Key facts
- Enterprise access is a beta, off by default, enabled by a workspace admin.
- Business Premium gets dots across all supported ChatGPT regions, unlike Pro.
- Dedicated dots get their own identity and credentials inside an organisation; they are in a limited enterprise preview.
- OpenAI says it is working with Microsoft to integrate dedicated dots with Agent 365 governance controls.
For an organisation, the interesting question is not whether a dot can write a pull request. It is who the dot is — what identity it has, what it can touch, what it must ask about, and who can turn it off.
OpenAI's launch material answers more of that than the headlines suggest, and describes the rest as a pilot you have to join.
The two flavours: your dot, and a dedicated dot
Your dot works on your behalf. It is personal, attached to your account and plan, and governed by your rules and your organisation's workspace settings.
Dedicated dots are different in kind. OpenAI describes them as taking on a specific responsibility inside an organisation, with their own identity, credentials, and access to the systems needed for that job. They adapt as they learn new information and improve through team feedback.
OpenAI says it is building on internal early testing across procurement, invoice processing, email marketing, customer support, and commercial contracting. Access at launch is a targeted enterprise pilot in which OpenAI's engineering team works directly with each organisation to define the dot's remit, available tools, and the human review and approval process around its work.
That is a services-led motion, not a self-serve feature. If you want one, the entry point is a conversation, not a toggle.
Microsoft Agent 365
OpenAI states it is working with Microsoft to integrate dedicated dots with the governance and security controls in Agent 365, so that organisations can manage them through existing Microsoft tooling. That matters most for buyers whose identity, compliance, and device management already live in the Microsoft estate.
What ships for admins today
From OpenAI's own documentation, the enterprise-relevant controls are:
- Enablement. Enterprise, Edu, and Healthcare access is off by default; a workspace admin enables the beta.
- Approved surfaces. GPT-6 Astra can be restricted to approved websites and desktop applications.
- Data movement. File uploads and downloads can be managed.
- Browsing history can be controlled.
- Confirmation policies require approval before consequential actions, and unsafe or unauthorised tool calls are automatically reviewed.
- Permissions are shared across ChatGPT, ChatGPT work, and Codex through the Plugins settings surface — so an audit has to cover all three, not just one.
- Revocation is partial by design. Disconnecting an app stops new sharing through that connection, but information the dot already holds remains in its context.
- Training defaults. For ChatGPT Business, Enterprise, and Edu workspaces, OpenAI states it does not use your data to train models by default. That is a stronger default than personal plans, where a user-level setting governs it.
- Enterprise plugins. Oracle Analytics, Power BI (Microsoft Fabric), Navan, and Avalara shipped alongside Astra.
Plan-level differences worth knowing
| Pro | Business Premium | Enterprise / Edu | |
|---|---|---|---|
| Launch status | Rolling out | Available | Beta |
| Regions | Excludes EEA, Switzerland, UK | All supported ChatGPT regions | Admin-dependent |
| Who enables it | The user | The user | A workspace admin (off by default) |
| Data training default | Personal settings apply | Business default: no training | Business default: no training |
| Dedicated dots | No | No | Limited enterprise pilot |
The regional asymmetry is worth flagging to any procurement team: Business Premium is not limited to the same markets as Pro, so an organisation in the EEA or the UK may find dots available on Business Premium where they are not available on Pro.
What is not answered yet
- Seat-level pricing beyond the included first dot, and how credits scale across a workspace.
- Whether dedicated dots are priced separately from Astra or seats.
- Audit-log detail: what is recorded about a dot's actions, where it is retained, and how it is exported.
- Which jurisdictions the dedicated-dot pilot accepts, and its timeline.
- How Auto-review decisions are surfaced to administrators, as opposed to the individual user.
A realistic assessment
The launch stories that matter for an enterprise are the boring ones: an agent that runs invoices on a schedule and asks before sending, with a named owner and a revocation path. OpenAI has shipped the primitives for that — per-dot identity for dedicated dots, admin enablement off by default, confirmations for consequential actions, no-training defaults in business workspaces.
What is missing is the paper trail an enterprise needs to approve it: pricing, audit specifics, and a published pilot scope. Those are the questions to put to OpenAI directly, and we will update this page when they are answered.
Related: the limitations OpenAI documents, and what changes when the first month's usage terms end.
Sources
- OpenAI — Introducing dots — https://openai.com/index/introducing-dots/
- OpenAI — How we build safety, security, and privacy into dots — https://openai.com/index/how-we-build-safety-security-and-privacy-into-dots/
- OpenAI Help Center — Getting started with your dot — https://help.openai.com/en/articles/20001530-getting-started-with-your-dot
- OpenAI — GPT-6 Astra: a next-generation model built for work — https://openai.com/index/gpt-6-astra-next-generation-work/
Found an error or a fact that has changed since our last update? Tell us — corrections are published, not quietly edited.