ChatGPT dots: what they are, who gets them and how to set yours up
Published

Short answerA dot is an always-on ChatGPT agent, powered by GPT-6 Astra, that has its own cloud computer and keeps working on tasks after you close the chat. It is available on Pro and Business Premium plans, not on Free, Go or Plus, so check your plan before you plan around it.
OpenAI announced dots at its DevDay event on 29 September 2026. If you use ChatGPT, this is the biggest change to how it works in a while: instead of answering and stopping, a dot keeps going. This guide explains what a dot is, who can get one, how to set one up, and how to keep it under control.
Everything here comes from OpenAI's dots overview and news reports from the launch, including 9to5Google and Engadget. OpenAI's own help pages weren't readable when we checked, so we haven't repeated details we couldn't confirm. Rollouts change quickly, so check OpenAI's getting started page for the latest.
What a dot is
OpenAI describes a dot as "an always-on agent that keeps work moving across your tools and projects." A few things make it different from a normal ChatGPT chat:
- It runs on GPT-6 Astra, OpenAI's top GPT-6 model.
- It has its own cloud computer and browser, so it can keep working while your laptop is off.
- It remembers. It keeps context across conversations through your chat history, ChatGPT memory and notes it saves about your preferences and decisions.
- It can work on a schedule. For example, "check the planning channel every weekday at 9 AM."
- It connects to your apps. OpenAI says it can use more than 4,000 apps through plugins, such as Gmail, Google Drive and GitHub.
A simple way to think about it: a normal chat is like asking a question at a shop counter. A dot is like handing a task to an assistant who goes away, does it, and comes back to report.
Who can get one
This is the part to read before you get excited.
| Plan | Can get a dot? |
|---|---|
| Pro plans | Yes, for users aged 18 and over, outside the EEA, UK and Switzerland |
| Business Premium | Yes |
| Enterprise | Yes, if a workspace admin turns it on (off by default) |
| Free, Go and Plus | Not yet, and OpenAI hasn't announced a date |
The rollout started on 29 September 2026 and reaches accounts gradually, so eligible users may wait several days. Regional reports say India is included in the current rollout for eligible plans, but you should confirm in your own ChatGPT account.
Your first dot is included with an eligible plan at no extra cost. OpenAI says more dots and a shared workspace called ChatGPT Space are coming later, and the price of extra dots hasn't been announced. Conversations with your dot don't count toward your normal ChatGPT limits. Tasks it hands to Work or Codex count toward those limits, and OpenAI has said extended allowances apply in the first month.
How to set up your dot
- Use a desktop. You create a dot in the ChatGPT desktop app or in ChatGPT on desktop web. Mobile setup and mobile web aren't supported.
- Follow the onboarding. Your dot introduces itself, looks at the context it can see, and suggests ways it can help.
- Personalise it. Choose a name (it gets a handle like
@yourname-dot) and pick its shape, colour, eyes, glasses and accessories. You can change them later. - Set your rules before you connect anything. Go to Settings, then Personalization, then Custom rules under Permissions. Add a rule, describe the action, and choose how the dot should handle it. More on this below.
- Give it a small first job. Something low-risk, such as summarising a document you paste in.
- Connect apps one at a time. Open Settings, then Plugins, choose an app, and sign in with the account you actually want the dot to use.
After the desktop setup, the same dot appears in the ChatGPT mobile apps once they support it.
How to talk to your dot
You can reach your dot in several places:
- ChatGPT on desktop, web and mobile.
- Phone calls in ChatGPT.
- Slack, by direct message or by mentioning it in a channel.
- Microsoft Teams.
- Text messages are listed as coming soon.
Each channel keeps its own messages, but the dot can use what it knows across them. On the desktop app, the Activity view in your dot's profile shows the work it has done, its results and any questions waiting for you. You can redirect it from there.
Rules and safety: the part not to skip
A dot can act on your behalf, so set limits before you rely on it. OpenAI's design includes several controls:
- Built-in approval checks before actions that affect your accounts.
- Custom rules that let you allow an action, require approval, or block it. Reports describe four levels: take action without asking, take action if pre-approved, ask before taking action, or hand off to you.
- Read-only research. OpenAI says the proactive research tools are read-only, so a dot can't send messages or change content without approval.
- Passwords stay with you. OpenAI says password changes stay with you, and that a dot can save and use passwords without exposing them to the model.
- Your computer is optional. Connecting a computer is off by default and needs the ChatGPT desktop app on that machine.
One thing to know from a practical guide by The Rundown: disconnecting an app does not erase information your dot already collected from it. So think before you connect a work account, and check your organisation's data policy first.
A good starting rule: let the dot draft, and you send. In your first prompts, write "Draft suggested replies for my review, but don't send them." OpenAI's own example prompt for planning an offsite says exactly that.
Is it worth it for you?
It's probably worth trying if you're on an eligible plan and you spend a lot of time on repeating chores: checking a shared channel, keeping a project list current, summarising long threads, tracking feedback, or preparing a morning brief.
Wait and see if you're on Free, Go or Plus. There's no date for those plans. In the meantime, the way you write a good dot task also makes a good ChatGPT prompt: say the job, the sources, the limits and when to ask you. Our dots prompts guide shows how.
Be careful if the work is sensitive. Don't connect accounts that hold confidential customer or employer data until you have checked the rules that apply to you.
Where to go from here
Ready to give your dot something to do? Start with our 10 ChatGPT dots prompts. To see how the model behind dots works, read our GPT-6 Sol and Luna guide. The ChatGPT prompts page has hundreds of free prompts, and the prompt builder writes structured prompts for you.
Quick checklist
- A dot is an always-on ChatGPT agent with its own cloud computer.
- Available on Pro plans and Business Premium; Enterprise if the admin allows it.
- Not on Free, Go or Plus yet.
- Create it on desktop, then message it in ChatGPT, Slack or Teams.
- Set Custom rules first, and start with drafts only.
- Connect one app at a time, and use the narrowest account you can.
- Check the Activity view to see what it did.
A dot is new, and OpenAI says it can make mistakes. Read what it produces before you act on it.



Comments
No comments yet. Be the first to share what worked for you.