and in Microsoft Teams, with context shared across all those channels. Sam Altman demoed the product on stage, saying his own Dot noticed on its own that he had forgotten to invoice a publication and prepared the charge for his approval. Dots ask for explicit permission before sensitive actions, like installing software or changing a password, and users can set Custom Rules to block or allow specific behaviors.
Who gets it first and what it costs
The rollout starts today for ChatGPT Pro subscribers, who pay $100 a month, plus Business Premium and Enterprise in eligible markets. Pro users can join a waitlist to message their Dot directly through iMessage or RCS. OpenAI also announced specialist Dots, versions tuned for specific corporate tasks like accounting, email marketing and legal analysis, with their own access identity and hardware provisioned by the client company.
Why this reshuffles the personal agent board
Meta's Muse shot to the top of app download charts in recent weeks, leaving OpenAI in the odd position of reacting to a rival instead of leading the category. Dots is the direct answer, after xAI had already opened this front in August with Grok Bot. The strategic difference is clear: Muse is free and chased volume, while Dots launches paid and targets enterprise and high ticket users first. For anyone weighing long running agents in production, the caution is the same one that applies to the rest of the GPT-6 Astra family: granting access to apps, messages and a personal computer widens the risk surface, and OpenAI itself recommends extra caution at this early stage.
MaxAssistant's read
Our strong-signal coverage got the name wrong but the thesis right: last week's technical artifacts correctly showed OpenAI had an always-on assistant ready to ship, even under the wrong label. That reinforces a point we keep making, a properly labeled rumor is not guesswork, it is reading real evidence ahead of the official announcement. Now that Dots is confirmed, the test that matters is not the shiny name on stage, it is how much real work it completes without a human babysitter and how much that autonomy costs in security risk, the same dilemma that already pushed OpenAI to shelve GPT-6.1 Astra days earlier.
Sources
OpenAI: Introducing dots; Wired: OpenAI's Dots Are Always-On AI Agents; Business Insider: OpenAI just launched its answer to Meta.