How ChatGPT Dots actually work (and why they're not just another chatbot)

Picture an employee who never clocks out, never forgets what you told them last week, and quietly keeps working on your to-do list even while you're asleep. That's not a hiring fantasy — that's basically what OpenAI just shipped inside ChatGPT with a feature called Dots.

If you've been using ChatGPT the normal way — open a chat, ask a question, get an answer, close the tab — Dots are going to feel like a completely different product. This guide breaks down what a Dot actually is, how it's different from a regular chat or a Project, how to talk to one, and how to set up your first one without accidentally handing over the keys to your entire digital life.

What is a ChatGPT Dot, exactly?

A Dot is a persistent AI agent that lives inside ChatGPT and keeps working on your goals even after you've closed the app. Your dot is an always-on agent in ChatGPT that can take on ongoing responsibility and keep making progress between conversations, powered by GPT-6 Astra, with its own cloud computer, working across the apps you choose to connect, and remembering context to help with ongoing work.

The easiest mental model: instead of asking a chatbot a question and doing the legwork yourself, instead of asking ChatGPT to help you accomplish something and then doing the rest yourself, the idea is that you can hand a Dot the job. You give it a goal, it figures out the steps, and it comes back to you when it needs a decision only you can make.

Under the hood, each dot comes with four possessions: a dedicated cloud computer, a dedicated ID, apps you've authorized it to connect to, and a communication channel. OpenAI has actually compared this setup to onboarding a new hire — OpenAI describes this as essentially the same environment you'd set up for a new employee. You're not just chatting with a smarter model. You're giving an agent a workstation.

You can check out the official announcement and documentation directly at OpenAI's Introducing Dots page and the Getting Started with Your Dot help article.

How is a Dot different from a regular chat or a Project?

Regular ChatGPT conversations are isolated — close the tab, and the context is gone unless you specifically bring it back up. Projects solved part of that problem by letting you group related chats together so they share context, which is great for ongoing business or personal topics.

A Dot goes several steps further. Dots lets you create a personal AI inside ChatGPT, give it a name, and assign it work. It continues operating outside of conversations and reports back with results and points that need your judgment. It's not just remembering context across chats — it's actively doing things in the background, on its own machine, without you needing to babysit the thread.

And because it's tied to your identity rather than a single conversation, you can start a project in ChatGPT, share context with the working team in Slack, and let your dot follow through, since dots carry context across every channel, so your dot will have the full context wherever you reach it.

How do you talk to your Dot?

One of the more practical upgrades here is flexibility in how you communicate. You're not locked into typing in a chat window. Users can interact with a Dot through ChatGPT, Slack, and Microsoft Teams, including by messaging it or having a voice conversation, and OpenAI also plans to add texting.

That means you can talk to your Dot the way Igor demonstrates in the original video walkthrough — literally having a voice conversation with it, the same way you'd brief a human assistant over a call. You describe what you need, it asks clarifying questions if necessary, and it gets to work.

Once it's set up, access gets even easier. After the initial setup, you'll be able to message it in the ChatGPT mobile app. So the setup happens on desktop, but day-to-day check-ins can happen from your phone.

What can you actually ask a Dot to do?

This is where it gets genuinely useful instead of just novel. Because a Dot has its own cloud computer, it's not limited to generating text — it can browse the web, use connected apps, and interact with software the same way a person would. Dots are persistent AI agents that get their own virtual computer in the cloud, allowing them to browse the web, use connected apps and continue working even when you're no longer sitting in front of ChatGPT.

OpenAI's own framing leans into this range, from trivial to ambitious: your dot can do simple things like book a table — or take on your most ambitious work with the initiative of a high-agency engineer or chief of staff, with its own computer and the ability to work across 4,000+ apps through ChatGPT, using the plugins you connect.

Real-world examples people are already testing include comparing restaurant delivery prices, tracking research threads over days, and catching administrative tasks before they're forgotten — that could mean comparing how much dinner would cost from several restaurants, keeping tabs on something you're researching or catching an administrative task you've forgotten about. One widely cited use case that mirrors the house-hunting example from the video involves a Dot scanning listings, accommodation, or flights, then reporting back only when a decision needs a human — exactly the "set a task and walk away" pattern that makes this feel different from a normal chatbot session.

Early testers are already reporting meaningful time savings. In practical testing, users report information gathering dropping from 20 minutes to zero, and research plus spreadsheet creation shrinking from 75 minutes to roughly 10 minutes.

Who gets access to Dots right now?

This is important before you go looking for the feature and can't find it. Dots are rolling out today in ChatGPT to Pro users in markets excluding the European Economic Area, Switzerland, and the UK, and are also available to Business Premium users across all supported ChatGPT regions. Enterprise customers aren't left out entirely — Enterprise users (including Edu and Healthcare) can try the beta when their workspace admin enables it.

If you're on a Plus, Go, or free plan, you won't see this yet. Availability centers on the Pro and Business Premium plans, with the first dot included at no extra charge, while Plus, Go, and free plans are excluded.

There's also good news on usage limits if you do have access: conversations with your dot don't count toward your ChatGPT usage limits, though when you ask your dot to start or manage tasks in Codex or ChatGPT Work, those tasks count toward your usage limits as usual.

How do you set up your first Dot?

Setup has to happen on a bigger screen first. You can only create dots using the ChatGPT desktop app or a PC browser — they cannot be created on smartphones, and it is not supported on the mobile web version either.

Here's the general flow based on OpenAI's own rollout guidance:

1. Open the ChatGPT desktop app or sign in through a desktop browser.
2. Look for the option to create your first Dot — "today, you can start with your primary dot, give it a name, and make it your own," as OpenAI put it in its announcement.
3. Connect the apps you want it to use. You choose exactly which tools it can touch — you choose which apps dots can access and can manage their permissions through your existing ChatGPT app controls.
4. Set your boundaries using Custom Rules. Users decide what a dot may do on its own, when it must ask first and what it must never do, using Custom Rules, and changing a password is among the tasks that always stay with the user.
5. Give it a goal or task and let it introduce itself. From there, switch over to the mobile app for day-to-day messaging and check-ins.

For a deeper walkthrough with screenshots, the official "Getting started with your dot" guide from OpenAI's help center is the most reliable source. There's also a free downloadable Dots quick-start PDF that walks through practical first use cases if you want a cheat sheet handy while you set things up.

What should you watch out for before handing over the keys?

Autonomy is the whole selling point of Dots — but that cuts both ways. OpenAI has built in layered safety nets, and it's worth understanding them before you connect your email or banking apps.

First, your machine stays separate by default. Each dot works on its own cloud computer, while your computer and its contents stay separate unless you choose to connect it, and for signing into supported websites, dots can use saved passwords without exposing them to the model.

Second, there's a built-in approval system for anything risky. Safety is built around a three-tier permission system: sending messages and sharing files require approval, while password changes and fund transfers are handed back to the user. That matches what OpenAI states directly: dots start with built-in rules for when to act independently and when to ask for approval, and certain sensitive tasks, such as changing a password, always stay with you.

Third, be realistic about current limitations. Early reviewers have flagged some rough edges worth knowing about upfront. Remaining challenges include technical glitches, memory retention limits, and the inability to delete individual memories. If you're the type who wants tight control over what an AI remembers about you, that memory limitation is worth reading up on before you connect anything sensitive.

None of this means you should avoid the feature — it just means you should treat the permission screen the way you'd treat a new hire's first week: give it clear instructions, limited access at first, and expand trust as it proves itself.

Where does this leave you?

Dots mark a real shift in what "using ChatGPT" means. You're no longer just the person typing the prompt and doing the follow-through yourself — you're the person setting direction for something that keeps working whether or not you're watching. Start small: connect one app, give it one real task, and see how it handles the judgment calls. That's the fastest way to find out whether a Dot earns a permanent seat on your team — or just stays a very polished party trick.