What Is OpenAI Dots? What It Is and How to Access It
An OpenAI dot is an always-on AI agent that lives inside ChatGPT, runs on GPT-6 Astra, and works on your behalf from its own cloud computer. You give it a goal, connect the apps it may use, and it keeps making progress between conversations — researching, drafting, and preparing decisions for your review instead of waiting for your next prompt.
OpenAI announced dots at DevDay on September 29, 2026 and began rolling them out to Pro and Business Premium users in eligible markets, with an Enterprise beta behind an administrator switch. Access is the part people actually get stuck on, because "released" and "available to you" are not the same thing. This guide separates the two: first what a dot is, then the exact path to turn one on — and what to do when you cannot.
What OpenAI dots are, at a glance
A dot is best understood as a standing commitment rather than a chat. A normal ChatGPT conversation ends when you stop typing. A dot keeps a cloud computer running, holds context across sessions, and returns with work already in progress. Two design choices define it: it has infrastructure of its own, and it is bounded by rules you set.
| Attribute | What it means |
|---|---|
| What it is | An always-on agent in ChatGPT that takes on ongoing work |
| Model | Powered by GPT-6 Astra |
| Where it runs | Its own cloud computer and browser |
| Channels | ChatGPT, Slack, and Teams; mobile messaging after setup |
| App access | Connects to 4,000+ apps through plugins you approve |
| Plans | Pro and Business Premium; Enterprise/Edu/Healthcare beta via admin |
| Cost to try | First dot included at no extra charge on eligible plans |
| Creation | Desktop app or desktop web only |
Two details in that table decide most access questions. Creation is desktop-only, so you cannot make a dot on mobile or mobile web. And the first dot is included, so the real barrier is eligibility, not price.
How a dot actually works
Open your dot's profile and you can watch it work: an In progress list, a Scheduled list, and a Completed list, plus its computer, which you can open and interact with at any time. That visibility is the point. The agent is autonomous, but it is not opaque.
The autonomy has a clear boundary. When you are not actively working with your dot, it does proactive research using connected apps — but only through tools that are restricted to read-only. It cannot send messages, change app content, or control your browser or computer on its own. Anything that could affect your accounts or share information passes through rules you set: take action without asking, act if pre-approved, ask first, or hand off to you. Certain sensitive actions, such as changing a password, always stay with you.
That gives a simple mental model for what you hand over:
Runs on its own: reading connected apps, researching, drafting, preparing files, keeping a project moving.
Pre-approved only: actions you explicitly requested in your prompt.
Needs your approval: anything with outside effects that you have not pre-authorized.
Stays with you: sensitive tasks such as credential changes.
Your dot also has its own cloud computer separate from yours. Connecting your local laptop is optional and starts turned off; you enable it deliberately from the desktop app and can revoke it later.
Your dot vs. specialist dots
The word "dots" covers two different things, and the difference is governance rather than raw capability. A personal dot is yours: you name it, it lives in your ChatGPT account, and you own its rules. A specialist dot is an enterprise construct — it has its own identity, its own credentials, and access to specific systems of record, provisioned so it can hold a defined role inside an organization.
| Dimension | Personal dot | Specialist dot |
|---|---|---|
| Owner | You | Your organization |
| Identity | @yourname-dot, renamed by you |
Its own corporate identity and credentials |
| Setup | Self-serve in the desktop app | IT-provisioned, with defined responsibilities |
| Access | Included with an eligible plan | Behind admin enablement and enterprise pilots |
| Governance | Your Custom Rules | Company controls and Microsoft Agent 365 integration |
| Best for | Individuals and small teams | Well-defined roles such as procurement or support |
If you are an individual, the personal dot is the one you are asking about, and the rest of this guide follows that path. If you work in IT or operations, the specialist-dot preview is the relevant branch — it runs through enterprise pilots rather than a self-serve button.
How to access OpenAI dots
The real question is not whether dots exist but whether your plan, market, and device line up. Answer those three and access usually follows.
| Your situation | Availability | Who turns it on |
|---|---|---|
| ChatGPT Pro | Rolling out now | You, self-serve |
| ChatGPT Business Premium | Rolling out now | You, within workspace settings |
| ChatGPT Enterprise / Edu / Healthcare | Beta | Workspace admin |
| ChatGPT Free / Plus | Not yet | — |
| EEA, Switzerland, UK | Not at launch | — |
Then follow the setup path in order:
Confirm your plan. Dots ship with Pro and Business Premium. On Free or Plus, there is no setup step to find yet.
Confirm your market. Rollout is by region, and Pro availability has not extended to the EEA, Switzerland, or the UK at launch.
Open a desktop device. Go to ChatGPT on desktop web or the desktop app, including Windows. Mobile cannot create a dot.
Create and name it. Follow the onboarding prompts. Your default handle is
@yourname-dot; naming your dot updates it, for example@yourname-alfred.Connect apps and set rules. Add the apps your dot may use, review each app's permissions, then define what it can do alone and what needs your approval.
Connect channels if you want them. Set up Slack or Teams from desktop. After setup, you can message your dot from the ChatGPT mobile app where mobile access has landed.
Once it exists, conversations with your dot do not count against your ChatGPT usage limits, and your plan includes an allowance for deeper work, with extended limits in the first month.
What blocks access, and how to fix it
Most access problems are rollout or eligibility, not bugs. The rollout is gradual, so an eligible account can still be missing the option for a few days.
| Symptom | Likely cause | Fix |
|---|---|---|
| No "create dot" option | Rollout has not reached your account | Wait and check back; eligibility is not instantaneous |
| Option missing on your phone | Creation is desktop-only | Use the desktop app or desktop web |
| Plan looks right but nothing appears | Market not yet in rollout | Check OpenAI's eligible-markets list |
| Workspace member can't start | Dots access is off by default for Enterprise | Admin enables "Use dots" for the member or role |
| Dot can't see an app | Plugin not connected or permission denied | Reconnect the app and review permissions on the Plugins screen |
| Texting unavailable | Limited beta, US Pro only | Use ChatGPT, Slack, or Teams instead |
| You want a clean slate | Disconnecting an app does not erase learned data | Reset the dot to clear conversations, memories, and tasks |
Two habits keep access smooth. Keep app permissions minimal, connecting only what a task needs. And when something looks wrong, check the profile's In progress, Scheduled, and Completed lists before assuming the dot failed; it may simply be waiting on a decision.
What an always-on agent runs on
Step back from ChatGPT for a moment. A dot is the clearest public example of a pattern spreading across software: an agent that thinks with a frontier model, acts through data and tools, and runs on an execution layer that returns traceable results. Whether you use a dot or build something like one, the same ingredients have to exist underneath — a reasoning model, data and tools it can call, and one execution surface that handles tasks, status, and output.
That is the layer BeatAPI provides. Where a dot is a finished product you subscribe to, BeatAPI is the capability layer developers place underneath their own agents and apps: a single API key that reaches 60+ model APIs, 1,000+ data APIs, and production workflows, so the model that reasons and the data it acts on come from one place instead of a stack of separate vendors.
The design maps onto what an agent needs. BeatAPI exposes three stable actions — search capabilities, inspect the contract, then run — so an agent can find the right model, data action, or workflow without loading an entire catalog. Each capability carries a live price and a public schema you can validate before you spend. Calls run over MCP or REST, return a task id, and resolve to a hosted file, so status, usage, and output stay on one record instead of scattered across providers.
For a builder, the practical difference is the glue code you do not write. Instead of separate keys, status formats, retries, and output buckets for every model, you get one key and one task shape across video, image, text, and workflows, with eligible failures refunded through the same ledger. It is not a replacement for a ChatGPT dot — it is what you reach for when you want the same kind of always-on behavior inside a product you control.
FAQ
Is OpenAI dots free, and can I use one right now? Your first dot is included with an eligible Pro or Business Premium plan at no extra cost, and dots are rolling out in eligible markets now. "Free to try" is accurate once your plan and market are covered; the gate is eligibility, not price.
Why don't I see the option even though I'm on Pro? The rollout is gradual. An eligible account can take several days to receive access. Confirm your market is included, then check back on desktop.
Can I create a dot on my phone? No. Creation is limited to the ChatGPT desktop app and desktop web, and dots are not supported on mobile web. Once a dot exists, you can message it from the mobile app where mobile access is available.
What's the difference between a dot and a normal ChatGPT chat? A dot keeps working between conversations. It has its own cloud computer, retains context, and performs read-only proactive research from connected apps unless a task needs your approval.
Is it safe to connect my apps and computer? You control the surface. App permissions are managed through your existing ChatGPT controls, local computer access starts off, and proactive tools are read-only. No public safeguard claims that prompt injection is fully solved, so keep the connected footprint minimal and review activity.
How do I build something that behaves like a dot? You need a model to reason, data and tools to act, and an execution layer to run tasks and return results. BeatAPI supplies all three behind one key — start from the docs and see how search, inspect, and run fit.
The short version
An OpenAI dot is an always-on agent with its own computer, and access comes down to three checks: an eligible plan, an eligible market, and a desktop device. Confirm those, create your dot in the desktop app, connect only the apps you need, and set the rules that keep you in the loop. If you would rather put the same capability inside your own product, the BeatAPI capability layer gives your agent the models and data it needs through one API key.