One primary dot with its own computer
OpenAI introduced dots at DevDay as agents that can continue work on a person’s behalf. A dot has its own cloud computer and browser, can use connected apps, and can be reached through ChatGPT, Slack or Teams. OpenAI says it is powered by GPT-6 Astra and can show the user its computer while it works. [1][2]
The September 29 rollout starts with one primary dot for each eligible user. Pro and Business Premium users in eligible markets can create one, while Enterprise workspaces, including Edu and Healthcare, can enable a beta through an administrator. OpenAI describes specialist dots with separate organizational identities as previews and focused enterprise pilots, not generally available teammates. [1]
What it can do without asking
Users choose connected apps and can set custom rules to allow, block or require approval for actions. OpenAI says background proactive research is restricted to read-only tools, while actions affecting accounts or sharing information pass through its action review. Certain sensitive tasks, such as changing a password, stay with the user. These are the company’s stated safeguards, not a measured reliability guarantee. [1]
OpenAI says a first dot is included in Pro or Business Premium with an allowance for deeper work. Tasks it starts in Codex or ChatGPT Work still count against those products’ usage limits. The initial setup is on ChatGPT desktop or a desktop browser; mobile messaging follows setup. [1]
Arriving after Muse and Grok Bot
Two rivals were already on the market. Grok Bot went into beta on August 11 and Muse launched on September 8. Same-day coverage set dots against both: BGR called OpenAI’s version very similar to Muse and Grok Bot, and 9to5Mac’s DevDay live coverage framed it as ChatGPT’s response to the two. Axios compared it with Muse alone. [3][4][2]
Who could get each one differed most. On September 29 dots began rolling out to ChatGPT Pro and Business Premium users in eligible markets, with an admin-enabled beta for Enterprise; BGR reported it was not on the cheaper $100 Pro tier. Muse had a free tier alongside $20 and $100 monthly plans. Grok Bot had started on the top Grok and Cursor plans and, after an August 26 expansion, was included with every SuperGrok plan, Cursor Pro, Pro+ and Ultra, and Cursor Teams, with enterprise access from September 3. [1][3][5][6]
The form factors overlap but are not the same. A person starts with one primary dot, with its own cloud computer and browser, reachable in ChatGPT on desktop, web and mobile, in Slack and Teams, or on a voice call; OpenAI describes teams of dots as a later goal. Grok Bot gives each account several named Bots that share one cloud computer, message each other in group chats and are reached through desktop and mobile apps, and since September 22 by voice in Tesla cars. BGR described Muse as a personal agent working from a dedicated virtual machine across the apps a person connects. [1][7][8][3]
All three converge on the same shape: an agent with a remote computer of its own that acts inside a person’s apps. They split on the entry point. Meta, as Axios put it, launched widely to its consumer user base. SpaceXAI sold a multi-agent product through its own and Cursor’s paid plans. OpenAI started dots with one agent per person on its most expensive plans. None of the sources here measures which one finishes tasks more reliably, so launch-day claims about quality remain vendor and user claims. [2][3][1][5]