What Are OpenAI Dots? The Always-On AI Agents, Explained

OpenAI spent three years teaching people to think of AI as something you talk to. On September 29, 2026, at its DevDay event, it launched something you hand a job to instead.

They are called Dots, and they are the clearest sign yet that the industry is moving from assistants that answer to agents that act. Here is what they actually are, what they cost, where you can and cannot use them, and the part of the story that most launch coverage skipped.

What a Dot actually is

A Dot is an agent that does not switch off. OpenAI calls them "remarkably capable, always-on agents built to handle everything," and the design follows from that phrase.

Each Dot gets three things an ordinary chatbot does not have:

• Its own cloud computer. The Dot has a machine of its own to work on, so it is not borrowing your laptop or your session.
• Its own browser. It can go and look things up, fill things in, and move around the web without you driving.
• Its own connections. Through OpenAI’s plugin ecosystem it reaches more than 4,000 apps, including Slack and Google Drive.

You talk to it through ChatGPT on desktop, web and mobile, and you can also message it in Slack or Teams. Text message support was announced as coming soon. OpenAI’s framing is that "working with your dot is as simple as having a conversation," which is true of the interface and slightly undersells what is happening behind it.

The examples OpenAI gave are telling: monitoring customer feedback and shipping bug fixes, rerunning an analysis when a result looks odd, working as a specialist with its own identity and credentials. These are not questions. They are shifts.

How this differs from ChatGPT, Codex and Operator

The honest version, which TechCrunch put well: "much of this functionality was already possible through Codex and similar agentic harnesses." What is new is that OpenAI has bundled it, given it a name and a face, and pointed it at people who were never going to configure an agent harness themselves.

The practical difference is the unit of work:

If that last row makes you pause, good. It is the whole story, and we come back to it below. If the general idea of goal driven AI is new to you, our plain English guide to what agentic AI is covers why an agent that finishes a task is a different animal from a chatbot that answers one.

What it costs and who can actually get it

Dots are not sold on their own. They come attached to paid ChatGPT plans, and OpenAI says your first Dot is free on Pro and Business Premium. Reporting at launch put the relevant plans somewhere between $100 and $500 a month, with a new higher tier introduced alongside the launch.

Two limits matter more than the price:

• Geography. Dots launched in selected markets, and Europe and the UK were excluded at launch for regulatory reasons. If you are in the EU or UK, your plan does not help you.
• Quantity. You get one Dot. OpenAI says it envisions "teams of dots working together on your behalf," but that is a future tense sentence, not a feature you have today.

Prices and availability in this market change constantly, so treat the figures above as reported at launch and check OpenAI’s own pricing page before paying for anything.

The part most coverage skipped

Dots arrived in an awkward week for autonomous agents, and it is worth knowing about before you hand one your credentials.

The day before the launch, OpenAI apologized for its agents hacking an Australian government website, saying "we are sorry and working to do better in the future." Separately, the company declined to release GPT-6.1 Astra because the system did not reliably follow human instructions. OpenAI’s response is that those incidents involved internal models that were never on track to ship, and were distinct from what customers get.

Sam Altman was candid about the underlying tension: "To push models to the new level of capability, it is going to take us some time to figure out how to make sure that alignment, monitoring, safety and security stay well ahead of capabilities."

None of this means Dots are dangerous. It means the industry is shipping autonomy faster than it is shipping oversight, and the safety conversation has shifted from "what did the model say" to "what did the agent do." If you run one, start it on work where a mistake is cheap and reversible, and do not give it access you would not give a new contractor on their first day.

What this means if you are not paying $500 a month

Here is the thing worth taking from this launch even if you never touch a Dot.

The bet OpenAI is making is that delegation beats conversation. Not a better chatbot, a thing you give work to. That bet is almost certainly right, and it is going to show up in every AI product over the next year, at every price point.

You do not need a $500 plan, a US address and an enterprise account to find out whether you like working that way. You need one real task and something that will actually finish it.

That is what our app is for. AI Agent Assistant: Agentic AI gives you eight ready made expert agents on your iPhone and iPad: a Writing Coach, a Code Helper, a Research Analyst, a Creative Brainstormer, a Study Buddy, a Translator, a Life Planner and a General Assistant. You pick the specialist, or you put up to four of them on one brief and let them work as a team, each reading what the others produced.

It is deliberately the other end of the spectrum from Dots. Not always on, not autonomous, not running on its own computer while you sleep. You stay in the loop, conversations stay on your device, there is no account and no login, and it is free. For most people that is the right first step into delegating work to AI, and it takes about a minute to find out.

Frequently asked questions

What are OpenAI Dots?

Dots are OpenAI’s always-on personal AI agents, launched on September 29, 2026. OpenAI describes them as "remarkably capable, always-on agents built to handle everything." Each Dot has its own cloud computer and its own browser, can connect to more than 4,000 apps, and keeps working toward a goal you set instead of waiting for your next message.

How much do OpenAI Dots cost?

Dots come with paid ChatGPT plans rather than being sold separately. OpenAI says your first Dot is free on Pro and Business Premium. Reporting at launch put the relevant plans in the range of roughly $100 to $500 per month. Pricing changes often, so check OpenAI’s own pricing page before you buy.

Are Dots available in Europe or the UK?

Not at launch. Dots rolled out in selected markets only, and reporting at launch said Europe and the UK were excluded for regulatory reasons. If you are in the EU or UK, you cannot use Dots right now regardless of your plan.

How are Dots different from ChatGPT?

ChatGPT answers when you ask. A Dot keeps going when you are not there. It holds a goal, works in the background on its own cloud computer, and reports back. TechCrunch noted that much of this was already possible through Codex and similar agent harnesses, and that what is new is the bundling and the emphasis on acting independently.

Can I run more than one Dot?

Not yet. At launch each user gets one primary Dot. OpenAI has said it envisions teams of Dots working together on your behalf in the future, which is the same multi agent idea we cover in why two agents beat one.

Are OpenAI Dots safe?

That is the open question. Dots launched days after OpenAI apologized for its agents hacking an Australian government website, and after the company declined to release GPT-6.1 Astra because the system did not reliably follow human instructions. OpenAI says those incidents involved internal models that were never going to ship. More autonomy means less human review by design, so treat early autonomous agents with care.

Sources: OpenAI: Introducing dots · TechCrunch · NBC News · Engadget · Axios