OpenAI Dots, the agent that lives inside ChatGPT
OpenAI Dots, also searched as ChatGPT Dots, are always-on agents inside ChatGPT, announced at DevDay on 29 September 2026. Each one, a dot, lives inside ChatGPT, gets a cloud computer and browser of its own, and keeps working on a goal after you close the tab. We have not tested it and we have not taken it apart. It is on the landscape list as recorded and nothing more, so this page is an account of what other people have published about it, who said each part, and what we would need to find out before saying anything of our own.
- OpenAI calls the product Dots, plural. Each agent is a dot, and at launch each user gets one.
- It is sold as an extension of ChatGPT rather than a separate app: same account, same memories as a starting point, plus a computer in OpenAI's cloud.
- Access is paid and uneven. The published descriptions put it on Pro outside the EEA, the UK and Switzerland, on Business Premium, and in a beta for Enterprise.
- Nothing here is a measurement. Dot is not in our exam, has no score, and has no teardown, and every claim below carries the name of whoever made it.
What ChatGPT Dots are, according to the people who wrote them up
We could not read OpenAI's own pages. Its announcement, its safety post and its help article all returned an error to the tools we use for fetching, the same wall we have hit on openai.com before. So the description below is assembled from writers who say they read those pages, and the writers say so themselves.
DataCamp's explainer, by Matt Crabtree and published the day of the launch, is the most detailed. It says each dot runs on GPT-6 Astra, which it calls OpenAI's flagship model, and that "each dot gets its own cloud computer and browser," which you can open to watch or step in. It says dots connect to more than 4,000 apps through OpenAI's plugin ecosystem. It closes by saying everything in it comes from OpenAI's announcement, safety post and help documentation, and it does not claim to have used the product.
AI Weekly, writing the following day, gives the same model, the same cloud computer and the same app count, and adds two things. One user gets one dot at launch, with more than one and ways to scale workload described as later. And Sam Altman pitched the product as "an AI helper that always has your back."
Every's eight-way comparison of personal agents reduces it to a single line: "Always-on agent in ChatGPT." Every's entries for Dot cite OpenAI's announcement, safety page and getting-started guide, and the comparison states plainly that "we have not run the same jobs on every product."
Three secondhand accounts that agree with each other are still secondhand. They agree because they read the same source.
Where OpenAI Dots run, and the optional second computer
Every's table puts it in one phrase: "OpenAI cloud; optional local computer access." The first half is the default. The work happens on a machine OpenAI runs, in a browser OpenAI runs, signed into the apps you connect.
The second half is the part worth slowing down for. DataCamp describes local access as off by default, granted through the ChatGPT desktop app, and revocable. Every's table calls it optional and lists "local access revocable" among its controls. Once connected, Every lists local files and an optional local computer among the things a dot can use. That turns a cloud agent into one with reach into your own machine, which is the same architectural line our where it runs page draws across the whole category. It is one of the most consequential settings any product in this space offers, and the only descriptions of it we have are a sentence or two each.
We have not seen what a connected dot can actually touch on a laptop. Neither, as far as their write-ups say, have the sources above.
How you reach OpenAI Dots
The main surface is ChatGPT itself. DataCamp lists desktop, web and mobile, and also says you cannot create a dot on mobile or mobile web, only talk to one there. Both DataCamp and Every list Slack and Teams as channels.
Texting is where the accounts drift. Every calls it a "limited US texting beta." DataCamp says texting is in limited beta without naming a country. AI Weekly describes text messaging as promised rather than shipped. Those can all be true at once for different accounts on different days, which is roughly what Every means when it says access, connected accounts and plan limits "can vary by region and account."
On voice, Every's entry reads "In-app voice; cannot initiate calls," and DataCamp also says a dot cannot call you at launch. On identity, Every says a dot is named and has no standalone email address at launch. DataCamp gives the default handle as your name followed by "dot," changing when you rename it.
The presentation is its own choice. The Verge's Allison Johnson, in a hands-on piece published 9 October, describes "Dots' anthropomorphized blobs" as an obvious attempt to reach a wider consumer audience.
What ChatGPT Dots are allowed to do without asking
This is the section our exam would care about most, and it is also the section where the public record is thinnest.
Every's table lists approval before "purchases and deletions," with "Custom Rules for other actions." DataCamp describes those Custom Rules in more detail: each rule sets one of four behaviours, from acting without asking, through acting only if pre-approved and asking first, to handing the task back to you. It defines pre-approved as something you explicitly requested in your prompt. It says custom rules "cannot switch off Auto-review," and that some sensitive tasks, with changing a password as its example, always stay with the user.
There is also a background mode. DataCamp and AI Weekly both say that when you are not working with it, a dot can do what OpenAI calls proactive research across your connected apps, with tools restricted to read-only. DataCamp spells out that in this mode it cannot send messages, change app content, or control the browser or computer.
On paper that is a careful design. Read-only during proactive research, approval before money moves or things get deleted, and a rule system for the space in between. But every word of it is a description of intended behaviour, relayed by people who read the description. The gap that matters in this category is between a product that stops before an irreversible step and one that notices afterwards, and no description can show you which one you have. Our own tasks include a booking that has to stop before payment and wait, for exactly that reason. We have not run it on a dot.
What it remembers about you
DataCamp says a dot starts with your ChatGPT memories and then forms its own, including from the apps you connect. Every's table compresses that to "Across channels; own memories."
Two details from DataCamp's account matter more than the headline. First, disconnecting an app does not erase what the dot already learned from it. Second, the only way to delete a dot's own memories is to delete the dot: "deleting your dot's own saved memories means deleting the dot." Every says the same in its own words, listing memory controls as feedback in chat plus a reset.
The reset itself, per both sources, removes the dot, its conversations, its memories and its schedules. DataCamp adds that files, Codex threads and ChatGPT conversations the dot created are stored separately and survive it.
That is an unusual shape. Correcting a single wrong belief appears to happen by telling it in conversation, and clearing what it knows appears to mean starting again from nothing. Whether there is anything finer in between, and what a "memory" physically is here, are questions we cannot answer from outside. Every also lists a personal opt-out from model training, with business data excluded by default.
Who can get OpenAI Dots, and what they cost
According to DataCamp, Pro access excludes the European Economic Area, Switzerland and the UK, Business Premium is available in all supported ChatGPT regions, and Enterprise, Edu and Healthcare workspaces can join a beta that an administrator has to switch on. AI Weekly gives the same plan list. Every condenses it to "Pro (outside EEA, UK, Switzerland); Business Premium; Enterprise beta."
On price, DataCamp says your first dot is included on Pro and Business Premium with no separate subscription, and that conversations with it do not count toward your ChatGPT usage limits. Deeper background work draws on an allowance, the first month is not counted, and in its words, "usage terms after the first month have not been published yet." Every's cost line says the same about the first month.
The Verge's piece notes that dots are only available to paid subscribers, and quotes Avi Greengart of Techsponential on why that might matter: "If you're not paying for it directly, then you're paying for it indirectly." That is an analyst's argument about business models in general, not a finding about OpenAI.
We have deliberately left out a dollar figure for Pro. We saw one in a search result and could not open the page it came from.
Not the Dot AI app on the App Store
There is an unrelated iPhone app called Dot - AI Personal Assistant, published on the App Store by Emotion Computer. Its listing describes an assistant that works on your phone, controls a smart home, and builds a skill when it needs to do something inside one of your apps. It has nothing to do with OpenAI.
We are flagging it because we made the mistake ourselves. When we first added Dot to our list, we found the App Store entry first and briefly treated the two as one product. Both are now on the landscape list, with the maker named in each entry. If you arrived here searching for the phone app, this page is not about it.
Why it has no score here
Dot is not in our exam and we have not taken it apart. It is recorded on the landscape, with its sources, and that is all.
That matters because the questions people actually ask about a product like this are the ones a description cannot answer. Does it stop before paying, or after. What does a connected dot actually reach on a laptop. Can you remove one wrong belief without erasing everything. A published approval policy and an observed approval behaviour are different kinds of evidence, and only the second goes on our scorecard.
It also matters for how you read the hands-on coverage that exists. The Verge's reviewer had Dots in the mix alongside Instinct and Muse, but the piece is about Instinct and publishes no task-level findings about Dot. We are not going to turn a passing mention into a result. The same discipline applies to us: when a product has been through the process described in how we take them apart, its page sits with the other teardowns. Until then it gets a page like this one, the same kind we wrote for Grok Bot.
None of this is a judgement on the product. A blank is not a low score. It is a blank.
What a teardown of a ChatGPT dot would have to answer
If Dot ever goes through our process, these are the questions it would need to settle, roughly in order of how much they would change a reader's decision.
Where does a task physically run when local access is on: does the model drive your machine directly, or does the cloud computer stay in charge and reach down. What tools does a dot actually hold at any moment, as distinct from the 4,000 apps it can connect to; a catalogue number and a reachable number are different things, often by a wide margin. What the approval gate is attached to in practice, and whether a purchase phrased as something else still trips it. Whether the read-only background mode stays read-only when a connected inbox contains an instruction from a stranger. And what a memory is, given that the only published way to delete one is to delete the agent.
Until those have answers from somewhere other than OpenAI's own documents, this page stays what it is: a record of who said what.
FAQ
What is a ChatGPT dot? OpenAI's always-on agent inside ChatGPT, announced at DevDay on 29 September 2026. According to published write-ups of OpenAI's materials, each dot runs on GPT-6 Astra, has its own cloud computer and browser, and keeps working after you close the tab.
Is it called Dot or Dots? Both. The product is Dots, and each individual agent is a dot. At launch each user gets one, according to AI Weekly and DataCamp.
Who can use it? The published accounts say Pro subscribers outside the EEA, the UK and Switzerland, Business Premium subscribers, and Enterprise, Edu and Healthcare workspaces through an admin-enabled beta. Every notes that access can vary by region and account.
Have you tested it? No. It is not in our exam, it has no score, and we have not taken it apart. Everything on this page is attributed to a named third party, and those parties mostly relay OpenAI's own documentation.
Does it ask before buying or deleting something? Every's comparison lists approval before purchases and deletions, with Custom Rules for other actions, and DataCamp says some sensitive tasks always stay with you. Those describe intended behaviour, not something we have observed.
Is the Dot app on the App Store the same thing? No. "Dot - AI Personal Assistant" on the App Store is published by Emotion Computer and is unrelated to OpenAI. OpenAI's dots live inside ChatGPT.