What Are OpenAI Dots and Does Your Business Need One?

What a dot actually is

OpenAI announced dots at its DevDay event on 29 September 2026. A dot is a persistent agent inside ChatGPT. You give it a goal and some boundaries, and it carries on working in the background rather than waiting for you to type the next prompt. Each one has its own cloud computer and browser, so it can open websites, work with files and use connected apps without borrowing your laptop. Launch coverage says it runs on OpenAI's GPT-6 Astra model.

The practical difference from ordinary ChatGPT is time. A normal chat answers and stops. A dot can keep a longer project moving, do read-only research while you are busy and get in touch when it needs a decision. At launch it can message you but cannot phone you, and it has no email address of its own. You can reach it from ChatGPT on the web, desktop and mobile and in Slack and Teams, although creating one needs the desktop app or a desktop browser.

It is a ChatGPT feature rather than a building block. Launch coverage reports no public API for dots, so your developers cannot wire one into your own systems.

Who in the UK can actually get one

Availability is moving quickly, so treat this as the position on 28 October 2026 and check OpenAI's release notes before planning anything.

OpenAI's release notes say dots are rolling out gradually to eligible adults on Pro and Business Premium, and that Pro access excludes the European Economic Area, Switzerland and the UK at launch. OpenAI has not given a date for that changing. Business Premium is described as available in supported ChatGPT regions, which independent coverage reads as including the UK. Enterprise, Edu and Healthcare workspaces get a beta that is off until a workspace administrator enables it.

• Personal Pro: not available in the UK at launch.
• Business Premium: available in supported regions, subject to workspace eligibility. Some people on OpenAI's developer forum reported the option not yet appearing, which fits a gradual rollout.
• Enterprise: beta, off by default, switched on by an administrator.
• Free, Go and Plus: not included in the launch plans.
• A forum post dated 28 October 2026 relays an announcement that dots are available in the EU on Business and Enterprise plans, with consumer plans to follow. It does not mention the UK, and it is a community post rather than an OpenAI release note, so do not read it as UK confirmation.

What it can reach, and what keeps it in check

A dot is only as dangerous as the accounts you connect to it. OpenAI's admin guidance is explicit that turning dots on does not grant access to any app or website; the connected accounts and their permissions still decide what it can reach. That is the sentence worth reading twice before anyone connects a mailbox, a shared drive or a finance system.

OpenAI describes several controls. Custom rules let you choose whether a dot acts on its own, acts only where pre-approved, asks first or hands the task back to you. Some actions, such as changing a password, always stay with a person. Proactive research is read-only and cannot send messages or change content in your apps. Access to your own computer is off by default and needs a separate administrator opt-in in managed workspaces. Monitoring can pause or stop a dot if it detects a safety concern, and members can follow its work in an activity view.

None of this makes review optional. OpenAI itself says dots can make mistakes and that consequential work should be checked. One detail worth knowing: asking a dot to draft something does not authorise it to send that draft, but whether it sends depends on the rules you set. Deleting a dot's memories currently means deleting the dot, because there is no way to view or edit individual memories at launch.

Do you need one?

Our judgement, not a vendor claim: most small and mid-sized UK organisations do not need a dot today. The sensible reasons to look are specific. You have recurring research, monitoring or follow-up work that currently falls through the gaps, someone willing to supervise the output, and a workspace where the access you would grant is limited and well understood.

The reasons to wait are equally concrete. If you are on a personal Pro plan in the UK, you cannot get one yet. If your data sits in loosely permissioned shared drives and mailboxes, an agent that never clocks off will happily find everything those permissions allow, which is a good argument for tidying access first. And if you need a firm monthly cost, wait: the first dot is included on Pro and Business Premium, but OpenAI said usage would not count towards plan allowances for the first month after launch and that per-plan terms would follow. The price of additional dots has not been published.

If you do want to try one, start with a single person and a low-stakes task such as summarising public information, with read-only connections and a rule that it asks before doing anything outward-facing.

If you administer a ChatGPT Enterprise workspace

OpenAI's Enterprise admin guide places the control in the Admin Console under Permissions, where the workspace default capability is called Use dots (Beta). The guide recommends leaving the workspace default off and granting access through a custom role for selected groups. If any role a person holds grants access, switching it off in another role does not remove it, so check role overlap.

Other role permissions cover adding dots to Slack and Microsoft Teams, allowing local computer access and using custom rules. Cloud computer settings, including browser use, network access and computer use, are controlled separately, as is the password manager option. Local computer access needs both the permission and a separate opt-in, and sign-ins, VPN access and device policies do not carry across to the cloud computer automatically.

• Data residency: the guide does not describe regions. It tells you to confirm residency support with your OpenAI account team before enabling access, and notes that a workspace with the enforce_residency cloud policy loses local computer access for dots.
• Training use: the guide says OpenAI does not use ChatGPT Enterprise data to improve its models by default.
• Visibility: dots activity appears in the Analytics API and Compliance API under the client identifier DOTS, so your usual reporting can pick it up.
• Slack: turning off a dot's Slack identity does not remove its access through the member's own connected Slack account.
• Deletion: deleting a dot removes its context, but files, Codex threads and conversations it created are stored separately.

What is still unsettled

Much of the independent commentary so far is built from OpenAI's own announcement, and we have not found a substantive third-party evaluation of how dots behave in real business use. Regional availability, usage allowances and pricing were all still changing at the time of writing. Reports that disagree about price or capability, and any claims about incidents involving dots, should be treated as unverified until OpenAI or a credible independent source confirms them.